THERMODYNAMICS OF STRONG-INTERACTION

MATTER FROM LATTICE QCD

arXiv:1504.05274v1 [hep-lat] 21 Apr 2015

Heng-Tong Ding
Key Laboratory of Quark & Lepton Physics (MOE), Institute of Particle Physics,
Central China Normal University, Wuhan, 430079, China

Frithjof Karsch
Physics Department, Brookhaven National Laboratory, Upton, NY 11973, USA
and
Fakultat fur Physik, Universitat Bielefeld, D-33615 Bielefeld, Germany

Swagato Mukherjee
Physics Department, Brookhaven National Laboratory, Upton, NY 11973, USA

We review results from lattice QCD calculations on the thermodynamics of strong-

interaction matter with emphasis on input these calculations can provide to the
exploration of the phase diagram and properties of hot and dense matter created
in heavy ion experiments. This review is organized as follows:
1) Introduction
2) QCD thermodynamics on the lattice
3) QCD phase diagram at high temperature
4) Bulk thermodynamics
5) Fluctuations of conserved charges
6) Transport properties
7) Open heavy flavors and heavy quarkonia
8) QCD in external magnetic fields
9) Summary

1. Introduction

It has long been recognized that under extreme conditions of high temperature or
densities matter interacting through the strong force strong-interaction matter
cannot exist in the form of nuclear matter formed from hadrons. The copious
production of new resonances1 and the intrinsic size of the nucleons put a natural

1
2 H.-T. Ding, F. Karsch and S. Mukherjee

limit on the range of validity of hadron physics.2,3 In the context of thermodynam-

ics of strong-interaction matter this found its expression in the formulation of the
hadron resonance gas model1,4 which led to the postulate of a limiting temperature
Hagedorn temperature for the thermodynamics of ordinary nuclear matter.
With the formulation of Quantum Chromodynamics (QCD)5,6 as the funda-
mental theory of strong interactions it has soon been realized that matter formed
from strongly interacting particles may be converted into a new form of matter7,8
the quark-gluon plasma9 in which the dominant degrees of freedom are the con-
stituents of hadrons, i.e. quarks and gluons. At low temperatures and densities
they are confined in hadrons. However, at high temperature and densities they can
move freely over macroscopic distances in bulk strong-interaction matter.
It has been speculated that the transition from confined to deconfined matter,
like in condensed matter systems, goes along10 with the restoration of global sym-
metries of the underlying microscopic theory describing the interaction among con-
stituents of strong-interaction matter, i.e. QCD. The analysis of these symmetries
led to the phenomenology of a complex phase structure of strong-interaction mat-
ter, leading to homogeneous and inhomogeneous as well as color-superconducting
phases. This bears many analogies to phase structures known in condensed matter

Fig. 1. QCD phase diagram in the temperature (T ) and baryon chemical potential (B ) plane.
At vanishing B lattice QCD calculations show that the transition is not a phase transition but a
continuous crossover, reflecting pseudo-critical behavior in the vicinity of the true chiral transition.
At B > 0 a second order critical end point (CEP) may exist, which would be followed by a line
of first order transitions at larger values of B . Constraints on the location of a CEP will come
from lattice QCD calculations.
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 3

systems.11,12 Only little of this complex structure is verified through first principle
lattice QCD calculations. In Fig. 1 we show a sketch of the QCD phase diagram that
only highlights those aspects that we believe can be addressed at present through
lattice QCD calculations. We will discuss the phase diagram in detail in Section 3.
Phase transitions as well as properties of matter close to such transitions arise
from complex long range interactions in multi-particle systems on multiple length
scales and at the same time often develop simple universal behavior. Their anal-
ysis requires non-perturbative techniques. With the formulation of lattice QCD,
i.e. the introduction of a discrete space-time lattice as regulator for quantum field
theories like QCD13 a framework for new non-perturbative approaches to the study
of strong-interaction matter had been established. It became possible to perform
first-principle numerical calculations lattice QCD simulations14 by which a sys-
tematic study of the phase structure and basic bulk properties of strong-interaction
matter1517 could be performed for the first time. We will give a brief introduction
to lattice QCD in Section 2. Today these calculations are highly advanced and con-
tinuum extrapolated results for the QCD transition temperature and the equation
of state exist at vanishing net baryon number density. We will review these results
and report on the status of the lattice QCD program that aims at an extension of
these calculations to non-zero net baryon number density or equivalently non-zero
baryon chemical potential in Section 4.
Properties of strong-interaction matter are studied intensively in large exper-
imental programs at the Relativistic Heavy Ion Collider (RHIC) at Brookhaven
National Laboratory (BNL), USA, and the Large Hadron Collider (LHC) as well
as the Super Proton Synchrotron (SPS) at the European Research Center (CERN).
Experiments at these accelerators are devoted to the exploration of the properties of
hot and dense matter created in collisions of ultra-relativistic heavy ions. Through
the variation of beam energies at RHIC, i.e. the beam energy scan (BES), it also
became possible to explore systematically the phase structure of strong-interaction
matter at non-zero net baryon number density. This allows to search systematically
for the chiral critical point or critical end point (CEP) in the QCD phase diagram.
The CEP is a postulated second order phase transition point which is expected to
mark the endpoint of a line of first order phase transitions that separates the low
temperature, low density hadronic phase from a low temperature, large baryon num-
ber quark-gluon plasma phase (also the existence of this first order line at present
is not confirmed through lattice QCD calculations). The existence of a CEP in
the QCD phase diagram is imprinted in the properties of fluctuations of conserved
charges, e.g. net baryon number, electric charge or strangeness fluctuations. These
conserved charge fluctuations can be calculated in lattice QCD at vanishing baryon
chemical potential and the calculations can be extended to non-zero values of the
chemical potential using Taylor expansions of thermodynamic observables. They
can directly be compared to experimental measurements of conserved charge fluc-
tuations performed at the LHC and the BES at RHIC and may provide insight into
4 H.-T. Ding, F. Karsch and S. Mukherjee

the existence and location of a CEP in the QCD phase diagram. We will report in
Section 5 on the status of lattice QCD calculations of conserved charge fluctuations
and their usage to analyze freeze-out conditions in heavy ion experiments.
Heavy ion experiments at RHIC provided striking evidence for the highly cor-
related, non-perturbative structure of strong-interaction matter in the high tem-
perature phase. In particular, the observation of strong elliptic flow and very ef-
ficient energy loss of high momentum quarks traversing through the medium (jet
quenching) paved the way for a picture of a strongly interacting medium that has
been called an almost perfect liquid. These measurements suggest that strong-
interaction matter at temperatures close to but above the transition temperature
has quite unique transport properties, (i) a small shear viscosity to entropy ratio
that may be close to the conformal limit value /s = 1/4 calculated in conformal
field theories (AdS/CFT limit), (ii) a large bulk viscosity that may diverge in the
massless QCD limit at the chiral phase transition temperature, (iii) small electrical
conductivity and diffusion coefficients. These transport properties can systemati-
cally be analyzed in lattice QCD, although this requires the reconstruction of con-
tinuous spectral functions from a finite set of observables, which is in general an
ill-posed problem. The calculation of transport properties in lattice QCD will pro-
vide crucial input to the modeling of the hydrodynamic expansion of hot and dense
matter formed in heavy ion collisions. Spectral functions also carry information on
in-medium modifications of hadrons. The dissociation of heavy quark bound states
has long been advocated as one of the striking signatures for the formation of a
deconfined color screened medium18 while the melting of light quark bound states
is intimately related with the restoration of chiral symmetry.19 We will report on
the status of calculations of transport coefficients and in-medium hadron properties
in Sections 6 and 7.
Finally, in Section 8 we will briefly review a relatively new but very exciting field
of QCD thermodynamics that addresses the behavior of strong-interaction matter in
strong external magnetic fields. Strong, static magnetic fields will certainly influence
the thermodynamics and lead to modifications of the transition temperature as well
as the equation of state. Strong fields are generated in the early phase of heavy ion
collisions20,21 but weaken rapidly during the subsequent expansion phase. While
lattice QCD calculations will allow to understand basic effects that occur in matter
exposed to strong fields it is at present not clear to what extent equilibrium lattice
QCD calculations can contribute to a quantitative analysis of experimental data.

2. QCD thermodynamics on the lattice

2.1. Path integral formulation of QCD thermodynamics

The equilibrium thermodynamics of elementary particles interacting only through

the strong force is controlled by the QCD partition function which can be expressed
in terms of a Euclidean path integral. The grand canonical partition function,
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 5

Z(T, V, ~ ), is given as an integral over the fundamental quark (, ) and gluon

(A ) fields. In addition to its dependence on volume (V ), temperature (T ) and a
set of Nf chemical potentials, ~ (u , d , s , ...) the partition function implicitly
depends on the masses, m ~ (mu , md , ms , ...), of the Nf different quark flavors. In
this review we will most of the time discuss the thermodynamics of QCD with 2
light quarks (u, d) which are assumed to have degenerate masses, mu = md , and a
heavier strange quark (s), with mass ms . This often is referred to as (2+1)-flavor
QCD, or Nf = 2 + 1.
In Euclidean space-time, which is obtained from the Minkowski formulation by
substituting t i with R, the QCD Lagrangian is given by

LE E E
QCD = Lgluon + Lf ermion
1 X 
E

= Fa (x)Fa
f (x) D/ + mf f (x) , (1)
4
f =u,d,s...

where Greek letters are spinor indices, a = 1, .., Nc2 1 is the color index, Nc is the
number of colors (Nc = 3 for QCD) and mf is the mass of quarks with flavor f .
E a
The covariant derivative D/ and the field strength tensor F are given by
 
E
/ = E DE = + ig a Aa E ,
D (2)
2
a
F = Aa Aa gf abc Ab Ac . (3)

Here Aa are the gauge fields, f (f ) are the quark (anti-quark) fields, a are
the generators of SU(Nc ), fabc are the corresponding structure constants, E are
the Euclidean Dirac matrices obeying {E , E } = 2 , and g is the bare coupling
constant.
In the Euclidean path-integral formalism the partition function of QCD is then
given by
Z Y Y
Z(T, V,
~) = DA Df Df eSE (T,V,~) , (4)
f =u,d,s...

with the Euclidean action

1/T
Z Z
~ ) dx0 d3 x LE (~
SE (T, V, ) . (5)
0 V

Here we have suppressed the dependence of the Euclidean Lagrangian and action on
the fields (A , f , f ) but have stressed explicitly their dependence on the various
quark chemical potentials that couple to the conserved quark number currents
X
LE (~
) = LE
QCD + f f 0 f . (6)
f =u,d,s..
6 H.-T. Ding, F. Karsch and S. Mukherjee

The thermal expectation value of physical observables O can be obtained

through
1
Z Y Y
hOi = DA Df Df O eSE (T,V,~) . (7)
Z(T, V,
~) f

Basic thermodynamic quantities like the pressure (P ), energy density () or

net-quark number density nf can be obtained from the logarithm of the partition
function using standard thermodynamic relations,
P 1
= ln Z(T, V,
~) , (8)
T4 V T3 
 1 ln Z(T, V, ~ ) 
= , (9)
T4 V T4 1/T 
~ /T fixed
nf 1 ln Z(T, V,
~)
3
= 3
, (10)
T VT f
where we introduced the chemical potentials in units of temperature, f = f /T .
In fact, the QCD partition function depends on chemical potentials only through
these dimensionless combinations, which are the logarithms of the fugacities, zf
exp(f /T ).
As will become clear in the next subsection, the numerical analysis of thermody-
namic observables at non-vanishing chemical potential, ~ 6= 0, is difficult in lattice
regularized QCD. A viable approach that circumvents the so-called sign-problem in
lattice QCD at ~ 6= 0, suitable for moderate values of the chemical potentials, is to
consider Taylor expansions for thermodynamic observables. The starting point for
such an analysis is the Taylor expansion of the pressure. In (2+1)-flavor QCD it
reads
P X 1
uds
 i  j  k
u d s
= (T ) , (11)
T4 i!j!k! ijk T T T
i,j,k

where the expansion coefficients are dimensionless, generalized susceptibilities that

can be evaluated at
~ = 0,

uds i+j+k P/T 4 
ijk (T ) = . (12)
iu jd ks 


~ =0

We will present results from Taylor expansions of QCD thermodynamics in several

chapters of this review. Other thermodynamic observables and expressions more
suitable for the calculation of P/T 4 and /T 4 in lattice regularized QCD will be
introduced in later sections where they appear.

2.1.1. The high temperature, ideal gas limit

At very high temperatures, all quark masses are small on the scale of the temper-
ature and massless QCD becomes a good approximation. Furthermore, because
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 7

of asymptotic freedom also interactions among quarks and gluons become small at
high temperature. QCD thermodynamics thus approaches that of a non-interacting,
massless quark-gluon gas. In the temperature range accessible to heavy ion colli-
sions the massless limit of 3-flavor QCD is most relevant (Nf = 3). In this limit
the pressure is given by
8 2 X  7 2
  
P 1  f 2 1  f 4
= + + + 2 , (13)
T 4 ideal 45 60 2 T 4 T
f =u,d,s

where the first term gives the contribution of gluons and the sum yields the contri-
bution of quarks with different flavor degrees of freedom.
The thermodynamics of QCD at high temperatures can be systematically an-
alyzed in perturbation theory. This, however, becomes complicated beyond O(g 2 )
due to the appearance of non-perturbative length scales of O(gT ) and O(g 2 T ) re-
flecting electric and magnetic screening lengths,22 which require the resummation
of diagrams leading to the so-called hard thermal loop perturbation theory,23,24
or the explicit integration over hard scales leading to the dimensional reduction
scheme.25,26

2.1.2. The low temperature hadron resonance gas approximation

At low temperature quarks and gluons are confined in colorless hadrons, i.e. baryons
and mesons are the relevant degrees of freedom. In fact, it turns out that even
at temperatures close to the transition region from hadronic matter to the high
temperature quark-gluon plasma phase a non-interacting gas constructed from all
experimentally known resonances does provide quite a good description of QCD
thermodynamics. This had been anticipated by R. Hagedorn when he formulated
the hadron resonance gas (HRG)1,4 model prior to QCD.
In an HRG model the pressure and other thermodynamic observables are easily
obtained from the logarithm of the partition function,
X X
M B
ln ZHRG (T, V,
~) = ln Zm i
(T, V,
~) + ln Zm i
(T, V,
~) , (14)
i mesons i baryons

where
~ = (B , Q , S ) is the set of baryon number, electric charge and strangeness
chemical potentials, respectively. The partition functions for mesons (M ) or baryons
(B) are given by
Z
V
M/B
ln Zmi (T, V, ~) = 2 dk k 2 ln(1 zi ei /T ) , (15)
2 0
with energies 2i = k 2 + m2i and fugacities


zi = exp (Bi B + Qi Q + Si S )/T . (16)

Of course, Bi = 0 for all mesons and Bi = 1 for baryons. The set of hadron
chemical potentials, related to conserved quantum numbers, and the set of quark
8 H.-T. Ding, F. Karsch and S. Mukherjee

flavor chemical potentials are easily related to each other,

1 2
u = B + Q ,
3 3
1 1
d = B Q ,
3 3
1 1
s = B Q S . (17)
3 3
With this it is straightforward to rewrite the Taylor series given in Eq. 11 in terms
of quark chemical potentials, also in terms of baryon number, electric charge and
strangeness chemical potentials,
P X 1  i  j  k
B Q S
4
= BQS
ijk (T ) , (18)
T i!j!k! T T T
i,j,k

where the expansion coefficientsa BQS

ijk can again be evaluated at ~ = 0,

i+j+k P/T 4 
BQS
ijk (T ) = . (19)
iB jQ kS ~ =0


2.2. Lattice Regularization and Continuum Limit

Lattice gauge theory suggested by K. G. Wilson in 197413 is based on the Euclidean
path integral formalism and provides a particular regularization scheme for QCD
by introducing a finite lattice spacing as cut-off. In this way space-time is dis-
cretized and the path integral becomes a finite, yet high dimensional integral over
the gauge and fermion field variables. The discretization of space and time intro-
duces a systematic cut-off dependence in all observables, which vanishes when the
lattice spacing is taken to zero, i.e. in the continuum limit. Many theoretical and
technical details can be found in excellent textbooks such as Ref. 2730. Here we
will only give a brief introduction that may help to make this review self-contained.
By introducing a hyper-cubic lattice of size N3 N with a small but finite lattice
spacing a the calculation of observables given by Eq. 7 reduces to the evaluation of
finite, but high dimensional integrals. Rather than calculating the partition function
directly one usually is concerned with the calculation of expectation values, Eq. 7.
This can be done by exploiting well-known Monte Carlo simulation techniques.14
In order to preserve gauge invariance of the discretized action one introduces the
gauge degrees of freedom as variables living on links between neighboring sites of the
lattice, Ux, exp(iagA (x)), while the fermionic degrees of freedom, x and x ,
are defined on the sites of the lattice. The latter are anti-commuting Grassmann
variables and the former are Nc Nc matrices that are elements of the SU (Nc )
color group.
a We will in general suppress super- and subscripts of the generalized susceptibilities, if a subscript
is zero, i.e. BQS BS BQS
102 12 or 200 2 .
B
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 9

The volume V and the temperature T are related to the spatial and temporal
extents of the lattice, respectively,
1
V = (aN )3 , T = . (20)
aN
Here N and N are the number of sites in spatial and temporal directions. A
consequence of introducing a discrete space-time lattice is that aside from Lorentz
symmetry also some of the symmetries of QCD get explicitly broken by the finite
lattice cut-off. They will be recovered in the continuum limit, a 0 at fixed V and
T . Asymptotically, for small values of the bare gauge coupling g this is controlled
by the QCD -function
 b1 /2b20
1 2
aL = e1/2b0 g , (21)
b0 g 2
Nc2 1
   
with b0 = 11 2 34 2 10

2
3 Nc 3 Nf /16 and b1 = 3 Nc 3 Nc + Nc Nf /(16 2 )2 .
Keeping physical observables constant when approaching the continuum limit
also requires a proper tuning of the bare quark masses. In the discretized version
of QCD (lattice QCD) all quark masses are naturally expressed in units of the
lattice spacing, mf mf a. Taking the continuum limit thus requires to take the
limit, g 2 0 and mf 0 while keeping some physical observables, i.e. one per
flavor degree of freedom, constant. This defines lines of constant physics (LCP)
in the space of quark masses and the gauge coupling. In the case of (2+1)-flavor
QCD with degenerate light quark masses, mu = md , one thus can determine the
bare quark mass parameters ml mu and the strange quark mass ms using two
physical observables. An often used approach is to fix the strange quark mass using
the (fictitious) ss meson mass, which only contains strange quarks and is quite
insensitive to light quark mass values. Thispmass is matched to the lowest order
chiral perturbation theory estimate mss = 2m2K m2 . The light quark mass is
then fixed using a constant ratio ms /ml , the value corresponding to the physical
pion mass value is ms /ml = 27.5.31
The lattice regularized partition function may be written as
Z Y Y
Z(T, V,
~) = dUx, dx,f dx,f eSf Sg ,
x, x,f
Z Y Y
= dUx, detMf (f ) eSg , (22)
x, f

where Sg and Sf are the discretized versions of the gluonic and fermonic part of the
action SE . As the action is bilinear in the quark fields these can be integrated out,
which gives rise to the determinant of fermion matrix Mf . Quite often exploratory
lattice QCD studies are performed in the so-called quenched approximation. This
refers to the calculation of observables that contain fermionic degrees of freedom
as valence quarks. However, back-reaction of fermions on the gauge fields (virtual
quark loops) are not included in the observables. This is achieved by fixing detMf =
10 H.-T. Ding, F. Karsch and S. Mukherjee

const. in Eq. 22, i.e. the determinant does not contribute in the generation of gauge
field configurations nor does it contribute in the calculation of expectation values.
The fermion matrix depends on the quark chemical potential f which is in-
troduced on the time-like links of the lattice.32 Although other formulations are
possible33 a common approach is to replace the link variables, Ux,0 exp (f a)Ux,0

and Ux, 0
exp (f a)Ux, 0
. The fermion determinant is positive definite only for
f 0. For non-zero values of the chemical potential detMf (f ) is a complex
function. Still the partition function is real, i.e. integration over the gauge fields
eliminates the imaginary contributions. However, the real part of the fermion de-
terminant still changes sign, which prohibits the application of conventional Monte
Carlo simulation techniques. This is known as the sign-problem in finite density
QCD. There are various attempts to circumvent this sign-problem in numerical
simulations performed directly at non-vanishing ~ . At present complex Langevin
3436
simulation techniques and the integration over Lefschetz thimbles3739 are most
actively being explored. In this review we will focus on the Taylor expansion ap-
proach40,41 for the evaluation of thermodynamic observables at non-zero chemical
potential. This only requires numerical simulations at vanishing baryon chemical
potential. However, it is naturally limited in applicability through the radius of
convergence of the Taylor series which remains to be determined for QCD.

2.3. Fermion discretization schemes

When discretizing the Euclidean action one has quite some freedom. Discretization
schemes that differ by higher order corrections in the cut-off are equivalent to the
extent that they yield the same physical answers in the continuum limit. The ba-
sic construction principle of a lattice discretized version of QCD is that one wants
to start with a theory that is manifestly gauge invariant. In addition one tries
to choose discretization schemes that preserve as many of the symmetries of QCD
as possible already at non-zero values of the lattice spacing. The discretization of
the gluonic part of the action, Sg , is rather straightforward and well-established im-
provement schemes (Symanzik improvement42,43 ) are known that allow to eliminate
systematically higher order cut-off errors.
Some difficulties already arise in the naive discretization of the fermionic part of
the Euclidean action, Sf . When one simply approximates the covariant derivative
in the fermion action by nearest-neighbor differences of the quark fields additional,
unphysical degrees of freedom arise even in the continuum limit owing to the pe-
riodicity of the fermion dispersion relation on the lattice. This is known as the
fermion doubling problem which is closely related to the explicit breaking of chiral
symmetry in most commonly used fermion discretization schemes.44
There are several discretization schemes for the fermionic part of the action such
as Wilson fermions,13 staggered fermions,45 Domain Wall fermions4649 as well as
overlap fermions.50,51 They differ to the extent they preserve chiral symmetry
and/or eliminate the problem of fermion doublers. Wilson fermions avoid the dou-
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 11

bling problem by adding a dimension five operator to the naively discretized fermion
action.13 A variant of this is the so-called clover-improved Wilson fermion action
where lattice cutoff effects are reduced from O(a) to O(a2 ).52 Most commonly
used in QCD thermodynamics calculations is the staggered fermion discretization
scheme in which the Dirac 4-spinor is spread over several lattice sites.45 This reduces
the doubler problem but does not eliminate it completely. However, it preserves a
U (1)even U (1)odd remnant of chiral symmetry, which allows independent rotations
of the quark fields on even and odd sites of the lattice, respectively. This is of great
advantage in the study of chiral symmetry breaking at non-zero temperature as it
allows to introduce an order parameter, the chiral condensate, that is sensitive to the
spontaneous breaking of this continuous symmetry. The reduction of the doubler
problem in the staggered fermion approach leads to the introduction of additional
heavier states for each fermion flavor. In the continuum limit these so-called taste
symmetry partners will become degenerate with the Goldstone mode that corre-
sponds to the broken U (1)even U (1)odd symmetry. The Goldstone boson thus has
another 15 heavier partners (tastes) which become degenerate with the Goldstone
boson only in the continuum limit. To reduce the taste symmetry breaking, i.e.
reduce the mass difference between physical states and their heavier taste partners
two improved staggered actions, the Highly Improved Staggered Quarks (HISQ)53
and the stout smeared action,54 are commonly used in lattice studies of QCD ther-
modynamics. In Fig. 2 (left) we show some results for the cut-off dependence of
the root-mean-square mass values MRMS of the Goldstone particle and its 15 taste
partners calculated in different staggered fermion discretization schemes. As seen in
the left panel of Fig. 2 the 4stout and HISQ actions have the smallest MRMS . The
right hand panel of this figure shows the influence of cut-off effects on the calculation
of the fermion contribution to the pressure in the high temperature ideal gas limit.
These cut-off effects are independent of the taste improvement schemes and solely
arise from the strategy used to discretize the covariant derivatives appearing in the
fermion action. The stout-smeared actions utilize the naive 1-link discretization
scheme that leads to O(a2 ) errors in the ideal gas limit. In the HISQ, asqtad and
p455 actions 3-link terms are added to the action that eliminate the leading order
cut-off effects in the ideal gas limit. In these cases discretization errors only start
at O(a4 ).
Domain Wall fermions (DWF) and overlap fermions allow for an exact repre-
sentation of chiral symmetry even at non-zero lattice spacing. In the Domain Wall
discretization scheme the physical chiral Dirac fermions are constructed on two 4-
dimensional hyper-surfaces at the edges of a 5-dimensional lattice with extent Ls in
that fifth direction. These hyper-surfaces represent the usual 4-dimensional space-
time and on each of the surfaces fermions with one given chirality exist. There
persists a small explicit breaking of chiral symmetry which vanishes exponentially
in the limit Ls . In DWF calculations this gives rise to a small additive renor-
malization of the quark masses, the so-called residual mass56 mres . Controlling and
12 H.-T. Ding, F. Karsch and S. Mukherjee

600 1.8
RMS M [MeV] naive & stout
P/Pideal asqtad & hisq
500 1.6 p4
HISQ/tree
2stout
400 asqtad 1.4
4stout
300
1.2

200
1
100
0.8
a [fm] N
0
0 0.05 0.1 0.15 0.2 5 10 15 20 25

Fig. 2. (Left) Root mean squared pion mass (MRMS ) as a function of lattice spacing a. RMS
M is defined as the rooted sum of the mass squared of 16 pseudo scalar states divided by 4. The
lattice cutoff effects are smaller with smaller MRMS . (Right) Ratio of quark contribution to the
pressure, obtained from lattice QCD calculations in the infinite temperature, ideal gas limit, to the
pressure of an ideal quark gas in the continuum (Pideal ) as a function of N . P/Pideal approaches
unity in the continuum limit.

reducing these residual mass effects is one of the important improvement steps in
calculations with DWF.57 Overlap fermions on the other hand preserve exact chiral
symmetry on a 4-dimensional lattice by obeying the Ginsparg-Wilson relation at
non-zero lattice spacing.58
Numerical calculations with Domain Wall as well as overlap fermions are quite
time consuming. However, with increasing speed of super-computers calculations
with physical light quark masses become feasible and chiral fermions have been used
recently also for QCD thermodynamics studies.5962 In particular in the analysis of
subtle aspects of the QCD transition related to the temperature dependence of the
axial anomaly, calculations with fermions obeying exact chiral symmetry already at
non-zero values of the cut-off are mandatory.

3. QCD phase diagram at high temperature

Our thinking about the phase structure of strong-interaction matter centers around
two very basic concepts in strong interaction physics confinement and chiral sym-
metry breaking. The former expresses the fact that only colorless states, baryons
and mesons, can exist in the vacuum and are observed experimentally. This gave
rise to the concept of a linearly rising, confining potential exhibited between quarks
and anti-quarks,
(r)
Vqq (r) = + r , (23)
r
with being the string tension and (r) the running coupling of QCD.
Chiral symmetry breaking, on the other hand is a mandatory feature of strong
interactions needed to explain the appearance of a light, almost massless, particle
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 13

in the hadron spectrum the pions. In the limit of vanishing quark masses the
QCD Lagrangian has a build-in chiral symmetry. It is invariant under independent
global rotations of quarks with left and right handed chirality in flavor space as well
as chiral rotations of single flavor quark spinors. This gives rise to the U (Nf )L
U (Nf )R chiral symmetry which is equivalent to SU (Nf )L SU (Nf )R U (1)A
UV (1). The latter UV (1) reflects the conservation of baryon number, and the axial
U (1)A symmetry, although an exact symmetry of the classical QCD Lagrangian, is
explicitly broken by quantum fluctuations. The global SU (Nf )L SU (Nf )R flavor
symmetry, on the other hand, is broken spontaneously in massless QCD, giving rise
to massless Goldstone modes and a non-vanishing chiral condensate,
T ln Z
hif = , f = 1, ..., Nf . (24)
V mf
The masses of the light pseudo-scalar (Goldstone) pions are then understood as
arising from the small non-zero values of the light up and down quark masses. This
is reflected in the Gell-Mann-Oakes-Renner relation between the pion mass (m ),
the quark masses (mu , md ), the pion decay constant (f ) and the non-vanishing
chiral condensate hi = hiu + hid that arises from the spontaneous breaking
of chiral symmetry,
1
f2 m2 = (mu + md )hi . (25)
2
The fact that QCD describes confinement as well as spontaneous breaking of chiral
symmetry is not evident from the QCD Lagrangian or the perturbative treatment
of strong interactions described by it. It requires a non-perturbative analysis
lattice QCD calculations to firmly establish the confining and symmetry breaking
features of QCD.
Asymptotic freedom of QCD suggests that non-perturbative effects are sup-
pressed at high temperatures and hot strong-interaction matter approaches ideal
gas behavior asymptotically. It thus is expected that non-perturbative condensates
also disappear at high temperatures. For exact global symmetries of QCD this is
expected to happen through a true phase transition. Basic features of this transition
can be understood by invoking universality arguments.10
The nature of the QCD transition depends crucially on the values of the quark
masses and the number of flavors (Nf ). Fig. 3 shows a sketch of the nature of the
QCD transition as functions of the quark masses for a theory with two degenerate
light (up and down) quarks with masses, mu,d ml = mu = md , and a heavier
strange quark with mass, ms , at zero baryon chemical potential. In the limit ml
and ms fermions decouple and the thermodynamics of a pure SU (3) gauge
theory (Nf = 0) is recovered. The SU (3) gauge theory has an exact Z(3) symmetry
and the deconfimement transition is first order. This first order region extends to
lower quark masses and ends at a second order critical line that belongs to the
universality class of the 3-d, Z(2) symmetric Ising model.63,64 For three degenerate
flavors Nf = 3 with small quark masses ml = ms 0 the chiral transition is
14 H.-T. Ding, F. Karsch and S. Mukherjee

Nf=2 PURE
ms nd GAUGE
2 order 1st
2nd order
Z(2) order
O(4)

physical point
Nf=3

mtri
s cross over Nf=1

mc

1st 2nd order

Z(2)
order

mu,d

Fig. 3. A sketch of the nature of the QCD transition as functions of the two degenerate light (up
and down) quarks with masses, mu,d ml , and a heavier strange quark with mass, ms , at zero
baryon chemical potential.

known to be first order.65,66 Recent lattice QCD studies67,68 with improved actions
suggest that the extent of this first order region is quite small, i.e. limited to
ml = ms . mphyss /270 where mphys
s is the physical value of the strange quark mass.
An additional ingredient in the discussion of the order of the transition in the
chiral limit arises from the role of the axial anomaly. The nature of the chiral
transition for the massless Nf = 2 theory , i.e. for ml 0 and ms , depends
on the magnitude of the axial UA (1) symmetry breaking. If this remains significant
close to the transition temperature then the relevant symmetry becomes isomorphic
to that of the 3-d O(4) spin model and the transition is expected to be second order
belonging to that universality class.10,69 However, if UA (1) symmetry breaking
becomes negligible near the chiral transition temperature, the relevant symmetry
becomes isomorphic to O(2) O(4) and the transition be either first order10 or
second order.70,71 In the intermediate quark mass region there is no true phase
transition, rather a crossover takes place from the hadronic to the quark-gluon
plasma phase.
All the first order regions are separated from the crossover region by lines of
second order phase transitions belonging to the 3-d Z(2) universality class. The
first order region for the Nf = 2 + 1 case, the second order Z(2) line separating
the Nf = 2 + 1 first order and the crossover regions and the second order O(4) line
for the Nf = 2 case are supposed to meet at a tri-critical point characterized by a
certain value of the strange quark mass, mtrics . Although, it is well established that
in the real world, i.e. for the physical values of the quark masses, the transition
is a crossover,61,72 the location of the physical point with respect to mtric
s has not
been established and even mtric s cannot be ruled out. More specifically, it
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 15

is yet unclear whether mphys

s > mtric
s or mphys
s = mtric
s or mphys
s < mtric
s . If
phys tric
ms > ms then in the limit of ml 0 one should find a second order transition
belonging to the 3-d O(4) universality class, if mphys
s = mtric
s the tri-critical point
6
is a Gaussian fixed point of the 3-dimensional model and its critical exponents
take the mean field values73 and if mphys
s < mtric
s then in the ml 0 limit one may
cross through a second order transition belonging to the 3-d Z(2) universality class
and then may end up in the first order transition region.
In the following discussion we assume mphys
s > mtric
s .

3.1. Chiral transition in (2+1)-flavor QCD

In the vicinity of the chiral phase transition, the free energy density may be ex-
pressed as a sum of a singular and a regular part
T
f = ln Z fsing (t, h) + freg (T, ml , ms ,
~) . (26)
V
The parameter h represents the dimensionless explicit chiral symmetry breaking
(magnetic) field and t incorporates all the thermal variables that do not explicitly
break the chiral symmetry, e.g. at leading order,
1 T Tc0
  2 
q 1 ml
t= + q , and h= , (27)
t0 Tc0 T h0 ms
where q u = d . Tc0 denotes the (unknown) phase transition temperature in the
chiral limit ml 0 and for q = 0. The scaling variables t, h are normalized by two
unknown parameters t0 and h0 . The other unknown parameter q controls how the
chiral transition temperature Tc0 changes as a function of q . All these 4 unknown
quantities (Tc , q , t0 , h0 ) are unique to QCD, similar to the low energy constants
in the chiral Lagrangian. These parameters can be determined by analyzing the
scaling behaviors, arising from the singular part of the free energy, of the chiral
order parameter and susceptibilities using lattice QCD.

3.1.1. Pseudo-critical behavior at vanishing chemical potential

Sufficiently close to the QCD chiral transition the renormalization group invariant
dimensionless order parameter Mb , constructed out of the light quark chiral con-
densate, defined in Eq. 24, and the strange quark mass obeys the scaling relation
ms hil
Mb = h1/ fG (z) + regular terms , (28)
T4
in terms of a single scaling variable z = t/h1/ . The critical exponents and and
the scaling function fG (z) uniquely characterize the universality class of the chiral
phase transition. As discussed before, for continuum QCD the relevant universality
class is expected to be the same as that of 3-d O(4) spin models. However, the
situation is more subtle for QCD on the lattice. For example, since for staggered
fermions away from the continuum limit there is only one Goldstone boson in the
16 H.-T. Ding, F. Karsch and S. Mukherjee

chiral limit the relevant universality class is that of the 3-d O(2) spin model. Table 1
summarizes the critical exponents and other relevant quantities for the 3-d O(2) and
O(4) universality classes.
Fig. 4 shows results for the chiral order parameter Mb calculated in (2+1)-flavor
QCD76 for several values of the light to strange quark mass ratio ml /ms while
keeping ms fixed to its physical value mphys s . In the continuum limit the physical
value of this ratio is mphys
l /m phys
s = 1/27.5. 31
The left hand panel in Fig. 4 shows
the variation of Mb with temperature calculated on rather coarse lattices (N = 4)
for several values of ml /ms . In the continuum limit the smallest of these ratios
would correspond to pion masses of about 80 MeV. Clearly, the transition sharpens
as the light quark masses become smaller. The right hand panel of Fig. 4 shows a
comparison of the scaled order parameter Mb /h1/ (see Eq. 28) with the O(2) scaling
function. As can be seen, for small enough light quark masses, ml . mphys s /20, the
scaled order parameter for different values of the light quark mass collapses on a
unique curve and compares well with the O(2) scaling function. While these lattice
results suggest that the chiral transition for Nf = 2 theory belongs to the 3-d O(N )
universality class one should bear in mind that these results are not continuum
extrapolated and discretization effects may change the conclusion. In fact, lattice
studies performed with the unimproved staggered action even further away from the
continuum limit suggest that in this case the transition may even be first order.77
The chiral order parameter Mb can be used to define two susceptibilities obtained
by either taking a derivative with respect to thermal variable t, which give the
mixed susceptibility t Mb /t 2 fs /th, or with respect to the symmetry
breaking variable, h Mb /h 2 fs /h2 . These two susceptibilities are the
only two second derivatives of the free energy of an O(N ) symmetric theory like
QCD that diverge at the critical pointb , i.e. at T = Tc0 for ml = 0. Their singular
0
behavior is controlled by two scaling functions, fG (z) = dfG (z)/dz, and f (z) =
0
(fG (z) (z/)fG (z))/, respectively. Away from the chiral limit, these two scaling
functions have maxima at universal values of z, i.e. z = zt and z = zp (see Tab. 1),
b In general, there is a third susceptibility, the specific heat, C , which is obtained as a second
V
derivative of the free energy with respect to the thermal variable t, CV 2 fs /t2 . In O(N )
symmetric theories, however, the relevant critical exponent is negative. The specific heat, thus,
does not diverge at Tc in the chiral limit.

Table 1. Critical exponents , , , for the 3-dimensional

O(N ) universality classes.74,75 Only two of the four critical
exponents are independent. They are related to each other
through the scaling relations = ( 1) and + 2 + = 2.
The last two columns give the location of the maxima of the
0 (z) and f (z), respectively.74,75
scaling functions fG

N zt zp
2 -0.017 0.349 1.319 4.779 0.46(20) 1.56(10)
4 -0.213 0.380 1.453 4.824 0.73(10) 1.35(3)
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 17

2.50
ml/ms=1/10 2.00
Mb/h1/ ml/ms=2/5
1/20 1/5
Mb 1/40 1/10
1/80 1/20
2.00 1.50 O(2) 1/40
chiral limit
1/80
1.50
1.00

1.00

0.50 all masses

0.50

T/Tc t/h1/
0.00 0.00
0.94 0.96 0.98 1.00 1.02 1.04 1.06 1.08 -5 -4 -3 -2 -1 0 1 2 3 4 5

Fig. 4. O(N ) scaling of the chiral order parameter Mb introduced in Eq. 28 for (2+1)-flavor
QCD.76

which can be used to define pseudo-critical temperatures.

For example, the scaling behavior of the chiral susceptibility for two light, de-
generate flavors,
hil
m,l (T ) = 2 , (29)
ml
is related to that of h . The renormalization group invariant product of the chiral
susceptibility m,l and the square of the strange quark mass is related to the scaling
functions by,
m2s m,l 1 1/1
4
= h f (z) + regular terms . (30)
T h0
At vanishing quark chemical potential the chiral susceptibility thus diverges at the
chiral critical temperature Tc0 in the chiral limit h 0. The peak location of the
chiral susceptibility will then be associated with the peak of the scaling function
f (z), located at z = zp , and defines the chiral pseudo-critical temperature Tc for
h > 0,
"  1/ #
0 zp ml
Tc = Tc 1 + + regular terms , (31)
z0 ms
1/
with z0 = t0 /h0 . As illustrated in Fig. 5, the chiral crossover temperature for
physical QCD with ms = mphys s and mphys
l = mphys
s /27.5 was studied in detail78 by
using such a scaling analysis of the chiral susceptibility, including the influence of
regular terms. By performing scaling fits to the chiral susceptibility, extrapolating
to the physical value of light quark mass and subsequently taking the continuum
limit one obtains a value of
Tc = (154 9) MeV (32)
for the chiral crossover temperature in (2+1)-flavor physical QCD.78 Studies with
other improved staggered fermion discretization schemes,7981 which used other
18 H.-T. Ding, F. Karsch and S. Mukherjee

275 195
190
250 HISQ/tree: N = 8, O(4)
m,l / T2 185 Tc [MeV] Physical ml/ms
225 180 HISQ/tree
ml / ms = 0.050
0.037 175 Asqtad
200
170
175 165
150 160
155
125 Combined continuum extrapolation
150 -2
HISQ/tree: quadratic in N
100 145 Asqtad: quadratic in N-2

75 T [MeV] 140 N-2

135
130 140 150 160 170 180 190 200 210 220 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035

Fig. 5. (Left) O(4) scaling of the chiral susceptibility and its extrapolation to the physical light
quark mass for (2+1)-flavor QCD.78 The peak location of the chiral susceptibility defines the chiral
crossover temperature. (Right) Continuum extrapolation of the chiral crossover temperature for
(2+1)-flavor physical QCD, yielding Tc = (154 9) MeV.78

criteria, not directly related to criticality, to define a crossover temperature, and

with chiral Domain Wall Fermions61 also yielded compatible results for the QCD
chiral crossover temperature.

3.1.2. Curvature of the pseudo-critical line

The scaling behavior of the mixed susceptibility, t , can be utilized to determine
the chiral phase boundary in the T - plane.82 As introduced in Eq. 27, to leading
order the chemical potential q does not contribute to the explicit chiral symmetry
breaking field h and can be treated as part of the thermal scaling variable t.
Thus, the derivative of the order parameter with respect to the chemical potential,
m, = 2 Mb /(q /T )2 , is equivalent to the mixed susceptibility, t . The scaling
behavior of the renormalization group invariant combination of m, multiplied
with the strange quark mass, which combines a thermal and field-like derivative
of the free energy, is given by
ms m, 2q (1)/ 0
2
= h fG (z) + regular terms . (33)
T t0
In the chiral limit, h 0, this susceptibility diverges at Tc0 for q = 0. For q > 0,
the chemical potential dependence of the chiral transition temperature Tc0 (q ) is
0
defined by the peak location, z = zt , of the scaling function fG (z) and in the
2
leading order in (q /T ) it is given by
"  2 "  ##
4
0 0 q q
Tc (q ) = Tc 1 q + O + regular terms . (34)
Tc0 Tc0

As illustrated in Fig. 6 (left), the curvature of the chiral phase transition line, q ,
can be determined by fitting the scaled susceptibility ms t0 h(1)/ m, /T 2 to the
0
universal scaling function fG (z).82 Such a scaling analysis provides a value of the
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 19

0.07
-2qfG(z) 200 62.4 39 27 19.6 14.5 11.6 9.1 7.7
- ms t0 h(1-)/ m, / T2 N=8: ml/ms= 1/20
0.06 N=4: ml/ms= 1/10 T [MeV] s [GeV]
1/20 170
0.05 1/40
1/80
160
0.04

0.03 150

0.02 140 Tc(B)

freeze-out
0.01 CEP: JHEP 0404, 050
CEP: PoS LATTICE2013, 202 B [MeV]
z=t/h1/
0
-2 -1 0 1 2 3 0 50 100 150 200 250 300 350 400

Fig. 6. (Left) O(N ) scaling of the susceptibility m, and determination of the curvature, q , of
the chiral phase boundary in the T -q plane.82 (Right) Chiral crossover temperature as function
of baryon chemical potential, Tc (B ), compared with the freeze-out temperature in heavy-ion
collisions. The parametrization of the freeze-out line is taken from Ref. 86, with the freeze-
out temperature in the limit of infinite collision energy adjusted to the crossover temperature

at vanishing chemical potential, Tf o ( s ) = 154 MeV. The black band for the crossover
line, Tc (B ), reflects the systematic uncertainty arising from a factor of 2 difference in current
estimates from calculations with real82,83 and imaginary84,85 B (see text). Data points are
from a reweighting analysis87 and Taylor expansion88 (see text). They also are normalized to
Tc = 154 MeV.

curvature q = 0.059(6).82 In terms of the baryon chemical potential B = 3q ,

the chiral transition temperature is then given by
"  2 " 4 ##
0 0 B B
Tc (B ) = Tc 1 0.0066(7) +O . (35)
Tc0 Tc0

This result is in very good agreement with the independent determination of the
continuum extrapolated results of the curvature of the chiral phase boundary for
(2+1)-flavor of physical QCD,83 using different criteria. However, analytic continu-
ation from purely imaginary chemical potential yields a factor two larger values for
the curvature.84,85 Fig. 6 (right) shows the proximity of the QCD phase boundary
to the freeze-out line in heavy-ion collisions86 in the T -B plane.

3.1.3. The critical end point

Information on the location of a critical end point (CEP) in the QCD phase dia-
gram (Fig. 1) from lattice QCD is ambiguous. Although considerable progress in
developing algorithms that would allow direct calculations in lattice QCD at B > 0
have been made recently (see Ref. 36 for a recent review) these techniques are not
yet applicable to realistic QCD parameter values.
The first calculations that provided hints for the existence of a critical point87
used a reweighting technique that is applicable only on rather small lattices. It
had been performed on coarse lattices with temporal extent N = 4 using only the
naive 1-link staggered fermion action. Taste violation effects thus are large and the
20 H.-T. Ding, F. Karsch and S. Mukherjee

reweighting is known to fail when used too far away from the parameters actually
used in the calculations. It has been argued that this overlap problem may have
led to the spurious identification of a signal for a critical point.89 Calculations
performed with an imaginary chemical potential do not find any evidence for the
existence of a critical point.90 However, also these calculations have been performed
on coarse lattices and utilized the naive staggered 1-link action.
The conclusions drawn in Ref. 90 are based on an analysis of the B -dependence
of the location of the boundary line that separates the first order transition region
shown in Fig. 3 at small quark masses from the crossover region. These calcula-
tions, performed with staggered fermions using an imaginary chemical potential,90
suggest that this boundary moves to smaller quark masses with increasing chemi-
cal potential. This disfavors the existence of a critical point connected to the Z(2)
boundary line at vanishing chemical potential. A similar analysis, performed with
clover-improved Wilson fermions, comes to the opposite conclusion91 and thus does
favor the existence of a critical point. The apparent differences between calcula-
tions performed with staggered fermions and Wilson fermions on coarse lattices
underscores that better control over systematic errors is needed before a definite
conclusion on the existence of a critical point can be drawn.
Analyzing the convergence properties of the Taylor series for the pressure
at non-zero chemical potential40,92 (Eq. 18) does, in principle, allow to relate
the location of the CEP to the radius of convergence of this series. Although
this approach has been used closer to the continuum limit (N = 8)88,93 than
the studies discussed above, the current estimates still are based on calcula-
tions with the naive staggered 1-link action. They still suffer from large taste
symmetry violations. These calculations yield for the location of the critical
point88 (T E /Tc , E E
B /T ) = (0.94(1), 1.68(5))
93
which is significantly lower than
the reweighting result (T /Tc , B /T ) = (0.99(1), 2.2(2)).87 We show these two
E E E

estimates for the location of the CEP in Fig. 6 (right). The systematic errors of
these estimates are at present difficult to estimate.
Further information on the location of a critical point comes from the calcu-
lations of (i) the equation of state at B > 0 performed with improved staggered
fermions, which we will discuss in Section 4, as well as (ii) the analysis of freeze-out
parameters based on Taylor expansions of cumulants of charge fluctuations, which
are also performed with improved staggered actions and which we will discuss in
Section 5, both suggest that a CEP located at B /T < 2 is unlikely.

3.2. Deconfining aspect of QCD

Pure SU (Nc ) Yang-Mills theory without quarks possess an exact global Z(Nc ) cen-
ter symmetry, which gets spontaneously broken in the high temperature deconfining
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 21

phase.15,16 The thermal expectation value of the renormalized Polyakov loop

* N
+
c(g 2 )N 1 X Y
Lren (T ) = e Tr U(x0 ,x),0 , (36)
V Nc x =1
~
x 0

is not invariant under the Z(Nc ) center symmetry. The Polyakov loop can be in-
terpreted as the free energy difference, F (T ), of a thermal system with and with-
out an infinitely heavy static quark anti-quark pair separated by infinite distance,
Lren (T ) = exp(F (T )/2T ). The renormalization constant c(g 2 ) can be fixed94 by
demanding that in the short distance limit the heavy quark free energy coincides,
up to a trivial additive constant, with the Coulombic short distance behavior of the
zero temperature heavy quark potential defined in Eq. 23. In the confined phase
F (T ) = , as a static quark and anti-quark pair cannot be separated by infinite
distance in this phase. Thus in the Z(Nc ) symmetric confined phase Lren (T ) = 0.
On the other hand, in the spontaneously Z(Nc ) broken deconfined phase a static
quark anti-quark pair can be separated by infinite distance due to the presence of
color screening and Lren (T ) 6= 0. Thus, for a pure SU (Nc ) gauge theory, i.e. in the
limit ml , the Polyakov loop serves as the order parameter for the deconfine-
ment transition.95 However, since the mere presence of quarks explicitly breaks the
Z(Nc ) center symmetry, the Polyakov loop does not serve as an order parameter
for QCD with realistic light quarks. Furthermore, since the Polyakov loop is not
related to a derivative of the QCD partition function with respect to the thermal
or the symmetry breaking field, its change or fluctuation across the QCD transition
may not capture the true singularities of the QCD partition function in any limit.
Thus, a deconfining temperature defined from the change of the Polyakov loop may
not reflect the pseudo-critical properties of QCD with realistic light quark masses.
As shown in Fig. 7 (left), the change of the Polyakov loop within the chiral crossover
region, Tc = 154(9) MeV, is rather gradual and smooth.

0.35 Lren(T) B B
0.30 2- 4 non-int. quarks
0.3 BS BS
0.25 31 - 11
HISQ: N=6
0.25 N=8
N=10 0.20
0.2 N=12
cont. 0.15 0.10
0.15 stout, cont.
0.10 0.05

0.1 0.00
0.05
0.05 uncorr. 130 145 160 175
0.00 hadrons T [MeV]
T [MeV]
0
120 140 160 180 200 120 160 200 240 280 320

Fig. 7. (Left) The renormalized Polyakov loop in (2+1)-flavor QCD.96 (Right) Appearance of the
fractionally charged degrees of freedom in the chiral crossover region Tc = 154(9) MeV (shaded
region) for the light as well as the strange quark. The black points show results obtained using
the stout action97 and the other points have been obtained using the HISQ action.98
22 H.-T. Ding, F. Karsch and S. Mukherjee

Deconfinement is associated with the liberation of degrees of freedom, mani-

fested by a rapid rise in the bulk thermodynamic observables such as the pres-
sure, energy density etc. Among such bulk thermodynamic observables fluctuations
and correlations of different conserved quantum numbers, e.g. the baryon number
(B), electric charge (Q) and strangeness (S), directly probe the liberation of the
quark degrees of freedom and the consequent appearance of fractionally charged
quantum number carriers. For example, both for an uncorrelated hadron gas and
in a free quark gas the difference of second to fourth order cumulants of baryon
number fluctuations probes the baryon number of degrees of freedom, specifically
B B 2 4
2 4 B B . For hadronic degrees of freedom carrying baryon number
B = 1 one thus has B B
2 4 = 0, but for quark-like degrees of freedom with
B B
B = 1/3 one finds 2 4 6= 0. Similarly, for the strange quark sector higher or-
der baryon-strangeness correlation such as BS BS 3
31 11 (B B)S is non-vanishing
if the strangeness carrying degrees of freedom are associated with quark-like baryon
number B = 1/3 and vanishes when the strangeness is carried by hadronic degrees
of freedom with B = 1. Thus, such combinations of higher order fluctuations and
correlations of quantum numbers are sensitive probes of appearance of fraction-
ally charged degrees of freedom in a deconfined medium.98100 The right hand
panel of Fig. 7 shows lattice QCD results for B B BS BS
2 4 and 31 11 from the
BNL-Bielefeld collaboration as well as the Budapest-Wuppertal collaboration.97
98

Clearly, both quantities start deviating strongly from zero in the chiral crossover
region, Tc = 154(9) MeV, indicating that fractionally charged degrees of freedom
start to appear at these temperatures and onset of deconfinement takes place for
the light as well as the strange quarks. Note, however, that also these fluctuation
observables are not order parameters in the strict sense.

3.3. Axial symmetry of QCD at high temperature

Although, the axial UA (1) symmetry is not an exact symmetry of QCD, as men-
tioned before, the magnitude of its breaking is expected to influence the order of
the chiral phase transition. Hence, knowledge of the temperature dependence of the
UA (1) breaking is essential for a comprehensive understanding of the QCD chiral
transition. The axial UA (1) symmetry of the massless QCD Lagrangian is broken
due to quantum fluctuations, resulting in non-conservation of axial current101,102
as well as explicit breaking of the global UA (1) symmetry induced by topologi-
cally non-trivial gauge field configurations, such as the instantons.103 Due to color
screening the instanton density gets suppressed as the temperature increases104 and
in the T limit the UA (1) symmetry becomes exact.
Since UA (1) is not an exact symmetry of QCD one cannot define an order pa-
rameter associated with its breaking. However, for two massless flavors, the pion
() and the iso-vector scalar (a1 ) mesons transform into each other under a UA (1)
rotation. Since the presence of an exact UA (1) will render these meson states degen-
erate, the difference of the integrated two-point correlation functions of pion and
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 23

meson,
Z 

d4 x + (x) (0) + (x) (0)


= , (37)

will also vanish in the limit of exact UA (1) symmetry. Corrections due to small non-
vanishing light quark masses will only contribute at O(m2l ). Thus, this quantity
can be taken as a measure of the UA (1) breaking.105

(MS
MS 2 m=135 MeV
250 - )/T
m=200 MeV
200

150

100

50

0
130 140 150 160 170 180 190 200
T [MeV]

Fig. 8. The UA (1) breaking measure renormalized in the MS scheme as function of

temperature. Results are from a calculation with Domain Wall fermions.61 The difference is non-
vanishing for T Tc = 154(9) MeV. It becomes independent of quark mass for T & 168 MeV and
rises with decreasing quark mass at lower temperature. In fact, it will diverge in the chiral limit
for T < Tc .

This measure of UA (1) breaking was studied in detail using improved p4 stag-
gered fermions,106 and it was found that remains non-vanishing for
T . 1.2Tc . However, for staggered fermions the issue of axial anomaly is quite
subtle and the correct anomaly may only emerge in the continuum limit.107,108
On the other hand, emergence of the axial anomaly is more straightforward for
the chiral DWF formulation.49 For DWF, axial symmetry is broken by the same
topologically non-trivial configurations as in the continuum. Lattice artifacts ap-
pear only at O(m2res ), due to the explicit chiral symmetry breaking residual mass,56
mres , arising from finiteness of the fifth dimension (Ls < ). Thus, the DWF
action turns out to be a natural candidate for investigation of the temperature de-
pendence of UA (1) breaking in QCD. The temperature dependence of the UA (1)
breaking measure, , has been extensively studied using the DWF formalism
for several volumes as well as quark masses.59,61,109 As depicted in Fig. 8, cal-
culations with DWF clearly show that does not vanish around the chiral
crossover temperature Tc and remains non-vanishing for 165 MeV . T . 195 MeV,
independent of the light quark masses. These results indicate that UA (1) symmetry
24 H.-T. Ding, F. Karsch and S. Mukherjee

may remain broken at these temperatures even in the chiral limit.

1
()/T 3
0.8
1.5 Tc 323 8
0.6 =0

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
/T

+ +5
0.06 0.06
0.06 0.04 0.06
0.04 0.02 0.04
0.04 0 0.02
0.02 -0.02 0
0.02 -0.04 -0.02
0 -0.06 -0.04
0 -0.06

4 4
3 3
x (fm) 2 0 x (fm) 2 0
1 1 1 1
2 2
0 3 y (fm) 0 3 y (fm)
4 4

Fig. 9. (Top) Accumulation of near-zero eigenmodes of the Overlap-Dirac fermion matrix at

T = 1.5Tc .62 (Bottom) Spatial profile of a typical near-zero mode (left) and the spatial profile of
the chirality of the same near-zero mode (right) at T = 1.5Tc .62

Since UA (1) breaking arises due to topology of the gauge fields, it is intimately
related to the infrared modes of the

Dirac
 fermions. In the limit of infinite volume,
both the chiral order parameter, l , and the UA (1) breaking measure, ,
can be represented in terms of the eigenvalue density, (), of the Dirac fermions
Z Z

 2ml () 4m2l ()
l = d 2 , and = d 2 . (38)
0 + m2l 0 (2 + m2 ) l


In the chiral symmetric phase and in the chiral limit l must vanish, but
can remain non-zero. Identification of the infrared fermionic modes, i.e. the form
of (), that give rise to such a phenomenon naturally leads to the underlying non-
perturbative mechanism of axial symmetry breaking. Studies with DWF59,109 as
well as with overlap fermions,62 possessing even better chiral properties and an exact
index theorem, suggest that an accumulation of near-zero eigenmodes of the form
() m2l () may largely account for the observed UA (1) breaking in at
high temperature. Such accumulation of the near-zero modes are depicted in Fig. 9
(top). More detailed studies62 of the space-time profiles, as illustrated in Fig. 9
(bottom), localization properties and distributions of these near-zero modes indicate
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 25

that their behavior is consistent with underlying presence of a dilute instanton gas
a gas of widely separated, weakly interacting, small instantons and anti-instantons.
This suggests that even at temperatures T 1.5Tc weakly interacting instanton
anti-instanton pairs are largely responsible for the UA (1) breaking.
The analysis of UA (1) symmetry breaking close to the chiral limit of 2 or (2+1)-
flavor QCD, however, is far from being settled. Detailed systematic studies of
the quark mass and volume dependence as well as the cut-off dependence are still
missing. Moreover, results obtained with chiral fermions are still controversial.
Calculations performed with dynamical overlap fermions60,110 and so-called optimal
domain wall fermions111 at present suggest that UA (1) does get restored already
at Tc , which would still allow for the occurrence of a first order transition when
approaching the chiral limit.

4. Bulk thermodynamics

4.1. The QCD equation of state at vanishing chemical potential

Lattice QCD calculations of the equation of state, or more general, of basic bulk
thermodynamic observables like the pressure (P ), energy density () or entropy
density (s), have reached now a precision where continuum extrapolated results
for physical light and strange quark masses can be obtained. This also allows
detailed comparisons with perturbative calculations at high temperature and model
calculations at low temperature.
The procedure to calculate bulk thermodynamic observables in lattice QCD
is well established. The starting point for these calculations is the evaluation of
the trace anomaly, (T )  3P . The trace anomaly is directly obtained as a
temperature derivative of P/T 4 which also is related to the QCD partition function,
Z(T, V ),
 3P dP/T 4
= T , (39)
T4 dT
with P/T 4 = limV (V T 3 )1 ln Z(T, V ). As the trace anomaly arises as a deriva-
tive of the logarithm of the QCD partition function with respect to temperature,
it can be expressed in terms of expectation values of rather simple observables, e.g.
a combination of the gauge action and light and strange quark chiral condensates.
Nonetheless, the calculations become rather time consuming because a subtrac-
tion of the zero temperature contributions is needed to eliminate divergent vacuum
contributionsc .
As can be seen from Eq. 39 up to an integration constant at a temperature T0
the pressure can be obtained from the trace anomaly,
Z T
P (T ) P (T0 )  3P
= dT 0 . (40)
T4 T04 T0 T 05
c For more details we refer, for instance, to Ref. 112.
26 H.-T. Ding, F. Karsch and S. Mukherjee

16
4 non-int. limit

12
3
HRG
Tc
8
2
stout HISQ 3p/T4
(-3p)/T4 /T4
p/T4 4 3s/4T3
1
s/4T3
T [MeV] T [MeV]
0 0
130 170 210 250 290 330 370 130 170 210 250 290 330 370

Fig. 10. (Left) Comparison of the trace anomaly ( 3P )/T 4 , pressure and entropy density
calculated with the HISQ (colored)114 and stout (grey)113 discretization schemes for staggered
fermions. (Right) Continuum extrapolated results for pressure, energy density and entropy den-
sity obtained with the HISQ action.114 Solid lines on the low temperature side correspond to
results obtained from hadron resonance gas (HRG) model calculations. The dashed line at high
temperatures shows the result for a non-interacting quark-gluon gas.

This allows to reconstruct the energy density as well as the entropy density s/T 3 =
( + P )/T 4 .
The determination of thermodynamic quantities in QCD is a parameter free
calculation. All input parameters needed in the calculation, e.g. the quark masses
(mu = md , ms ) and the relation between the lattice cut-off, a, and the bare gauge
coupling, = 6/g 2 , are determined through calculations at zero temperature. Like-
wise, there is only a single independent thermodynamic observable that is calculated
in a lattice QCD calculation, for instance the trace anomaly, (T ). All other bulk
thermodynamic observables are obtained from (T ) through standard thermo-
dynamic relations. In Fig. 10 (left) we show recent results for the trace anomaly
of (2+1)-flavor QCD113,114 obtained with two different discretization schemes by
two different groups. The results are extrapolated to the continuum limit and are
obtained with a strange quark mass tuned to its physical value and light quark
masses that differ slightly (ms /ml = 27113 and 20114 ). The right hand panel in this
figure shows results for the pressure, energy density and entropy density obtained
from the trace anomaly by using Eqs. 39 and 40.
Also shown in Fig. 10 are results obtained from a hadron resonance gas (HRG)
model calculation of bulk thermodynamics. As can be seen this describes the QCD
equation of state quite well also in the transition region, although it may be noted
that the HRG calculations yield results for all observables that are at the lower error
band of the current QCD results. It has been speculated that this may indicate
contributions from additional, experimentally not yet observed resonances which
could contribute to the thermodynamics.115 Indeed evidence for the contribution
of a large number of strange baryons has recently been found in lattice QCD calcu-
lations of conserved charge fluctuations116 (see also the discussion in Section 5 and
7).
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 27

1
QCD
[GeV/fm3] HRG
0.8

0.6 Tc

0.4
c

0.2

T [MeV]
0
130 140 150 160 170 180

Fig. 11. The critical energy density c in (2+1)-flavor QCD. The band gives the continuum
extrapolated result for the energy density taken from Ref. 114. The HRG curve is based on all
resonance with mass less than 2.5 GeV listed by the Particle Data Group.31

In Fig. 11 we show the energy density in the low temperature region. The
box highlights the transition region characterized by the crossover temperature,
Tc = (1549) MeV. From this we deduce the energy density in the crossover region,
3
c = (0.34 0.16) GeV/fm . This is a rather small value for the energy density
needed to convert ordinary hadronic matter into a medium made up from quarks
and gluons. It may be compared to the energy density of ordinary nuclear matter,
3
nuclear matter ' 0.15 GeV/fm or the energy density inside a nucleon, nucleon '
3
0.45 GeV/fm , assuming the radius of nucleon RN ' 0.8 fm. In fact, c is close to
the energy density reached in the dense packing limit of nucleons with radius RN .
The simple bulk thermodynamic observables like pressure and energy density
give the impression that an HRG model also provides a good description of the
equation of state for temperatures above the crossover region, i.e. for T & 165 MeV.
However, as discussed in the previous section the analysis of conserved charge fluc-
tuations clearly shows that at temperatures T & 160 MeV thermodynamics can
no longer be described in terms of hadronic degrees of freedom (see for instance
Fig. 7 (right) and the discussion in Section 7). This also becomes evident in sec-
ond order derivatives of the QCD partition function with respect to temperature.
The speed of sound, c2s = dp/d, is related to the inverse of the specific heat,
CV = d/dT ,
dp dp/dT s
c2s = = = , (41)
d d/dT CV
(/T 4 ) 
   
CV   
= 4 4 +T . (42)
T3 T V T T V
Both quantities are shown in Fig. 12. The specific heat does not develop a pro-
nounced peak in the transition region as one could have expected from pseudo-
28 H.-T. Ding, F. Karsch and S. Mukherjee

0.35 70
non-int. limit
non-int. limit
c2s 60
0.30 HRG
50

0.25 40

c 30 CV/T3
0.20 4/T4
20 Td(/T4)/dT
0.15 HRG 10
[GeV/fm3]
T [MeV]
0
0.10 150 200 250 300 350
0.1 0.2 0.5 1 2 5 10 40

Fig. 12. The velocity of sound in (2+1)-flavor QCD (left) and the specific heat CV /T 4 together
with the two components (see Eq. 42) contributing to it (right). Solid black lines in the low and
high temperature regions show the corresponding hadron resonance gas (HRG) and non-interacting
quark-gluon gas results, respectively.

critical behavior of energy density fluctuations close to a critical point. This may
be understood114 from the temperature dependence of the two terms contribut-
ing to CV /T 3 . The dominant singular contribution arises from the temperature
derivative of /T 4 , which has a peak. This, however, is overwhelmed by the large
energy density contribution at high temperature which reflects the liberation of
many partonic degrees of freedom. Furthermore, even in the chiral limit, where
QCD is expected to have a second order phase transition belonging to the uni-
versality class of 3-d, O(4) symmetric spin models, the specific heat will not di-
verge as the relevant critical exponent ' 0.2 that controls its singular behavior,
CV /T 3 (|T Tc |/Tc ) + const., is negative for this universality class (see Ta-
ble 1). The speed of sound will therefore stay non-zero at Tc also in the chiral limit.

4.2. The QCD equation of state at non-vanishing chemical potential

In Section 3 we have discussed lattice QCD results on the dependence of the QCD
crossover temperature on the baryon chemical potential and its relation to the
freeze-out temperatures determined in heavy ion experiments. These experiments,
in particular the beam energy scan program performed at RHIC, will probe prop-
erties of strong-interaction matter at non-vanishing baryon chemical potential in
the temperature range 0.9 . T /Tc . 2 and 0 . B /T . 3, with Tc denoting
the crossover temperature at B = 0. For the hydrodynamic modeling of matter
in this (T, B ) regime it thus is of importance to also know the equation of state
at non-vanishing B /T . As direct numerical calculations at non-zero B are not
yet possible, a viable approach is to analyze the equation of state using a Taylor
expansion in terms of chemical potentials117,118 as given in Eq. 18. In this way
some results for the EoS at non-zero baryon chemical potential have already been
obtained on coarse lattices.40,41,118,119 Continuum extrapolated results for Taylor
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 29

expansion coefficients of the pressure at O(2f ) have been obtained.120,121 These

have been used to construct the EoS at O(2B )122 which implements the strangeness
neutrality constraint119 (hnS i = 0) and the electric charge to baryon number rela-
tion (hnQ i = 0.4hnB i) suitable for conditions met in heavy ion collisions.

0.35 1.2
B B
B free 4 /2
2
0.3 hadron resonance gas
1

0.25 stout cont.

0.8 HISQ, N=6
0.2 8
continuum extrap.
N=12 0.6 data:
0.15 BNL-Bielefeld-CCNU
8
6 0.4 preliminary
0.1 PDG-HRG

0.05 0.2
free quark gas
T [MeV] T [MeV]
0 0
120 140 160 180 200 220 240 260 280 120 140 160 180 200 220 240 260 280

Fig. 13. Expansion coefficients of the pressure at non-zero baryon chemical potential. The left
hand figure shows the leading order correction121 and the right hand figure shows the relative
contribution of the next to leading order correction. The continuum extrapolated result obtained
with the stout action is taken from Ref. 123.

We will discuss here only the case Q = S = 0 in some detail in order to

illustrate the relative importance of higher order corrections in different temperature
and B regions. The Taylor series for the pressure is given by,

P (T, B ) P (T, 0) 1 B  2 
B 1 B
4 (T ) B
 2 
= (T ) 1 + + O(6B ) . (43)
T4 2 2 T 12 B
2 (T ) T

The leading order correction to the pressure thus is proportional to the quadratic
fluctuations of net baryon number. The next to leading order corrections are pro-
portional to the quartic fluctuations. When written in the form given in Eq. 43
it becomes easy to identify the relative importance of leading and next to leading
order corrections. In Fig. 13 we show B B B
2 (T ) (left) and 2 (T )/4 (T ) (right). With
4
increasing temperature the O(B ) correction rapidly looses importance relative to
the leading O(2B ) term.
In Fig. 14 we show preliminary results for the B -dependent contribution to
the total pressure evaluated for different values of B /T and taking into account
corrections up to O((B /T )4 ).124 These results suggests that an O((B /T )4 ) Taylor
expansion of the pressure (and energy density) is well controlled for all values of the
chemical potential below B /T = 2. This covers a wide range of the QCD phase
diagram accessible in the beam energy scan (BES) at RHIC, i.e. the region of beam

energies s 20 GeV.
30 H.-T. Ding, F. Karsch and S. Mukherjee

1 4.5
(P(T,B)-P(T,0))/T4 4 P(T,B)/T
4

0.8 3.5
3
0.6
2.5

B/T=1.0 2 B/T=0.0
0.4
2.0 1.5 1.0
2.5 2.0
0.2 1 2.5
0.5
T [MeV] T [MeV]
0 0
150 200 250 300 350 150 200 250 300 350

Fig. 14. (Left) The B -dependent part of the pressure at O((B /T )2 ) (black) and O((B /T )4 )
(colored).124 The latter is shown only in the temperature regime where neglected corrections at
O((B /T )6 ) are estimated to contribute less than 10%. (Right) When combined with the B = 0
contribution for the pressure shown in Fig. 10 these neglected terms contribute less than 3% to
the total pressure.124 The grey band gives the uncertainty on P (T, 0)/T 4 and the central line in
the band is the parametrization of P (T, 0)/T 4 given in Ref. 114.

4.3. Perturbation theory, hadron resonance gas and the strongly

interacting liquid
In our discussion of bulk thermodynamic observables we have compared with HRG
model calculations. We have seen that the HRG does provide a rather good de-
scription of bulk thermodynamics below and even in the crossover region. However,
the HRG does depend on the resonance spectrum used in the calculation. This is of
some significance when one looks at more selective observables like, e.g., fluctuations
and correlations of strange baryons (see Sections 5 and 7). Spectrum independent
HRG results are, for instance, related to fluctuations of net baryon number. The ra-
tio of fourth to second order net-baryon number cumulants shown in Fig. 13 (right)
is unity in any HRG model irrespective of details of the spectrum. This figure thus
shows that an HRG model calculation can be expected to provide a good descrip-
tion of QCD thermodynamics on a 10% level only at temperatures less than about
T ' (140 145) MeV.
At asymptotically high temperatures hard thermal loop, perturbative calcula-
tions24 or dimensionally reduced QCD (EQCD)125 provide a well-established frame-
work for the analysis of bulk thermodynamics and fluctuations of conserved charges.
We show in Fig. 15 a comparison of the trace anomaly (left) and the pressure
(right) calculated in lattice QCD with HTL and EQCD calculations. These cal-
culations seem to give a good description of bulk thermodynamics at tempera-
tures T & 400 MeV. However, even in this regime there remain differences between
HTL and EQCD calculations that are of the order of 10%. Moreover, within the
HTL calculations there seems to be no unique choice for the renormalization scale
, that would allow to match all observables simultaneously. Similar conclusions
can be drawn from the HTL analysis of net-baryon number fluctuations shown in
Fig. 16 (right). Although observables that are dominated at high temperature by
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 31

1.0
(-3p)/T4
HISQ
4 p/pideal
O(g6) EQCD 0.8
3-loop HTL
3
0.6

2 0.4
HISQ
O(g6) EQCD
1 0.2 3-loop HTL

T [MeV] T [MeV]
0 0.0
130 200 300 400 500 600 800 1000 130 200 300 400 500 600 800 1000

Fig. 15. Comparison of the (2+1)-flavor calculation114 of the trace anomaly (left) and pressure
(right) with HTL and EQCD (dashed line) calculations. The black line corresponds to the HTL
calculation24 with renormalization scale = 2T . Note that this solid line would move up for the
trace anomaly and move down for the pressure if the scale in HTL is reduced.

9B B
4/ 2

HRG

st
rongly
i
nteract
ing

li
quid

per
tur
bat
iveQGP
T [MeV]

Fig. 16. The ratio of quartic and quadratic net-baryon number fluctuations versus temperature.
The left hand panel shows temperature ranges in which HRG and resummed perturbative calcu-
lations, respectively, provide good approximations to lattice QCD results. The right hand panel
shows the result from a HTL-resummed calculations.24

quark rather than gluon contributions seem to approach perturbative behavior ear-
lier, it still is evident that agreement with lattice QCD results on the 10% level only
is possible for T & (250300) MeV. In general the temperature range Tc T 2 Tc
is highly non-perturbative and obviously not accessible to hadronic model calcula-
tions. This is highlighted in the left hand panel of Fig. 16. We will discuss in the
following sections properties of strong-interaction matter in this temperature range.

5. Fluctuations of conserved charges

Proximity of a second order criticality, such as the O(4) chiral phase transition
or the QCD critical point, is universally manifested through long-range correla-
tions at all length scales, resulting in increased fluctuations of the order parameter.
32 H.-T. Ding, F. Karsch and S. Mukherjee

These fluctuations can be quantified through the Gaussian (variance) as well as

non-Gaussian (skewness, kurtosis etc.) cumulants of the distribution of the order
parameter. Higher order non-Gaussian cumulants become increasingly sensitive to
proximity of a critical point100 which is reflected in their grows with higher powers
of the correlation length.126 Moreover, even qualitative features, such as the sign
change and the associated non-monotonicity, of these non-Gaussian cumulants can
encode the presence of a nearby critical region.127129 Non-Gaussian cumulants can
be accessed in heavy-ion experiments via the event-by-event fluctuations of vari-
ous conserved charges and particle multiplicities.130134 In this vein, a major focus
of the Beam Energy Scan (BES) program135 at the RHIC is measurements of the
event-by-event fluctuations of particle multiplicities and conserved charges.136141
Among several conserved charges, the net electric charge is of special interest.
Cumulants of net electric charge fluctuations can be measured in experiments138,141
and are calculable in lattice QCD.120,121 Cumulants of net electric charge fluctua-
tions and correlations with other conserved charges are well defined in lattice QCD
at vanishing chemical potentials where standard lattice QCD techniques can be used
to compute them. Furthermore, Taylor expansions in powers of B around B = 0,
can be employed to obtain generalized susceptibilities for B > 0,

X 1 BQ  k
B
Q
n (T, B ) = (T ) . (44)
k! kn T
k=0
These susceptibilities directly relate to cumulants of the net electric charge fluctu-
ations,
1
Q1 (T, B ) = hNQ i ,
V T3
1 D
2
E
Q2 (T, B ) = (NQ ) ,
V T3
1 D
3
E
Q3 (T, B ) = (NQ ) ,
V T3 D
1 E D E2 
4 2
Q4 (T, B ) = (NQ ) 3 (NQ ) , (45)
V T3
where NQ is the net (positive minus negative) charge and NQ = NQ hNQ i.
On the other hand, through the measurements of the event-by-event distribu-
tions of the net electric charge, heavy-ion experiments provide various cumulants,
mean (MQ ), variance (Q ), skewness (SQ ), and kurtosis (Q ), of the electric charge

fluctuations for given beam energy ( s)138

2

D 2
E
MQ s = hNQ i , Q s = (NQ ) ,
D E D E
3 4

(NQ )
(NQ )
SQ s = 3 , Q s = 4 3. (46)
Q Q
Thus, the charge susceptibilities obtained from lattice QCD calculations and the
cumulants measured in the heavy-ion experiments are directly related to each other
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 33

through the appropriate volume-independent ratios142

MQ ( s) Q
1 (T, B ) Q
2
= R12 , (47a)
Q ( s) Q
2 (T, B )
3

SQ ( s) Q ( s) Q (T, B ) Q
= 3Q R31 . (47b)
MQ ( s) 1 (T, B )
As in many other cases, one still has to worry about directly confronting thermody-
namic QCD calculations performed in equilibrium and in the thermodynamic limit
to the rapidly expanding dense matter created in heavy ion collisions in a small
volume. In the analysis of higher order cumulants additional complications may
arise from more complicated efficiency corrections and the limited experimental
acceptance windows.143145
In heavy-ion collision experiments the only tunable parameter is the beam en-

ergy, s. However, to gain access to the information regarding the QCD phase
diagram this only tunable parameter needs to be related to the thermodynamic
variables, e.g. temperature and baryon chemical potential. Traditionally, this

s (T, B ) mapping has been done by relying on the statistical hadronization

model based analysis.86 Recent advances in heavy-ion experiments as well as in
lattice QCD calculations have placed us in a unique situation where, for the first
time, this mapping can now be obtained through direct comparisons between the
experimental results and rigorous (lattice) QCD calculations. Recently, it has been
Q
shown142,146 that by directly comparing lattice QCD calculations for R31 (Eq. 47b)
Q
and R12 (Eq. 47a) with their corresponding cumulant ratios measured in heavy-ion
experiments it is possible to extract the thermal parameters, namely the freeze-out
temperature, T f , and the freeze-out baryon chemical potential fB . The feasibility
of such a procedure has been demonstrated in Refs. 123,147,148. Fig. 17149 illus-
trates a recent example of such a comparison and subsequent determination of the
freeze-out parameters.
As can be seen from Fig. 17, due to large errors on the experimental results
3
for (SQ Q )/MQ , at present, only an upper limit on the freeze-out temperature
can be determined using the method described above. Thus, a complementary
procedure for the determination of T f , relying on a separate observable that can
be extracted both from heavy-ion experiments and lattice QCD calculations, is
certainly welcome. Recently, such a complementary procedure for the determination
of T f has been proposed in Ref. 116. This procedure takes advantage of the fact
that the initially colliding nuclei in heavy-ion collisions are free of net strangeness.
Thus, the conservation of strangeness under strong interaction ensures that the
QGP medium created during the collisions of these heavy-ions is also strangeness
neutral.
By Taylor expanding the net strangeness density, hnS i (B , S ), in B and S
and subsequently imposing the strangeness neutrality condition, hnS i (B , S ) = 0,
for a homogeneous thermal medium the strangeness chemical potential, S , can be
34 H.-T. Ding, F. Karsch and S. Mukherjee

0.16
19.6
RQ
31
LQCD: T=150(5) MeV
B/T=0.0 RQ
12
STAR
2.5 B/T=1.0
HRG: hadron yield 27
B/T=1.3 0.12

2.0 STAR: s=62.4 GeV

s [GeV]: 39
0.08
1.5
62.4

1.0 200
0.04

0.5
T [MeV]
0.00
0 50 100 B [MeV] 150 200
145 155 165 175 185 195 205 215

Q
Fig. 17. (Left) A comparison between the lattice QCD results149 for R31 and the STAR data138
3 )/M
for (SQ Q Q at s = 62.4 GeV. The overlap of the experimental results with the lattice
QCD calculations provides an upper bound on the freeze-out temperature T f 155 MeV. (Right)
Q
Lattice QCD results149 for R12 as a function of B compared with the STAR data138 for MQ /Q 2
f
in the temperature range T = 150(5) MeV. The overlap regions of the experimentally measured
results with the lattice QCD calculations provide estimates for the freeze-out chemical potential fB

for a given s. The arrows indicate the corresponding values of fB obtained from the statistical
hadronization model fits to experimentally measured hadron yields.86

obtained as116
 
S  2
B B 4
= s1 (T ) + s3 (T ) +O . (48)
B T T
The coefficients s1 , s3 , etc. consist of various generalized baryon, charge and
strangeness susceptibilities defined at vanishing chemical potentials and can be cal-
culated through standard lattice QCD computations at zero chemical potentials.116
Fig. 18 (left) shows the leading order contribution to S /B , i.e. s1 (T ). A compar-
ison of the lattice result with the predictions from the hadron resonance gas model
reveal that the inclusion of only experimentally observed hadrons, as listed by the
Particle Data Group,31 fails to reproduce the lattice results around the crossover
region. Note that, while S /B is unique in QCD, for a hadron gas it depends on
the relative abundances of the open strange baryons and mesons. For fixed T and
B , a strangeness neutral hadron gas having a larger relative abundance of strange
baryons over open strange mesons naturally leads to a larger value of S . Aston-
ishingly, the inclusion of additional, unobserved strange hadrons predicted within
the quark model150,151 provides a much better agreement with lattice results, hint-
ing that these additional hadrons become thermodynamically relevant close to the
crossover temperature.116 As will be discussed in Section 7 , other lattice thermody-
namics studies also indicate that additional, unobserved charm hadrons do become
thermodynamically relevant close to the QCD crossover.152
On the other hand, the experimentally measured yields of the strangeness S
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 35

0.30
(S/B)LO S/B
0.30 T=161 MeV T=170 MeV

cont. est. T=158 MeV T=166 MeV

0.25
0.25 PDG-HRG T=155 MeV T=162 MeV Me
V
QM-HRG 155
T=
. 5 MeV
T=149 MeV 52
0.20 QM-HRG: - BS S
11 / 2 LQCD: T=155(5) MeV T=1
V
LQCD: T=145(2) MeV T=147 MeV 0 Me
N =6: open symbols 0.20 T=15
PDG-HRG T=145 MeV
N =8: filled symbols QM-HRG
0.15 39 GeV (STAR prlim.)
T [MeV] 17.3 GeV (NA57) B/T
0.15
140 150 160 170 180 190 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Fig. 18. (Left) Lattice QCD results116 for S /B at the leading order, i.e. s1 (T ) (see Eq. 48). The
dotted line (PDG-HRG) shows the results of a hadron resonance gas model containing only hadrons
listed by the Particle Data Group.31 The solid line (QM-HRG) depicts the result for a hadron gas
when additional, experimentally yet unobserved, quark model predicted strange hadrons150,151
are included. The shaded region indicates the chiral crossover region Tc = 154(9) MeV. (Right)
A comparison between experimentally extracted values of (fS /fB , fB /T f ) (filled symbols) and
lattice QCD results for S /B (shaded bands).116 The lattice QCD results are shown for B /T =
fB /T f . The temperature range where lattice QCD results match with fS /fB provide the values

of T f , i.e. T f = 155(5) MeV and 145(2) MeV for s = 39 GeV and 17.3 GeV, respectively.

anti-baryon to baryon ratios, RH , at the freeze-out are determined by the thermal

freeze-out parameters (T f , fB , fS ),86
" !#
2fB fS
RH ( s) = exp f 1 f |S| . (49)
T B
By fitting the experimentally measured values of R , R and R , corresponding
to |S| = 1, 2 and 3, the values of fS /fB and fB /T f , as observed in a heavy-ion

experiment at a given s, can easily be extracted. Matching these experimentally
extracted values of fS /fB with the lattice QCD results for S /B as a function of
temperature, one can determine the freeze-out temperature T f . Fig. 18 (right) il-
lustrates this procedure. Once again, the inclusion of additional unobserved strange
hadrons in the hadron resonance gas model (QM-HRG) leads to very similar val-
ues of the freeze-out temperatures as obtained using the lattice data. However,
including only the hadrons listed by the Particle Data Group31 (PDG-HRG) yields
freeze-out temperatures that are 5-8 MeV larger.

6. Transport properties

As we have seen in previous sections the analysis of bulk thermodynamics and charge
fluctuations provides plenty of evidence that thermodynamics of strong-interaction
matter above the crossover transition temperature, Tc , and up to temperatures
of about (1.5 2)Tc is highly non-perturbative. In this temperature range the
36 H.-T. Ding, F. Karsch and S. Mukherjee

velocity of sound is significantly smaller than that in a non-interacting quark gluon

gas suggesting also significant modifications of its transport properties. One of
the striking results of heavy-ion experiments at RHIC and LHC is that strong-
interaction matter in this temperature range shows features of a strongly coupled
medium with a small shear viscosity to entropy density ratio and small diffusion
coefficients. This led to the notion of strong-interaction matter close to the crossover
transition being a nearly perfect liquid.
Transport properties of strong-interaction matter can be explored in lattice QCD
calculations through the analysis of current-current correlation functions, e.g. for
some current J(, ~x),
Z
d
Z
3 ipx
GJ (, p) = d x e hJ (0, 0)J (, x)i = (, p) K(, ), (50)
0 2 J
where the integration kernel K(, ) is
cosh(( 1/2T ))
K(, ) = , (51)
sinh(/2T )
and , are appropriately chosen (scalar, vector, tensor) indices specifying the
current.
In Euclidean space-time these correlation functions have a spectral representa-
tion. The low frequency and low momentum structure of the spectral functions

J (, p) then gives access to transport properties of strong-interaction matter

while at higher frequencies the spectral functions provide information on in-medium
properties of bound states. We will discuss the latter in the Section 7. Extracting
the spectral function from a Euclidean correlation function is, however, difficult.
This is due to the fact that Euclidean correlation functions are calculated in finite
temperature lattice QCD only at a few discrete Euclidean time separations, e.g.
0 N d . It requires the application of inversion methods like the Maximum
Entropy Method (MEM)153 or a modeling of the structure of the spectral function
at low frequencies.
According to the Kubo relations, which express the response of a medium to
small perturbations, transport coefficients can be determined from spectral func-
tions calculated in equilibrium thermodynamics. They are given by the slope of the
corresponding spectral functions at vanishing energy. For instance, using in Eq. 50
the electromagnetic current J (t, x) JV (t, x) = (t, x) (t, x) for a fermionic
charge carrier (quarks) with charge e (e.g. Qd = e/3 for a down quark) one can
extract the electrical conductivity which quantifies the response of the QGP to
small perturbations induced by an electromagnetic field,
3
Cem X ii (, p, T )
= lim lim . (52)
6 0 p0 i=1
d Note that Eq. 51 is symmetric around = 1/2T and as a consequence only half the number

of data points of the Euclidean temporal correlation function, i.e. N /2, provide independent
information to extract the spectral function.
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 37

Here Cem is the sum of the square of the elementary charges of the quark flavor
f , Cem = f Q2f . The electrical conductivity and the vector spectral function are
P

also related to the emission rate of soft thermal photons,

dR 3
lim = (T ) T em . (53)
0 d3 p 2 2
and the contribution of quark-antiquark annihilation to the thermal dilepton rate,
2
dW 5em 1
3
= 3 2 /T
V (, p, T ), (54)
d d p 54 (e 1)
where em is the electromagnetic fine structure constant.
Using the vector current JV (, x) for heavy quarks one obtains from Eq. 50 the
corresponding spectral function of heavy quarkonium in the vector channel, ii V,
which allows to determine the heavy quark diffusion coefficient D,
3
1 X ii
V (, p, T )
D= h
lim lim . (55)
62 0 p0 i=1

Here h2 is the heavy quark number susceptibility which is defined through the
zeroth component of the temporal meson correlator in the vector channel. E.g., for
charm quarks this is just the net charm number susceptibility introduced in Eq. 12.
Similarly, in the light quark sector, the electric charge diffusion coefficient DQ is
defined as the ratio of to electric charge susceptibility Q
2 defined in Eq. 19, i.e.
DQ = /Q 2 .
The heavy quark diffusion coefficient D can be related to the momentum diffu-
sion coefficient and drag coefficient through the Einstein relation
2T 2 T
D= = . (56)
M
It is used to describe the Brownian motion of a heavy quark in the hot medium and
can also be related to the ratio of shear viscosity over entropy density /s.154
The shear () and bulk () viscosities of strong-interaction matter are extracted
from correlation functions of the energy-momentum tensor, T = F F
1
4 F F . I.e. using the current J (, x) T (t, x) in Eq. 50 one obtains
12,12 (, p)
= lim lim , (57)
0 p0
X kk,ll (, p)
= lim lim . (58)
9 0 p0
k,l

At present only a few calculations of transport coefficients have been performed

in QCD with dynamical quark degrees of freedom. Most calculations utilized the
quenched approximation at vanishing baryon number density.155 In the following
subsections we will summarize current results for the transport coefficients intro-
duced above. For further details on the calculation of transport coefficients on the
lattice and earlier results see, for instance, Ref. 156.
38 H.-T. Ding, F. Karsch and S. Mukherjee

6.1. Electrical conductivity, charge diffusion and dilepton rates

Probably the best analyzed spectral function is that of the light quark vector current
which gives access to the electrical conductivity, (T ) (Eq. 52), soft photon emission
rates (Eq. 53) and thermal dilepton rates (Eq. 54) as well as the electric charge
diffusion constant DQ = /Q 2 . The vector spectral function V () V (, 0, T )
has been calculated first in quenched QCD157 using the MEM approach153,158 in
which the analysis had only been constrained at large frequencies through a default
model based on the free fermion spectral function and the largest lattices used only
had N = 16 points in the temporal direction. It has, however, been noticed159
that this is not sufficient to get access to transport coefficients, which require the
spectral functions at low frequencies to be linear in . Furthermore, a larger time
extent160 is needed to control the low frequency part of the spectral functions.
First continuum extrapolated results for the vector spectral function were ob-
tained in quenched QCD using clover-improved Wilson fermions160 at T ' 1.4 Tc .
These results have been extended to three different temperatures recently.161163
In these calculations the thermal spectral functions are obtained by fitting the
continuum-extrapolated two-point Euclidean correlation functions to an ansatz for
the spectral function that is motivated by kinetic theory and perturbation theory
3  
model () = q cBW 2 + (1 + k) 2
tanh . (59)
+ 2 /4 2 4T

Here q is similar to h2 appearing in Eq. 55 but for light quarks and it is obtained
by summing the zeroth component of the light temporal vector correlation function
over space-time coordinates. The unknown parameters are thus the amplitude cBW
and the width of the Breit-Wigner function as well as the parameter k, which
parametrizes deviations from the free theory behavior at large . The thermal
spectral functions obtained in this way and the resulting thermal dilepton rates
are shown in Fig. 19. In the high frequency region, i.e. for /T & 7, all the
thermal spectral functions at the three temperatures above Tc are well described
by leading order perturbation theory (Born rate) as well as Hard Thermal Loop
(HTL) calculations. At small frequencies the lattice QCD results are significantly
enhanced over the Born rate, indicating the presence of a transport peak, but are
smaller than the HTL rates. In fact, the latter rises too rapidly at low frequency
and would give rise to a divergent electrical conductivity.
The finite intercept of ()/(T ) at = 0 seen in Fig. 19 gives the electrical
conductivity (see Eq. 52), /(Cem T ) [0.21, 0.44].163 In these calculations /T
does not show any significant temperature dependencee , which may be due to the use
of the quenched approximation. The results are consistent with a calculation that
used staggered fermions, but no renormalized vector currents,159 /(Cem T ) ' 0.4(1)
for 1.5 . T /Tc . 2.25. These results are, however, substantially smaller than earlier
calculations performed on coarse lattices.164
e This is also reflected in the temperature dependence of the temporal correlation functions divided
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 39

1.0e-05
4.0 1.1Tc
dW/dd3p
1.0e-06 1.2Tc
/(T)
cut(0/T=3.0)/(T) 1.4Tc
3.0 HTL 1.0e-07 HTL
Born
1.0e-08
2.0
1.0e-09
1.0
1.0e-10
/ /T
0.0 1.0e-11
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8

Fig. 19. (Left) Thermal spectral function at 1.1 Tc calculated in quenched QCD.163 (Right)
Temperature dependence of the quark-antiquark annihilation contribution to the thermal dilepton
rate above Tc .163

T /Tc
0.75 1.00 1.25 1.50 1.75 2.00
0.6 2.5
(T) Nf=0
0.5 CemT (Aarts et al.)
2.0
Nf=0
0.4 (Ding et al.)
Nf=2 1.5
2T DQ

0.3 (Brandt et al.)

Nf=2+1 1.0
(Aarts et al.)
0.2

0.5
0.1

0 T/Tc 0.0
100 150 200 250 300 350 400
T [MeV]
0.5 1 1.5 2 2.5 3 3.5

Fig. 20. (Left) Lattice QCD results for the electrical conductivity /(Cem T ) as function of T /Tc
including results obtained in quenched QCD,159,160,163 2-flavor QCD with m ' 270 MeV165
and (2+1)-flavor QCD with m ' 380MeV.166 Note that the values of Tc obtained in these
N =0 N =2 N =2+1
calculations are different, i.e. Tc f ' 270 MeV, Tc f ' 208 MeV and Tc f ' 185 MeV.
(Right) Temperature dependence of charge diffusion coefficient 2T DQ = 2T /Q 2 . The results
are obtained in (2+1)-flavor QCD calculations on anisotropic lattices at one value of the spatial
lattice cut-off, a ' 0.12 fm and with light quark masses corresponding to m ' 380 MeV.166

Recently the calculation of the electrical conductivity has also been performed
in QCD with dynamical quarks. The first calculation was done in 2-flavor QCD
using O(a) improved Wilson quarks corresponding to m ' 270 MeV. The electrical
conductivity of the quark-gluon plasma at T ' 250 MeV was found to be similar to
the quenched result, /(Cem T ) = 0.40(12).165
First results from a calculation in (2+1)-flavor QCD performed on anisotropic
lattices with quark masses corresponding to a pion mass m ' 380 MeV166,167 are
shown in Fig. 20. The simulation is performed at one value of the lattice cut-off.
by T 3 which are temperature independent for this temperature range.163
40 H.-T. Ding, F. Karsch and S. Mukherjee

Varying the temporal extent of the lattice, N , from between 48 and 16 a tempera-
ture range from 0.63 Tc to 1.90 Tc is covered. The left hand panel of this figure shows
temperature dependence of /(Cem T ) (square points). It stays constant within er-
rors in the transition region and increases with temperature for T > Tc . Also shown
in the left hand panel of Fig. 20 are electrical conductivities obtained in quenched
QCD159,160,163 and 2-flavor QCD.165 The right hand panel of Fig. 20 shows the tem-
perature dependence of charge diffusion coefficient DQ multiplied by 2T . 2T DQ
shows a dip near Tc which arises from the rapid exponential drop of the electric
charge susceptibility at low temperature wherethe electrical conductivity of a pion
gas168 remains nonzero and vanishes only like T .

6.2. Heavy quark diffusion

Contrary to earlier expectations it has been found in heavy ion collision experiments
at RHIC169,170 and LHC171 that heavy-quark mesons show a substantial elliptic
flow that is comparable to that of light-quark mesons. Moreover, also somewhat
unexpectedly, heavy quarks are found to lose a significant amount of energy while
traversing through the QGP. The latter has been called the heavy quark energy loss
puzzle. Phenomenological explanations of these phenomena try to model the heavy
quark diffusion in a hot and dense medium. This requires knowledge about the
heavy quark diffusion coefficients D154,172,173 which can be determined in lattice
QCD calculations.
The charm-quark diffusion coefficient has been calculated in quenched QCD at
three temperatures in the deconfined phase174 using lattices with temporal extent
N = 48, 32 and 24 at a fixed value of the cut-off. Results from this calculation are
shown in Fig. 21. The left hand panel shows the transport peak in the charmonium
vector spectral function at three different temperatures above Tc . These transport
peaks are obtained using a MEM analysis where an ansatz of Lorentzian form is
used as the prior information for the low frequency part of the spectral function, i.e.
/( 2 + 2 ). The intercept at /T = 0 gives the charm quark diffusion coefficient
(c.f. Eq. 55). It is found that 2T D ' 2 and, with current understanding of
uncertainties, this product is independent of temperature. These values of D are
much smaller than the results from perturbative calculations175 and are very close
to the value obtained from the AdS/CFT correspondence.176 These different results
are compared in the right hand panel of Fig. 21.
A direct determination of the diffusion coefficient for bottom quarks from bot-
tomonium correlation functions is difficult as it would require quite small lattice
spacings to accommodate the bottomonium states on the lattice. The method of
choice here is to start with the infinite quark mass limit where one can use Heavy
Quark Effective Theory (HQEFT) to compute the diffusion coefficient of a static
quark. In this approach the propagation of a heavy quark carrying color charge
and its response to a colored Lorentz force can be described using linear response
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 41

16
2TD charm quark
14
(Ding et al.)
12 static quark
(Banerjee et al.)
10 NLO pQCD
s0.2 static quark
8 (Kaczmarek et al.)
6 AdS/CFT

4
2
0
1 1.5 2 2.5 3
T/Tc

Fig. 21. (Left) The low frequency transport part of the charmonium vector spectral function
obtained from a MEM analysis of vector correlation functions in quenched lattice QCD.174 The
central lines are mean values while the bands reflect the statistical uncertainties. (Right) Temper-
ature dependence of heavy quark diffusion coefficients multiplied by 2T in pure gluonic matter.
The charm diffusion coefficients are obtained from the spectral functions shown in the left plot.174
Also shown are results for the static quark diffusion coefficient obtained by using the lattice dis-
cretized versions of HQEFT.177,178 The boxes show the statistical error while the error bars reflect
the systematic uncertainties. The horizontal dotted line labels the value of 2T D in the heavy
quark limit from the AdS/CFT correspondence 176 and the short, horizontal solid line indicates
the value of 2T D from next-to-leading order pQCD calculations at s ' 0.2.175
.

theory. This leads to the spectral analysis of a color-electric correlator179,180

3
1 X hReTr [U (1/T ; )gEi (, 0) U ( ; 0)gEi (0, 0)]i
GE ( ) = , (60)
3 i=1 hRe Tr [U (1/T ; 0)]i
where U (2 ; 1 ) is a Wilson line in the Euclidean time direction and gEi (, x) denotes
the color-electric field at time-space (, x). The momentum diffusion coefficient
can then be extracted from the slope of the corresponding spectral function in the
limit of vanishing frequency ,
2 E ()
/T 3 = lim . (61)
T 2
0

Here E () is the corresponding spectral function that is related to GE ( ) via

Eq. 50. In the non-relativistic limit, i.e. for a heavy quark mass M  T , the
momentum diffusion coefficient is related to the (spatial) heavy quark diffusion
coefficient D via Eq. 56.
The color-electric correlator GE ( ) is defined in terms of gluonic observables
only. It thus suffers from a small signal to noise ratio, a problem similar to what
one encounters in the calculation of viscosities181 (see next subsection). In both
cases efficient noise reduction techniques are needed for a calculation of the gluonic
correlation functions. In a recent study177 the correlator GE ( ) has been calculated
using multi-level updates181,182 as well as the link integration technique.183,184 The
improved correlator Gimp ( T ) obtained this way at T ' 1.4 Tc for four different
42 H.-T. Ding, F. Karsch and S. Mukherjee

4 4
Gimp/Gnorm Gimp/Gnorm
continuum 3.5
3.5 19233x48
144 x36 3
9633x24
3 64 x16 2.5
NNLO
NLO
2.5 2

1.5 continuum
2 /T3=2.5(4)
1 A*NNLO+B*3
NNLO
1.5 /2T
0.5
T T
1 0
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5

Fig. 22. (Left) Continuum extrapolation of improved color-electric correlator GE ( ) Gimp ( )

(see Eq. 60) normalized to the high temperature, free field theory result Gnorm ( ) and the
comparison with NNLO and NLO perturbative QCD calculations at 1.4 Tc on quenched lat-
tice QCD.177 (Right) 2 fit to the continuum extrapolated Gimp /Gnorm using an ansatz
model 3

E () = max A N N LO () + B , 2T . Also shown are the separate contributions to the
correlator arising from the different parts of the fit ansatz for the spectral function.177

values of the lattice cut-off in quenched QCD has been extrapolated to the contin-
uum limit. Results are shown in the left hand panel of Fig. 22. In order to extract
the momentum diffusion coefficient from this correlator an ansatz for the spectral
function has been used.177 The corresponding fit is shown in the right hand panel
of Fig. 22. It yields,

/T 3 = 2.5(4) at T 1.4 Tc . (62)

This corresponds to 2T D = 5.0(8) which is consistent with similar lattice QCD

studies using HQEFT.178 However, it is a factor 2 to 3 larger than the charm quark
diffusion coefficient discussed above174 (see Fig.21).
The heavy quark diffusion coefficient D has also been calculated in perturbative
QCD in both leading and next-to-leading order as well as from the AdS/CFT corre-
spondence. At s 0.2 the leading order pQCD calculation185 gives 2T D 71.2
while 2T D 8.4 is obtained in next-to-leading order calculations.175 In the strong
coupling limit 2T D = 1 is obtained from the AdS/CFT correspondence176 (see
Fig.21).

6.3. Shear & bulk viscosities

The computation of shear and bulk viscosities has been attempted already in the
80s.186 However, it turned out that the signal to noise ratio in calculating Euclidean
energy-momentum correlation functions is very small and high statistics or improved
algorithms are needed. The latter became available for quenched QCD calculations
with the multi-level algorithms182 which have been used for the calculation of shear
and bulk viscosities.181,187 Results from such an analysis by Meyer181 and those
with direct high statistics calculations by Nakamura & Sakai188 are shown in Fig. 23.
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 43

The ratio obtained in the former calculation are smaller than those obtained in the
earlier direct simulations; /s = 0.102(56) at 1.24 Tc and 0.134(33) at 1.65 Tc
obtained in quenched QCD on lattices with temporal extent N = 8.181 This shows
that noise reduction techniques like the multi-level algorithm are mandatory for a
successful calculation of viscosities. Results on the bulk viscosity to entropy ratio
/s show that this ratio rapidly becomes small above Tc .189 At T & 1.2 Tc it is
smaller than /s and, in fact, within errors it is consistent with zero.181,188,189
Still the calculations of viscosities are performed on lattices with rather small
temporal extent compared to those used in calculations of the electrical conductivity
and diffusion constants. Systematic uncertainties in these calculations need to be
better controlled in future. Here it will also be helpful to combine lattice QCD
calculations with information from analytic approaches that put constraints on the
structure of spectral functions, e.g. QCD sum rules.190192

0.7
/s Nakamura & Sakai 2 /s
0.6 Meyer
KSS bound: 1/4
0.5 1.5

0.4
1
0.3

0.2 0.5

0.1
0
0
1 1.5 2 2.5 3 1 1.2 1.4 1.6 1.8 2
T/Tc T/Tc

Fig. 23. (Left) Temperature dependence of the shear viscosity, /s, of gluonic matter obtained by
using noise-reduction techniques (squares)181 and direct high statistical calculations (circles)188
on N = 8 lattices. The dotted line corresponds to the conjectured lower bound of /s from
AdS/CFT correspondence.176 (Right) Temperature dependence of the bulk viscosity, /s, of
gluonic matter.189 The square points include the statistical uncertainties while the solid black
bars denote the systematic uncertainties. Results for the same quantity from Ref. 188 which are
not shown are consistent with zero at T [1.4 Tc , 1.8 Tc ]. Data are taken from Refs. 181,188,189.

6.4. Transport coefficients of second order hydrodynamics

Due to the importance of viscous effects in the evolution of hot and dense matter
created in heavy ion collision there is considerable interest in extending the hydrody-
namic modeling beyond leading order gradient expansions of the energy momentum
tensor. A second order gradient expansion is parameterized by additional transport
coefficients,193 which may become accessible to lattice QCD calculations.194196
One of these new transport coefficientsf is , which controls the momentum depen-
f Although this particular second order coefficient is also called it should be noted that it is not

related to the momentum diffusion coefficient introduced in Eq. 56

44 H.-T. Ding, F. Karsch and S. Mukherjee

dence of space-like correlations of the energy momentum tensor. It is defined as

the leading order coefficient of the Taylor expansion of the retarded correlator of
energy-momentum tensor in momentum p about zero,192,197
Z
GR ( = 0, p) = dt dx eipx h[T12 (, x), T12 (0, 0)]i(t)

2
= GR (0) + |p| + O(|p|4 ) . (63)
2

Since the difference between the retarded correlator GR and the Euclidean correlator
at vanishing frequency = 0 is just a contact term, which has been shown to be
momentum independent,198 can directly be extracted from the corresponding
Euclidean correlator,

dGE ( = 0, p)
= 2 lim . (64)
|p|0 d|p|2

This is considerably simpler to compute than first order coefficients such as the shear
or bulk viscosity as it does not require analytic continuation from imaginary time
to real time. The only complication is that large spatial lattice sizes are required
to get access to small momenta in the lattice QCD calculation.

N = 8
1

N = 6 N = 6
N = 6
0.5
/T 2

0 = 7.1
= 6.68
= 6.14
-0.5 AdS/CFT
LPT

1 2 3 4 5 6 7 8 9 10 11
T /Tc

Fig. 24. Transport coefficient of second order hydrodynamics /T 2 as a function of T . The lines
denote the results from AdS/CFT correspondence197 and lattice perturbation theory (LPT),199
respectively.

The second order hydro-coefficient has been computed in quenched QCD on

lattices with temporal extent N = 8 and 6 and a large spatial size.199 Results from
this calculation are shown in Fig. 24. It yields /T 2 = 0.36(15). Within the current
large errors no further temperature dependence is observed in the temperature range
2Tc < T < 10Tc and results are compatible with both perturbative QCD192,195 and
AdS/CFT results.200
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 45

7. Open heavy flavors and heavy quarkonia

Owing to their large masses heavy flavors, such as charm and bottom, are produced
only during the earliest stages of a heavy ion collision and get affected by the hot and
dense medium during its entire evolution. They then emerge as hadrons with open
heavy flavors or as quarkonia. Thus, heavy flavors provide unique identifiable probes
against which the temperature and coupling of the medium created during heavy-
ion collisions can be calibrated. The melting of charmonium states has long been
considered as an important signature for the formation of quark-gluon plasma.18
Understanding its production rates and abundance can provide detailed information
on properties of the matter formed in a heavy ion collision. We will discuss current
results on the temperature dependence of quarkonium bound states in subsection
7.2. In addition, understanding the fate of open flavor bound states also is of
importance for the understanding of the strongly interacting regime in the QGP
close to the transition temperature. The question whether or not heavy flavor
mesons can survive above Tc plays a crucial role in the phenomenological modeling
of the heavy quark energy loss.201203 Lattice QCD results on the dissociation of
open charm hadrons will be discussed in the next subsection.

7.1. Melting and abundances of open charm hadrons

Similar to the case of strangeness,98 discussed in Section 3.2, the deconfinement or

melting of open charm hadrons can be probed152 by using fluctuations of the charm
quantum number (C) and its correlations with baryon number (B) fluctuations.
To this end one may consider appropriately chosen ratios of correlations between
net-baryon and net-charm number fluctuations. One such ratio is, for instance,
BC BC 152
22 /13 . Since at low temperatures, for a gas of uncorrelated charmed hadrons,
the thermodynamic contributions of the much heavier |C| = 2, 3 baryons can be
safely neglected, this particular ratio is dominated by |C| = 1 baryons and thus
yields BC BC
22 /13 = |B| = 1. On the other hand, at high temperatures, for a free
gas of massive charm quarks with masses much larger than T , the above ratio
isg BC BC
22 /13 = |B| = 1/3. Thus, this quantity can be viewed as a measure for
the baryon number associated with the predominant degrees of freedom that are
carriers of the charm quantum number in a given temperature region. Its behavior
can indicate the liberation of quark-like degrees of freedom with fractional baryon
number.
Fig. 25 (left)152 shows that close to the chiral crossover temperature, Tc =
154(9) MeV, the quantity BC BC
22 /13 starts deviating from unity. This indicates
that beyond this temperature fractionally charged quark-like degrees of freedom
start appearing and open charm hadrons start to melt. The left hand panel of
g Here one makes use of the fact that the Boltzmann gas approximation is still a good approximation
for a charm quark gas at a few times the transition temperature. This is the temperature range
of interest to heavy ion collision experiments.
46 H.-T. Ding, F. Karsch and S. Mukherjee

Fig. 25 also shows a comparison with similar quantities involving both baryon-
strangeness correlations as well as baryon-electric charge correlations. The first of
these quantities is sensitive to the deconfinement of strange quarks and the second
one is also sensitive to the deconfinement of light up and down quarks. It is clear
from the figure that the onset of deconfinement for the charm, strange as well as
the up/down quarks happens in the same chiral crossover region.

0.5 BC C BC
13 /( 4 - 13 )
un-corr. 0.4
1.0 hadrons QM-HRG-3
BC BC
22 /13 0.3
QM-HRG
PDG-HRG
0.8 BS BS
31 /11 0.2
BQC QC BQC non-int.
BQ BQ 0.5 112 /( 13 - 112 )
31 /11
quarks
0.6
0.4

0.4 0.3

BSC SC BSC
0.7 - 112 /( 13 - 112 )
0.2
0.5 N :8 6

0.0 T [MeV]
0.3 T [MeV]

140 180 220 260 300 340 140 150 160 170 180 190 200 210

Fig. 25. (Left) Ratios between cumulants of correlations of net baryon number with net charm,
net strangeness and net electric charge fluctuations.152 Deviations of these ratios from unity in
the crossover region (shaded band) suggest that the onset of the melting/deconfinement of open
charm hadrons as well as open strange hadrons and hadrons with light up/down quarks starts in
the same temperature range (see text for details). (Right) Thermodynamic contributions of all
charmed baryons (top), all charged charmed baryons (middle) and all strange charmed baryons
(bottom) relative to that of corresponding charmed mesons.152 The dashed lines (PDG-HRG) are
predictions for an uncorrelated hadron gas using only the PDG states31 . The solid lines (QM-
HRG) are similar HRG predictions including also the states predicted by the quark model.204,205
The dotted lines (QM-HRG-3) are the same QM predictions, but only including states having
masses less than 3 GeV.

The right hand panel of Fig. 25 shows ratios of generalized susceptibilities that
are constructed such that in a gas of uncorrelated hadrons the numerator would
correspond to the partial pressure of baryons with quantum numbers selected by
the other indices, while the denominator would correspond to the partial pressure of
corresponding mesons. E.g. BC 13 will give the partial pressure of charmed baryons
and C4 BC
13 filters out the corresponding partial pressure of charmed mesons.
The ratios 13 /(4 13 ), BQC
BC C BC QC BQC BSC SC
112 /(13 112 ) and 112 /(13 112 )
BSC

thus correspond to ratios of partial pressures arising from charmed baryons to that
from open charm mesons, charged-charmed baryons to open charm charged mesons
and strange-charmed baryons to strange-charmed mesons, respectively. As can be
seen in Fig. 25 (right) results for these ratios at low temperatures are well described
by a hadron gas that uses open charm resonances obtained in quark model calcu-
lations (QM-HRG) while, not surprisingly, it differs strongly from a simple hadron
resonance gas based only on the very few experimentally known charmed hadrons
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 47

(PDG-HRG). This observation is similar to that made for strange hadron fluctu-
ations in Section 5 (right panel of Fig. 18). These observations clearly provide
evidence for contributions from experimentally yet unobserved charmed/strange
hadrons to charm and strangeness fluctuations in the vicinity of the crossover tran-
sition temperature. In additions, the agreement between lattice QCD data and QM-
HRG model calculations below Tc and the onset of deviations just in the transition
region also suggests that open charm hadrons start to dissociate in the crossover
region. This conclusion is consistent with that drawn from the temperature depen-
dence of the ratio in the left hand panel of Fig. 25, which has the advantage of being
independent of details of the charmed hadron spectrum.

7.2. Heavy Quarkonia

Inside the QGP the interaction between a heavy quark anti-quark pair gets weak-
ened due to the screening effects of the intervening deconfined colored medium.
This telltale signature of the presence of a color deconfined medium is expected to
be manifested through melting of heavy quarkonium states,18 i.e. bound states of
a quark anti-quark pair. Melting of heavy quarkonium states in the QGP is also
expected to follow a distinctive sequential pattern with the smallest, most tightly
bound quarkonium state surviving up to the highest temperature, effectively serv-
ing as a thermometer to probe the temperature of the medium created at RHIC
(for a recent review see Ref. 206). Since quarkonia do not carry flavor quantum
numbers, unlike the open heavy flavor hadrons, their melting cannot be accessed
in lattice QCD calculations through quantum number correlation based studies as
discussed in Section 7.1. Currently, at least three approaches are being actively
pursued to study thermal modifications of heavy quarkonia on the lattice. The
first approach is to use potential models with heavy quark potentials computed on
the lattice. The second approach relies on the extraction of quarkonium spectral
functions from Euclidean temporal correlators. The third approach involves spa-
tial correlation functions of quarkonia and the study of their in-medium screening
properties. Here, we summarize the recent progresses and developments concerning
these three approaches. More details about earlier lattice results can also be found
in Refs. 206208.

7.2.1. Heavy quark potential

Properties of heavy quarkonia in the vacuum have been successfully described by the
potential model approach. In this approach, the interaction between a heavy quark
pair forming the quarkonium is described by an instantaneous potential.209,210 Due
to its success at zero temperature, the potential model approach has also been ap-
plied at nonzero temperature, by making the potential between the heavy quarks
temperature dependent.211 The temperature dependent potentials used in these
calculations are based either on model calculations or on finite temperature lattice
48 H.-T. Ding, F. Karsch and S. Mukherjee

QCD results for the free energy and the internal energy. However, justification
for the use of these potentials in a Schrodinger equation is mostly based on phe-
nomenological arguments and has no rigorous connections to the real-time evolution
of heavy quarkonia described by the Schrodinger equation.
Recently, a non-relativistic effective theory approach at nonzero temperature,
relying on a particular hierarchy of the relevant scales, has been developed.212214
Generally by integrating out the hard energy scale, i.e. the rest mass of the heavy
quark m, Non-Relativistic QCD (NRQCD) effective theory215,216 is obtained, and
by further integrating out the soft scale, i.e. the typical momentum exchange be-
tween the bounded quarks of O(mv), so-called potential non-relativistic QCD (pN-
RQCD), is obtained.217 In pNRQCD a heavy quark bound state can be described
by a two-point function satisfying a Schrodinger equation,212214
 
> 1 2 >
it DN (t, ~
r ) = + V (t, r) DN R (t, ~
r), (65)
R
m r
>
where DN R (t, ~
r) is the real-time forward heavy quark pair correlation function in
the non-relativistic limit. The potential V (t, r) in such a Schrodinger equation is
well defined and turns out to be complex valued.212214 Its real part reflects color
Debye screening effects while the imaginary part is related to Landau-damping, i.e.
the scattering of quarks with the constituents of the medium and the absorption
of gluons from the medium via singlet-octet transition.214 It has been shown in
Ref. 218 that in the limit m the leading part of the potential V (t, r) can be
obtained as
it W (t, r)
V (r) = lim , (66)
t W (t, r)

where W (t, r) is the real-time thermal Wilson loop. Through a Fourier transforma-
tion one obtains,
Z +
W (t, r) = d (, r) eit , (67)

where (, r) is the spectral function of the real-time Wilson loop. The above
equation can be analytically continued to the Euclidean time, giving
Z +
W (, r) = d (, r) e . (68)

In Eq. 68 the Wilson loop, W (, r) is the usual Euclidean-time Wilson loop which
can be obtained from lattice QCD calculations. As if there exists a well defined
lowest lying peak structure in the spectral function (, r) it would dominate the
dynamics in the Wilson loop in the late time limit thus giving to the complete
information of the heavy quark potential as shown in Eq. 66. However, as argued in
Ref. 219 the late time dynamics is not completely separated from the early time non-
potential physics and to take these effects into account a skewed Lorentzian function
including the early time contribution for (, r) is required to fit the Wilson loop.
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 49

0.4
5.5 210 MeV 315 MeV 503 MeV
252 MeV 360 MeV 629 MeV
5 0.35
280 MeV 419 MeV 839 MeV

4.5 0.3

4 0.25
Re[V] [GeV]

Im[V] [GeV]
3.5 0.2

3 0.15

2.5 0.1

2 839 MeV 419 MeV 280 MeV

0.05
629 MeV 360 MeV 252 MeV
503 MeV 315 MeV 210 MeV
1.5 0
0 0.1 0.2 0.3 0.4 0.5 0.6 r[fm] 0 0.5 1 1.5 2 2.5 3 r[fm]

Fig. 26. Real (left) and imaginary (right) parts of the heavy quark potential calculated in
quenched lattice QCD.221 Values are shifted for better visibility. Gray circles denote color singlet
free energies and solid lines are leading order hard thermal loop results.

The position and width of this skewed Lorentzian peak are connected to the real and
the imaginary parts of the potential, respectively. Thus, by extracting the spectral
function, (, r), from the Euclidean-time Wilson looph , W (, r), obtained from
lattice QCD calculations and by subsequently fitting this (, r) with a skewed
Lorentzian ansatz the real and the imaginary part of the heavy quark potential,
V (r), can be obtained.
This procedure has been tested using the Wilson loop calculated using hard
thermal loop perturbation theory220 as well as for the pure gauge theory.221 Fig. 26
shows the real (left) and imaginary (right) parts of the heavy quark potential in the
pure gluonic medium.221 The real part of the heavy quark potential turns out to be
close to the heavy quark singlet free energy, defined through the spatial correlation
function of the Polyakov loop and its conjugate,222 at all temperatures. At high
temperatures the imaginary part of the heavy quark potential is close to the leading
order hard thermal loop result, but at low temperatures it lies below hard thermal
loop result.
Using the same procedure the real part of the heavy quark potential has also been
extracted in (2+1)-flavor QCD221 on lattices with temporal extent of only N = 12.
Of course, on such small lattices it is hardly possible to extract reliable information
on the imaginary part of the heavy quark potential. Nonetheless, similar to the
quenched limit result the real part of the heavy quark potential is found to be close to
the singlet free energies. However, string breaking is not observed at T < Tc 174
MeV up to 1.2 fm. This could be due to the large pion mass (m 300 MeV)
employed in this exploratory QCD calculation.
The heavy quark potential has also been extracted from the Wilson line correla-
tion function calculated on 483 12 lattices at T = 250 MeV and 305 MeV using a
fit ansatz motivated by hard thermal loop calculations.223 These calculations used
h As illustrated in Ref. 220 in order to avoid the cusp divergence and the large noise to signal ratio
in computations of Wilson loops, some other observables, e.g. Wilson line correlation functions
defined in the Coulomb gauge, can be a substitution and are actually used in lattice calculations.
50 H.-T. Ding, F. Karsch and S. Mukherjee

HISQ action with quark masses that would correspond to m 160 MeV in the
continuum limit. In this calculation the real part of the potential was found to
be equal or larger than the singlet free energies at these two temperatures and it
always was smaller than the zero temperature potential. Moreover, the imaginary
part is found to be of similar size as predicted by leading order hard thermal loop
perturbation theory.223

7.2.2. Spectral functions of quarkonia

The spectral function of quarkonium can be extracted from lattice QCD calculations
of the two-point correlation functions of the heavy quark currents in the Euclidean
time
Z    
G(, T ) = d ~x3 ~ ~
q(, ~x) q(, ~x) q(0, 0) q(0, 0) , (69)

where q represents the heavy quark field. are Dirac matrices defining the spin
structure of quarkonium, in particular = 1, 5 , , 5 corresponds to the scalar,
pseudo-scalar, vector and axial-vector quarkonium states, respectively. Signatures
of medium modification and melting of quarkonium are reflected in the structure of
the spectral density (, T ). It can be obtained from the quarkonium correlator in
the Euclidean-time
Z
d cosh ( ( 1/2T ))
G(, T ) = (, T ) . (70)
0 2 sinh (/2T )
While (, T ) is a continuous function of the frequency over an infinite range,
G(, T ) is calculable on the lattice only at a finite number of discrete points along
the temporal direction due to the finiteness of the lattice. As has been discussed in
Section 6 inverting Eq. 70 to get spectral function (, T ) from the correlation func-
tion G(, T ) is a typical ill-posed problem and a suitable inversion method is needed
to solve the problem. The commonly used inversion methods for the extractions of
quarkonium spectral functions is the Maximum Entropy Method (MEM).153 Vari-
ants of this have recently been suggested which include a modified MEM with an
extended search space224,225 and a Bayesian method which is analogous to MEM
but uses a different prior distribution.226
The charmonium spectral functions at finite temperature have been studied
extensively for quenched lattice QCD.228232 Recently quenched lattice QCD calcu-
lations227 have been performed using lattices with large temporal extents, leading
to quite reliable extractions of the charmonium spectral functions. Fig. 27 illus-
trates the spectral functions of J/ (left) and c (right) for various temperatures.
These studies suggest that all charmonium states are dissociated for T & 1.5Tc in a
gluonic plasma.227 Investigations on charmonium spectral functions in lattice QCD
calculations with dynamical quarks lead to consistent conclusions.233,234
Lattice QCD calculations for bottom quarks are technically challenging since
present computational limitations do not permit calculations with sufficiently fine
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 51

Fig. 27. Temperature dependence of J/ (left) and c (right) spectral functions obtained from
quenched lattice QCD calculations.227

lattice spacings that are much smaller than the inverse of the heavy bottom quark
mass. Large discretization errors arise from the large heavy-quark bare mass in
units of the lattice spacing, mb a. Nonetheless, although it is computationally very
demanding relativistic bottomonium correlations functions have been computed re-
cently in quenched QCD using isotropic lattices with N =192 and large temporal
extent,236 N = 96 and 48 at the same value of the lattice spacing, a ' 0.01 fm,
i.e. mb a ' 0.2. The temporal lattice sizes correspond to temperatures T = 0.7 Tc
and 1.4 Tc . In this calculation it has been found that at 1.4 Tc only the bottomo-
nium correlation function in the vector channel show significantly smaller thermal
modifications than the charmonium correlation function in the same channel.236
To circumvent the problem of large cut-off effects in calculations with bottom
quarks without performing calculations at extremely small lattice spacings, one
needs to adapt a lattice discretization scheme for heavy quarks that is capable
to describe heavy quark physics in some chosen kinematic regime also at mod-
erately small values of the cut-off. One such approach is Non-Relativistic QCD
(NRQCD)237,238 where the heavy quark mass is assumed to be much larger than
the inverse lattice spacing, but the momentum dependence of the heavy quark en-
ergy is included in the non-relativistic limit. NRQCD does not possess a proper
continuum limit because the radiative corrections to coefficients of the NRQCD La-
grangian diverge as mb a 0. However, since the non-relativistic quark field is not
compactified along the Euclidean-time direction, the relation between Euclidean-
time quarkonium correlation functions and the corresponding spectral functions
becomes simpler,
Z
d
G(, T ) = (, T ) e , (71)
0 2
giving rise to a temperature independent integration kernel, exp( ). Moreover,
the transport peak is also absent in the NRQCD spectral function, which makes its
determination much easier.
52 H.-T. Ding, F. Karsch and S. Mukherjee

12 0.84 0.95
10 T /Tc = 0.76 0.95 1.09
0.84
8
6
4
2
(!)/m2b

0
5 1.09 1.27 1.52
1.27 1.52 1.90
4
3
2
1
0
9 10 11 12 13 14 15 9 10 11 12 13 14 15 9 10 11 12 13 14 15
! (GeV)

0.4 b1 0.84 0.95

T /Tc = 0.76 0.95 1.09
0.3 0.84
0.2
0.1
(!)/m4b

0
0.4 1.09 1.27 1.52
1.27 1.52 1.90
0.3
0.2
0.1
0
9 10 11 12 13 14 15 9 10 11 12 13 14 15 9 10 11 12 13 14 15
! (GeV)

Fig. 28. Temperature dependence of spectral functions for the (top) and b1 states (bottom).
Results are obtained by using a lattice discretized form of NRQCD.235

Treating the bottom quark within the NRQCD framework the fate of bottomo-
nium states has been studied on anisotropic lattices using Nf = 2 and 2 + 1 of
improved Wilson quarks with m 400 MeV and m 500 MeV, respec-
tively.235,239242 Fig. 28 shows the temperature dependence of the (top) and
b1 (bottom) spectral functions extracted using the MEM.235 This indicates that
the S-wave ground state survives up to 1.9 Tc while the P-wave ground state
b1 melts just above Tc .235 However, a recent lattice QCD study243 that uses a
different Bayesian method suggested in Ref. 226 for the extraction of the spectral
functions finds that the P-wave ground state b1 may survive up to 1.6 Tc . This
analysis, however, has been performed on lattices with a rather small temporal
extents, N = 12.
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 53

7.2.3. Spatial correlation functions of quarkonia

Similarly as the temporal correlation functions spatial quarkonium correlation func-
tions are defined as the sum of heavy quark current-current correlation functions
instead over a funny space of , x and y
Z 1/T Z    
G(z, T ) = d dx dy q(, ~x) q(, ~x) q(0, ~0) q(0, ~0) , (72)
0

are also related to the quarkonium spectral functions as follows

Z
dpz ipz z d (, pz , T )
Z
G(z, T ) = e . (73)
2 0 2
The spatial correlation functions, however, are sensitive also to the momentum de-
pendence of spectral functions which is reflected in an additional integration over
the momentum pz along the z-direction. Despite of its more complicated connec-
tion to the spectral function compared to that of temporal correlators (see Eq.
70), the spatial correlation function has some distinct advantages. Since the spa-
tial separation is not limited by the inverse temperature, the spatial correlation
function can be studied at separations larger than 1/T and thus is more sensitive
to in-medium modifications of quarkonia. Furthermore, in contrast to the tem-
poral correlation functions, the spatial correlation functions can be directly com-
pared to the corresponding vacuum correlation function to quantify modifications
of the in-medium spectral function as the entire temperature dependence of spa-
tial correlation functions emerges from the temperature dependence of the spectral
function (Eq. 73), while in the case of temporal correlation functions (Eq. 70) this
temperature dependence is folded with the temperature dependence of the kernel,
cosh ( ( 1/2T )) / sinh (/2T ).
While the relation between spectral functions and spatial meson correlators is
more involved, in some limiting cases it becomes more intuitive. At large distances
the spatial correlation functions decay exponentially, G(z, T ) exp(M (T )z),
characterized by the screening mass M . At low temperatures this exponential drop
of a mesonic correlator projecting on a given quantum number channel is con-
trolled by the bound state with smallest mass in this channel. The spectral function
then is dominated by this state,
(, pz , T ) ( 2 p2z m20 ), (74)
where M becomes equal to the pole mass m0 of the meson. At high temperatures,
when the mesonic excitations have completely dissolved and become uncorrelated,
the spatial meson correlation functions describe the propagation of a free quark
anti-quark pair and the screening mass is given by244
q q
Mfree = m2q1 + (T )2 + m2q2 + (T )2 , (75)
where mq1 and mq2 are the masses of the quark and anti-quark that form the meson.
This form of the screening mass in the non-interacting limit is a consequence of the
54 H.-T. Ding, F. Karsch and S. Mukherjee

1.2 1.2
GNO(z,T)/GNO(z,0), sc-, 1 GNO(z,T)/GNO(z,0), cc-, 1
1.0 1.0

0.8 0.8

0.6 0.6
T [MeV] T [MeV]
0.4 149 0.4 149
171 171
0.2 197 0.2 197
220 220
248 z [fm] 248 z [fm]
0.0 0.0
0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5

4.2
M [GeV] sc- M [GeV] cc-
3.5 1+ 4 1++
+ ++
0 0
1 3.8 1
3 0

3.6 0
+

3.4
2.5
3.2
3
2
T [MeV] T [MeV]
2.8
100 150 200 250 300 350 400 450 500 100 150 200 250 300 350 400 450 500

Fig. 29. (Top) Temperature dependence of ratios of spatial correlation functions to those at zero
temperature for Ds (top left) and J/ (top right).245 (Bottom) Temperature dependence of
screening masses of ground state sc mesons (left) Ds1 (1+ ), D (0+ ), D (1 ) and D (0 )
s0 s s
and ground state charmonia (right) c1 (1++ ), c0 (0++ ), J/ (1 ) and c (0+ ).245 The
horizontal solid lines denote the corresponding zero-temperature meson masses and the dashed
line depicts the non-interacting theory limit of a freely propagating quark anti-quark pair. The
yellow band denotes the chiral crossover temperature region, i.e. Tc = 154 9 MeV.

fermionic (anti-periodic) boundary condition along the Euclidean-time direction,

which leads to the appearance of a smallest non-zero Matsubara frequency, T , in
the quark and anti-quark propagators. As the bosonic meson state is dissolved in
the non-interacting limit the screening mass reflects two independently propagating
fermionic degrees of freedom. Thus, the transition between these two limiting values
for the screening mass can be used as an indicator for the thermal modification and
eventual dissociation of charmonia and open charm mesons.
A first analysis of spatial correlation functions of charmonia was performed
using staggered fermions with the p4 action.246 Recently, a detailed comparative
study of the spatial correlation functions of charmonia and open charm mesons was
performed using the HISQ action.245 Fig. 29 shows ratios of the spatial correlation
functions for the Ds (top left) and the J/ (top right) at different temperatures
to the corresponding zero temperature results. As discussed before, such a ratio
can directly probe the thermal modifications of the spectral functions. If there is
no change in the meson spectral functions, these ratios will be equal to unity and
deviations from unity will indicate in-medium modification of the meson spectral
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 55

functions at non-zero temperature. As can be seen, even for T Tc 154 MeV

the ratio for the open charm sector (Fig. 29 (top left)) shows significant deviation
from unity. This is completely consistent with the previously discussed quantum
number correlation based studies in Section 7.1 which concluded that open charm
mesons start to melt in the vicinity of Tc .152 On the other hand, for the charmonia
the ratio of non-zero to zero temperature correlation functions remains very close
to unity for T . 170 MeV and significant deviations, comparable to the open
charm sector at T Tc , only show up for T & 200 MeV (Fig. 29 (top right)).
This indicates that up to T ' 1.1 Tc J/ exists almost like a vacuum state and
significant thermal modifications appears only for T & 1.3Tc . The same physics
is also reflected in temperature dependence of the screening masses of open charm
mesons and charmonia as shown in Fig. 29 (bottom). The screening masses of all
open charm mesons (bottom left) start to deviate from unity significantly already
in the chiral crossover region. The J/ and c screening masses (bottom right),
however, remain to be almost identical to their vacuum pole masses up to T '
1.1 Tc . The screening masses for c0 and c1 deviate from their vacuum masses
even below Tc indicating significant thermal modifications of these states. At higher
temperatures, i.e. T & 2 Tc the screening masses in all channels increase linearly
with temperature. This rise agrees well with that of a freely propagating quark
anti-quark pair.

8. QCD in external magnetic fields

In recent past it has been realized that large magnetic fields created during the early
stages of heavy-ion collisions may give rise to fascinating observable effects induced
through the coupling between the magnetic field and the chiral anomaly.20,21 This
observation also motivated a plethora of activities including lattice QCD studies of
hot-dense strong-interaction matter under the influence of external magnetic fields.
This section presents a very brief summary of these lattice QCD studies. More
comprehensive reviews on this topic can be found in Refs. 247250.
An external magnetic field along the z-direction, B ~ = B z, can be induced
by choosing electro-magnetic gauge fields: Ay = Bx and Ax = Az = A = 0.
On the lattice this can be implemented simply by multiplying the SU (3) gauge
2
field variables Un, with the corresponding U (1) phase factors: un,y = eia qBnx
and un,x = un,z = un, = 1. Here, q is the (electric) charge of a quark
and n = (nx , ny , nz , n ) denotes a lattice site, with nx,y,z = 1 . . . N and
n = 1 . . . N . This choice ensures that the magnetic flux, a2 B, is constant through
all the plaquettes in the x-y plane, except at the boundary (N , ny , nz , n ) owing
to the periodic boundary condition on the gauge fields along the spatial directions.
Thus, to preserve the smoothness of the external magnetic field across the bound-
ary and the gauge invariance of the fermion action the U (1) factor on the boundary
2
links must be modified: u(N ,ny ,nz ,n ),x = eia qN Bny . By making such a choice
one pays the price that the magnetic flux becomes quantized, a2 B = 2nB /(qN2 )
56 H.-T. Ding, F. Karsch and S. Mukherjee

B x100
4 anisotropy method
Bali et al., PRL 14

half-half method
Levkova & DeTar, PRL 14

2
finite diff. method
Bonati et al., PRD 14

integral method
Bali et al., JHEP 14
0
T [MeV]

100 150 200 250 300

Fig. 30. (Left) The QCD crossover temperature as a function of the magnetic field obtained from
the inflection point of the quark condensate (red band) and strange quark susceptibility (blue
band).253 (Right) The magnetic susceptibility B , i.e. the correction to the pressure proportional
to quadratic power of the magnetic field, as a function of temperature.258

with nB = 0 . . . N2 . Thus, at a given temperature T and on a finite volume there

is a minimal value for the constant, external magnetic field B = 2T 2 N2 /(qN2 ).
On the other hand, for given temporal extent, N , of the lattice the maximal value
of the magnetic field achieved in this case is B = 2T 2 N2 /q. Most importantly,
since the external magnetic field is induced by multiplying the SU (3) gauge field
variables with purely imaginary phase factors the fermion determinant remains real
and positive, allowing direct Monte-Carlo sampling.
The QCD crossover transition in an external magnetic field has been extensively
studied using lattice QCD. In contrast to the earlier effective model based predic-
tions and the results of first lattice QCD calculations251,252 using unimproved stag-
gered fermions with heavy pion masses, the QCD crossover temperature Tc turns
out to decease with increasing magnetic field.253255 Fig. 30 (left) illustrates the
decrease of Tc as function of the magnetic field.253 The reduction of Tc follows from
the reduction of the chiral condensate with increasing B, a phenomenon known
as the inverse magnetic catalysis, for the relevant intermediate temperature range.
However, for low as well as high enough temperatures the chiral condensate shows
the expected magnetic catalysis, i.e. an increase with increasing B.255257
The response of the hot strong-interaction matter to an external magnetic field
has also been investigated through lattice QCD calculations of the magnetic suscep-
tibility. The magnetic susceptibility is the second derivative of the QCD partition
function with respect to the external magnetic field B and, thus, gives the order B 2
corrections to the pressure. Since the standard approach leads to discrete, quantized
values for B on the lattice, unlike for the susceptibilities of the chemical potentials,
taking direct derivatives of the QCD partition function with respect to B is not
straightforward. Lattice calculations of the magnetic susceptibility have been per-
formed using three different approaches: (i) by computing the free energy difference
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 57

through an integration over the varying magnetic field;259,260,260 (ii) through com-
putations of the magnetization via pressure anisotropies parallel and orthogonal to
the direction of the magnetic field;261,262 (iii) by adopting a non-uniform external
field such that the net magnetic flux across the x-y plane of the whole lattice is zero,
resulting in a non-quantized magnetic field and allowing for a direct computation of
the derivatives of the partition function with respect to B.263 As shown in Fig. 30
(right), all these different approaches produce more or less consistent results. The
QGP phase turns out to be paramagnetic. However, for low enough temperatures
QCD is weakly diamagnetic, in accordance with the hadron resonance gas model.264
These results also suggests that for T . 300 MeV the order B 2 magnetic contribu-
tion to the QCD pressure is at best a few percent for a magnetic field of the order
of 1015 T, which is relevant for heavy-ion collision experiments.21
Lattice QCD studies have also been carried out to investigate the chiral magnetic
effect,21 namely the phenomenon of (electric) charge separation along the direction
of the magnetic field in presence of chiral anomaly induced topological fluctuations.
Various different avenues have been pursued by studying: (i) charge fluctuations on
given topological background;265 (ii) enhanced fluctuations of the electric current
along the magnetic field,266 (iii) dependence of the electric current on a given axial
chemical potential,266 (iv) correlations of the electric polarization and the topolog-
ical charge.267 In all cases, the chiral magnetic effect has been observed, however,
in a suppressed magnitude compared to the model based expectations. Obviously,
in lattice QCD studies of the chiral magnetic effect, which is closely related to the
chiral anomaly of QCD, it would be of great interest to have lattice QCD calcu-
lations perfomed with a chiral fermion formulation. Fully dynamical calculations
with DWF or overlap fermions at present do not yet exist.

9. Summary

In this review we have discussed selected topics in finite temperature QCD that
have a direct link to the ongoing experimental study of the phase diagram of strong-
interaction matter and the properties of matter formed in heavy ion collisions. We
naturally focused on topics to which lattice QCD calculations can contribute and
have lead or may lead to definite, quantitative answers in the future.
Lattice QCD calculations at finite temperature started more than 30 years ago
with simulations of pure SU (N ) gauge theories on computers that delivered less
than one Mega-Flop/s peak-performance. Nowadays these calculations are per-
formed with dynamical quarks and the correct mass spectrum of QCD on comput-
ers that reach a peak-performance of more than ten Peta-Flop/s. The field thus
could utilize computing resources with a peak-performance that increased by more
than ten orders of magnitude, or doubled the speed almost every year. Although we
know quite well from the physics of the hadron gas that an exponential rise cannot
continue for ever, we still do not seem to have reached the critical point and we may
58 H.-T. Ding, F. Karsch and S. Mukherjee

also in the future expect to gain further inside into the physics of strong-interaction
matter through numerical lattice QCD calculations.
During the last years lattice QCD calculations have accomplished two long-
standing goals: (i) the pseudo-critical temperature of strong-interaction matter
with physical light and strange quark masses has been determined at vanishing
net-baryon number, Tc = (154 9) MeV, and (ii) the equation of state has been
calculated in a wide temperature range. It now is frequently used in the hydrody-
namic modeling of heavy ion collisions. The extension of these results to non-zero
chemical potential is well controlled at least in leading order Taylor expansion of
the pressure in terms of baryon, electric charge and strangeness chemical poten-
tials and we will soon have results for bulk thermodynamics in a range of baryon
chemical potentials B /T <3, which will cover most of the parameter space that
can be explored in heavy ion experiments at RHIC and the LHC. These Taylor ex-
pansions also provide specific information on fluctuations of and correlations among
conserved charges. Lattice QCD calculations of these observables have reached a
stage where guidance can be provided to the experimental search for the critical end
point and, with some care, a comparison between experimental results for measured
proton number, strangeness and, in particular, electric charge fluctuations on the
one hand and lattice QCD calculations on the other hand starts to become possible.
A major focus in the ongoing experimental heavy ion program is to explore in
detail the properties of matter formed in heavy ion collisions. The analysis of trans-
port coefficients, thermal masses and screening lengths does play an important role
in this effort. Lattice QCD calculations of these quantities are possible, although in
many cases they are complicated as an analytic continuation from Euclidean time
to Minkowski time is needed. For this reason most of the existing calculations have
been performed in the quenched approximation where fermionic degrees of freedom
are taken into account only as valence quarks and the influence of virtual quark loops
on the gluonic background is neglected. Still these calculations prepare the ground
for future fully dynamical calculations of transport properties of strong-interaction
matter, for which in some cases exploratory work has already started.
There remain many obvious, open questions in the study of the phase diagram
of strong-interaction matter and the properties of the strongly interacting medium
created in heavy ion collisions which can be addressed in future lattice QCD cal-
culations and which will profit from future increasing computing resources as well
as algorithmic advances. Calculations at non-zero baryon number density and the
search for the critical point, the calculation of transport properties with fully dy-
namical, light quarks as well as calculations of thermodynamic quantities using chi-
ral fermion discretization schemes are among the most demanding problems which
hopefully can be addressed in the coming years with improved techniques and re-
sources.
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 59

Acknowledgements

This review would not have been possible without the input from many of our
colleagues, in particular, Prasad Hegde, Olaf Kaczmarek, Christian Schmidt and
Mathias Wagner, with whom we could collaborate over many years in the HotQCD
and Bielefeld-BNL-CCNU collaborations. This work has been supported in part
through contract DE-SC0012704 with the U.S. Department of Energy, the BMBF
under grant 05P12PBCTA and the GSI BILAER grant.

References

1. R. Hagedorn, Statistical thermodynamics of strong interactions at high-energies,

Nuovo Cim.Suppl. 3 (1965) 147186.
2. G. Baym, Confinement of quarks in nuclear matter, Physica A 96 (1979) 131.
3. T. Celik, F. Karsch, and H. Satz, A percolation approach to strongly interacting
matter, Phys.Lett. B97 (1980) 128130.
4. K. Redlich and H. Satz, The Legacy of Rolf Hagedorn: Statistical Bootstrap and
Ultimate Temperature, arXiv:1501.0752.
5. H. D. Politzer, Reliable Perturbative Results for Strong Interactions?,
Phys.Rev.Lett. 30 (1973) 13461349.
6. D. J. Gross and F. Wilczek, Ultraviolet Behavior of Nonabelian Gauge Theories,
Phys.Rev.Lett. 30 (1973) 13431346.
7. N. Cabibbo and G. Parisi, Exponential Hadronic Spectrum and Quark Liberation,
Phys.Lett. B59 (1975) 6769.
8. J. C. Collins and M. Perry, Superdense Matter: Neutrons Or Asymptotically Free
Quarks?, Phys.Rev.Lett. 34 (1975) 1353.
9. E. V. Shuryak, Quantum Chromodynamics and the Theory of Superdense Matter,
Phys.Rept. 61 (1980) 71158.
10. R. D. Pisarski and F. Wilczek, Remarks on the Chiral Phase Transition in
Chromodynamics, Phys.Rev. D29 (1984) 338341.
11. K. Rajagopal and F. Wilczek, The Condensed matter physics of QCD,
hep-ph/0011333.
12. K. Fukushima and T. Hatsuda, The phase diagram of dense QCD, Rept.Prog.Phys.
74 (2011) 014001, [arXiv:1005.4814].
13. K. G. Wilson, Confinement of Quarks, Phys.Rev. D10 (1974) 24452459.
14. M. Creutz, Monte Carlo Study of Quantized SU(2) Gauge Theory, Phys.Rev. D21
(1980) 23082315.
15. L. D. McLerran and B. Svetitsky, A Monte Carlo Study of SU(2) Yang-Mills
Theory at Finite Temperature, Phys.Lett. B98 (1981) 195.
16. J. Kuti, J. Polonyi, and K. Szlachanyi, Monte Carlo Study of SU(2) Gauge Theory
at Finite Temperature, Phys.Lett. B98 (1981) 199.
17. J. Engels, F. Karsch, H. Satz, and I. Montvay, High Temperature SU(2) Gluon
Matter on the Lattice, Phys.Lett. B101 (1981) 89.
18. T. Matsui and H. Satz, J/ Suppression by Quark-Gluon Plasma Formation,
Phys.Lett. B178 (1986) 416.
19. R. Rapp and J. Wambach, Chiral symmetry restoration and dileptons in relativistic
heavy ion collisions, Adv.Nucl.Phys. 25 (2000) 1, [hep-ph/9909229].
20. D. E. Kharzeev, K. Landsteiner, A. Schmitt, and H.-U. Yee, Strongly interacting
60 H.-T. Ding, F. Karsch and S. Mukherjee

matter in magnetic fields: an overview, Lect.Notes Phys. 871 (2013) 111,

[arXiv:1211.6245].
21. D. E. Kharzeev, L. D. McLerran, and H. J. Warringa, The Effects of topological
charge change in heavy ion collisions: Event by event P and CP violation,
Nucl.Phys. A803 (2008) 227253, [arXiv:0711.0950].
22. A. D. Linde, Infrared Problem in Thermodynamics of the Yang-Mills Gas,
Phys.Lett. B96 (1980) 289.
23. E. Braaten and R. D. Pisarski, Soft Amplitudes in Hot Gauge Theories: A General
Analysis, Nucl.Phys. B337 (1990) 569.
24. N. Haque, A. Bandyopadhyay, J. O. Andersen, M. G. Mustafa, M. Strickland,
et al., Three-loop HTLpt thermodynamics at finite temperature and chemical
potential, JHEP 1405 (2014) 027, [arXiv:1402.6907].
25. E. Braaten and A. Nieto, Effective field theory approach to high temperature
thermodynamics, Phys.Rev. D51 (1995) 69907006, [hep-ph/9501375].
26. A. Hietanen, K. Kajantie, M. Laine, K. Rummukainen, and Y. Schroder,
Three-dimensional physics and the pressure of hot QCD, Phys.Rev. D79 (2009)
045018, [arXiv:0811.4664].
27. I. Montvay and G. Muenster, Quantum Fields on a Lattice . Cambridge
Monographs on Mathematical Physics, 2010.
28. H. J. Rothe, Lattice Gauge Theories: An Introduction (Fourth Edition). World
Scientific Lecture Notes in Physics, 2012.
29. T. Degrand and C. Detar, Lattice Methods for Quantum Chromodynamics. World
Scientific, 2006.
30. C. Gattringer and C. Lang, Quantum Chromodynamics on the Lattice-An
Introductory Presentation. Springer, 2010.
31. Particle Data Group Collaboration, K. Olive et al., Review of Particle Physics,
Chin.Phys. C38 (2014) 090001.
32. P. Hasenfratz and F. Karsch, Chemical Potential on the Lattice, Phys.Lett. B125
(1983) 308.
33. R. Gavai, Chemical Potential on the Lattice Revisited, Phys.Rev. D32 (1985) 519.
34. G. Aarts, F. A. James, E. Seiler, and I.-O. Stamatescu, Complex Langevin: Etiology
and Diagnostics of its Main Problem, Eur.Phys.J. C71 (2011) 1756,
[arXiv:1101.3270].
35. G. Aarts, L. Bongiovanni, E. Seiler, D. Sexty, and I.-O. Stamatescu, Controlling
complex Langevin dynamics at finite density, Eur.Phys.J. A49 (2013) 89,
[arXiv:1303.6425].
36. D. Sexty, New algorithms for finite density QCD, PoS Lattice2015 (2014) 016,
[arXiv:1410.8813].
37. AuroraScience Collaboration, M. Cristoforetti, F. Di Renzo, and L. Scorzato,
New approach to the sign problem in quantum field theories: High density QCD on
a Lefschetz thimble, Phys.Rev. D86 (2012) 074506, [arXiv:1205.3996].
38. H. Fujii, D. Honda, M. Kato, Y. Kikukawa, S. Komatsu, et al., Hybrid Monte Carlo
on Lefschetz thimbles - A study of the residual sign problem, JHEP 1310 (2013)
147, [arXiv:1309.4371].
39. M. Cristoforetti, F. Di Renzo, G. Eruzzi, A. Mukherjee, C. Schmidt, et al., An
efficient method to compute the residual phase on a Lefschetz thimble, Phys.Rev.
D89 (2014) 114505, [arXiv:1403.5637].
40. R. V. Gavai and S. Gupta, Pressure and nonlinear susceptibilities in QCD at finite
chemical potentials, Phys.Rev. D68 (2003) 034506, [hep-lat/0303013].
41. C. Allton, S. Ejiri, S. Hands, O. Kaczmarek, F. Karsch, et al., The Equation of
THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 61

state for two flavor QCD at nonzero chemical potential, Phys.Rev. D68 (2003)
014507, [hep-lat/0305007].
42. K. Symanzik, Continuum Limit and Improved Action in Lattice Theories. 1.
Principles and phi**4 Theory, Nucl.Phys. B226 (1983) 187.
43. K. Symanzik, Continuum Limit and Improved Action in Lattice Theories. 2. O(N)
Nonlinear Sigma Model in Perturbation Theory, Nucl.Phys. B226 (1983) 205.
44. H. B. Nielsen and M. Ninomiya, No Go Theorem for Regularizing Chiral Fermions,
Phys.Lett. B105 (1981) 219.
45. J. B. Kogut and L. Susskind, Hamiltonian Formulation of Wilsons Lattice Gauge
Theories, Phys.Rev. D11 (1975) 395408.
46. D. B. Kaplan, A Method for simulating chiral fermions on the lattice, Phys.Lett.
B288 (1992) 342347, [hep-lat/9206013].
47. Y. Shamir, Chiral fermions from lattice boundaries, Nucl.Phys. B406 (1993)
90106, [hep-lat/9303005].
48. Y. Shamir, The Euclidean spectrum of Kaplans lattice chiral fermions, Phys.Lett.
B305 (1993) 357365, [hep-lat/9212010].
49. V. Furman and Y. Shamir, Axial symmetries in lattice QCD with Kaplan fermions,
Nucl.Phys. B439 (1995) 5478, [hep-lat/9405004].
50. H. Neuberger, More about exactly massless quarks on the lattice, Phys.Lett. B427
(1998) 353355, [hep-lat/9801031].
51. H. Neuberger, Exactly massless quarks on the lattice, Phys.Lett. B417 (1998)
141144, [hep-lat/9707022].
52. B. Sheikholeslami and R. Wohlert, Improved Continuum Limit Lattice Action for
QCD with Wilson Fermions, Nucl.Phys. B259 (1985) 572.
53. E. Follana, Q. Mason, C. Davies, K. Hornbostel, G. P. Lepage, J. Shigemitsu,
H. Trottier, and K. Wong, Highly improved staggered quarks on the lattice with
applications to charm physics, Phys. Rev. D 75 (2007) 054502.
54. C. Morningstar and M. J. Peardon, Analytic smearing of SU(3) link variables in
lattice QCD, Phys.Rev. D69 (2004) 054501, [hep-lat/0311018].
55. U. M. Heller, F. Karsch, and B. Sturm, Improved staggered fermion actions for
QCD thermodynamics, Phys.Rev. D60 (1999) 114502, [hep-lat/9901010].
56. RBC, UKQCD Collaboration, D. J. Antonio et al., Localization and chiral
symmetry in three flavor domain wall QCD, Phys.Rev. D77 (2008) 014509,
[arXiv:0705.2340].
57. D. Renfrew, T. Blum, N. Christ, R. Mawhinney, and P. Vranas, Controlling
Residual Chiral Symmetry Breaking in Domain Wall Fermion Simulations, PoS
LATTICE2008 (2008) 048, [arXiv:0902.2587].
58. P. H. Ginsparg and K. G. Wilson, A Remnant of Chiral Symmetry on the Lattice,
Phys.Rev. D25 (1982) 2649.
59. M. I. Buchoff, M. Cheng, N. H. Christ, H. T. Ding, C. Jung, et al., QCD chiral
transition, U(1)A symmetry and the dirac spectrum using domain wall fermions,
Phys.Rev. D89 (2014) 054514, [arXiv:1309.4149].
60. G. Cossu, S. Aoki, H. Fukaya, S. Hashimoto, T. Kaneko, et al., Finite temperature
study of the axial U(1) symmetry on the lattice with overlap fermion formulation,
Phys.Rev. D87 (2013) 114514, [arXiv:1304.6145].
61. T. Bhattacharya, M. I. Buchoff, N. H. Christ, H.-T. Ding, R. Gupta, et al., QCD
Phase Transition with Chiral Quarks and Physical Quark Masses, Phys.Rev.Lett.
113 (2014) 082001, [arXiv:1402.5175].
62. V. Dick, F. Karsch, E. Laermann, S. Mukherjee, and S. Sharma, Microscopic
Origin of UA (1) Symmetry Violation in the High Temperature Phase of QCD,
62 H.-T. Ding, F. Karsch and S. Mukherjee

arXiv:1502.0619.
63. WHOT-QCD Collaboration, H. Saito et al., Phase structure of finite temperature
QCD in the heavy quark region, Phys.Rev. D84 (2011) 054502, [arXiv:1106.0974].
64. H. Saito, S. Ejiri, S. Aoki, K. Kanaya, Y. Nakagawa, et al., Histograms in
heavy-quark QCD at finite temperature and density, Phys.Rev. D89 (2014) 034507,
[arXiv:1309.2445].
65. F. Karsch, E. Laermann, and C. Schmidt, The Chiral critical point in three-flavor
QCD, Phys.Lett. B520 (2001) 4149, [hep-lat/0107020].
66. P. de Forcrand and O. Philipsen, The Chiral critical line of N(f ) = 2+1 QCD at
zero and non-zero baryon density, JHEP 0701 (2007) 077, [hep-lat/0607017].
67. G. Endrodi, Z. Fodor, S. Katz, and K. Szabo, The Nature of the finite temperature
QCD transition as a function of the quark masses, PoS LAT2007 (2007) 182,
[arXiv:0710.0998].
68. H.-T. Ding, A. Bazavov, P. Hegde, F. Karsch, S. Mukherjee, et al., Exploring phase
diagram of Nf = 3 QCD at = 0 with HISQ fermions, PoS LATTICE2011
(2011) 191, [arXiv:1111.0185].
69. A. Butti, A. Pelissetto, and E. Vicari, On the nature of the finite temperature
transition in QCD, JHEP 0308 (2003) 029, [hep-ph/0307036].
70. A. Pelissetto and E. Vicari, Relevance of the axial anomaly at the finite-temperature
chiral transition in QCD, Phys.Rev. D88 (2013) 105018, [arXiv:1309.5446].
71. M. Grahl and D. H. Rischke, Functional renormalization group study of the
two-flavor linear sigma model in the presence of the axial anomaly, Phys.Rev. D88
(2013) 056014, [arXiv:1307.2184].
72. Y. Aoki, G. Endrodi, Z. Fodor, S. Katz, and K. Szabo, The Order of the quantum
chromodynamics transition predicted by the standard model of particle physics,
Nature 443 (2006) 675678, [hep-lat/0611014].
73. E. K. Riedel and F. J. Wegner, Tricritical exponents and scaling fields, Phys. Rev.
Lett. 29 (1972) 349352.
74. J. Engels, S. Holtmann, T. Mendes, and T. Schulze, Finite size scaling functions for
3-d O(4) and O(2) spin models and QCD, Phys.Lett. B514 (2001) 299308,
[hep-lat/0105028].
75. J. Engels, L. Fromme, and M. Seniuch, Correlation lengths and scaling functions in
the three-dimensional O(4) model, Nucl.Phys. B675 (2003) 533554,
[hep-lat/0307032].
76. S. Ejiri, F. Karsch, E. Laermann, C. Miao, S. Mukherjee, et al., On the magnetic
equation of state in (2+1)-flavor QCD, Phys.Rev. D80 (2009) 094505,
[arXiv:0909.5122].
77. C. Bonati, P. de Forcrand, M. DElia, O. Philipsen, and F. Sanfilippo, Chiral phase
transition in two-flavor QCD from an imaginary chemical potential, Phys.Rev. D90
(2014) 074030, [arXiv:1408.5086].
78. A. Bazavov, T. Bhattacharya, M. Cheng, C. DeTar, H. Ding, et al., The chiral and
deconfinement aspects of the QCD transition, Phys.Rev. D85 (2012) 054503,
[arXiv:1111.1710].
79. Y. Aoki, Z. Fodor, S. Katz, and K. Szabo, The QCD transition temperature:
Results with physical masses in the continuum limit, Phys.Lett. B643 (2006) 4654,
[hep-lat/0609068].
80. Y. Aoki, S. Borsanyi, S. Durr, Z. Fodor, S. D. Katz, et al., The QCD transition
temperature: results with physical masses in the continuum limit II., JHEP 0906
(2009) 088, [arXiv:0903.4155].
81. Wuppertal-Budapest Collaboration, S. Borsanyi et al., Is there still any Tc
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 63

mystery in lattice QCD? Results with physical masses in the continuum limit III,
JHEP 1009 (2010) 073, [arXiv:1005.3508].
82. O. Kaczmarek, F. Karsch, E. Laermann, C. Miao, S. Mukherjee, et al., Phase
boundary for the chiral transition in (2+1) -flavor QCD at small values of the
chemical potential, Phys.Rev. D83 (2011) 014504, [arXiv:1011.3130].
83. G. Endrodi, Z. Fodor, S. Katz, and K. Szabo, The QCD phase diagram at nonzero
quark density, JHEP 1104 (2011) 001, [arXiv:1102.1356].
84. P. Cea, L. Cosmai, and A. Papa, Critical line of 2+1 flavor QCD, Phys.Rev. D89
(2014) 074512, [arXiv:1403.0821].
85. C. Bonati, M. DElia, M. Mariti, M. Mesiti, F. Negro, et al., Curvature of the chiral
pseudocritical line in QCD, Phys.Rev. D90 (2014) 114025, [arXiv:1410.5758].
86. A. Andronic, P. Braun-Munzinger, K. Redlich, and J. Stachel, The thermal model
on the verge of the ultimate test: particle production in Pb-Pb collisions at the
LHC, J.Phys. G38 (2011) 124081, [arXiv:1106.6321].
87. Z. Fodor and S. Katz, Critical point of QCD at finite T and mu, lattice results for
physical quark masses, JHEP 0404 (2004) 050, [hep-lat/0402006].
88. S. Datta, R. V. Gavai, and S. Gupta, QCD at finite chemical potential with Nt = 8,
PoS LATTICE2013 (2014) 202.
89. S. Ejiri, Lee-Yang zero analysis for the study of QCD phase structure, Phys.Rev.
D73 (2006) 054502, [hep-lat/0506023].
90. P. de Forcrand and O. Philipsen, The QCD phase diagram for small densities from
imaginary chemical potential, Nucl.Phys. B642 (2002) 290306, [hep-lat/0205016].
91. X.-Y. Jin, Y. Kuramashi, Y. Nakamura, S. Takeda, and A. Ukawa, Curvature of the
critical line on the plane of quark chemical potential and pseudo scalar meson mass
for three-flavor QCD, arXiv:1504.0011.
92. C. Allton, M. Doring, S. Ejiri, S. Hands, O. Kaczmarek, et al., Thermodynamics of
two flavor QCD to sixth order in quark chemical potential, Phys.Rev. D71 (2005)
054508, [hep-lat/0501030].
93. R. Gavai and S. Gupta, QCD at finite chemical potential with six time slices,
Phys.Rev. D78 (2008) 114503, [arXiv:0806.2233].
94. O. Kaczmarek, F. Karsch, P. Petreczky, and F. Zantow, Heavy quark anti-quark
free energy and the renormalized Polyakov loop, Phys.Lett. B543 (2002) 4147,
[hep-lat/0207002].
95. L. D. McLerran and B. Svetitsky, Quark Liberation at High Temperature: A Monte
Carlo Study of SU(2) Gauge Theory, Phys.Rev. D24 (1981) 450.
96. A. Bazavov and P. Petreczky, Polyakov loop in 2+1 flavor QCD, Phys.Rev. D87
(2013) 094505, [arXiv:1301.3943].
97. R. Bellwied, S. Borsanyi, Z. Fodor, S. D. Katz, and C. Ratti, Is there a flavor
hierarchy in the deconfinement transition of QCD?, Phys.Rev.Lett. 111 (2013)
202302, [arXiv:1305.6297].
98. A. Bazavov, H. T. Ding, P. Hegde, O. Kaczmarek, F. Karsch, et al., Strangeness at
high temperatures: from hadrons to quarks, Phys.Rev.Lett. 111 (2013) 082301,
[arXiv:1304.7220].
99. V. Koch, A. Majumder, and J. Randrup, Baryon-strangeness correlations: A
Diagnostic of strongly interacting matter, Phys.Rev.Lett. 95 (2005) 182301,
[nucl-th/0505052].
100. S. Ejiri, F. Karsch, and K. Redlich, Hadronic fluctuations at the QCD phase
transition, Phys.Lett. B633 (2006) 275282, [hep-ph/0509051].
101. S. L. Adler, Axial vector vertex in spinor electrodynamics, Phys.Rev. 177 (1969)
24262438.
64 H.-T. Ding, F. Karsch and S. Mukherjee

102. J. Bell and R. Jackiw, A PCAC puzzle: pi0 -gt; gamma gamma in the sigma model,
Nuovo Cim. A60 (1969) 4761.
103. G. t Hooft, Symmetry Breaking Through Bell-Jackiw Anomalies, Phys.Rev.Lett. 37
(1976) 811.
104. D. J. Gross, R. D. Pisarski, and L. G. Yaffe, QCD and Instantons at Finite
Temperature, Rev.Mod.Phys. 53 (1981) 43.
105. E. V. Shuryak, Which chiral symmetry is restored in hot QCD?, Comments
Nucl.Part.Phys. 21 (1994) 235248, [hep-ph/9310253].
106. M. Cheng, S. Datta, A. Francis, J. van der Heide, C. Jung, et al., Meson screening
masses from lattice QCD with two light and the strange quark, Eur.Phys.J. C71
(2011) 1564, [arXiv:1010.1216].
107. S. R. Sharpe, Rooted staggered fermions: Good, bad or ugly?, PoS LAT2006 (2006)
022, [hep-lat/0610094].
108. G. C. Donald, C. T. Davies, E. Follana, and A. S. Kronfeld, Staggered fermions,
zero modes, and flavor-singlet mesons, Phys.Rev. D84 (2011) 054504,
[arXiv:1106.2412].
109. HotQCD Collaboration, A. Bazavov et al., The chiral transition and U (1)A
symmetry restoration from lattice QCD using Domain Wall Fermions, Phys.Rev.
D86 (2012) 094503, [arXiv:1205.3535].
110. A. Tomiya, G. Cossu, H. Fukaya, S. Hashimoto, and J. Noaki, Effects of near-zero
Dirac eigenmodes on axial U(1) symmetry at finite temperature, arXiv:1412.7306.
111. TWQCD Collaboration, T.-W. Chiu, W.-P. Chen, Y.-C. Chen, H.-Y. Chou, and
T.-H. Hsieh, Chiral symmetry and axial U(1) symmetry in finite temperature QCD
with domain-wall fermion, PoS LATTICE2013 (2014) 165, [arXiv:1311.6220].
112. F. Karsch, Lattice QCD at high temperature and density, Lect.Notes Phys. 583
(2002) 209249, [hep-lat/0106019].
113. S. Borsanyi, Z. Fodor, C. Hoelbling, S. D. Katz, S. Krieg, et al., Full result for the
QCD equation of state with 2+1 flavors, Phys.Lett. B730 (2014) 99104,
[arXiv:1309.5258].
114. HotQCD Collaboration, A. Bazavov et al., Equation of state in ( 2+1 )-flavor
QCD, Phys.Rev. D90 (2014) 094503, [arXiv:1407.6387].
115. A. Majumder and B. Muller, Hadron Mass Spectrum from Lattice QCD,
Phys.Rev.Lett. 105 (2010) 252002, [arXiv:1008.1747].
116. A. Bazavov, H. T. Ding, P. Hegde, O. Kaczmarek, F. Karsch, et al., Additional
Strange Hadrons from QCD Thermodynamics and Strangeness Freezeout in Heavy
Ion Collisions, Phys.Rev.Lett. 113 (2014) 072001, [arXiv:1404.6511].
117. R. V. Gavai and S. Gupta, Quark number susceptibilities, strangeness and
dynamical confinement, Phys.Rev. D64 (2001) 074506, [hep-lat/0103013].
118. C. Allton, S. Ejiri, S. Hands, O. Kaczmarek, F. Karsch, et al., The QCD thermal
phase transition in the presence of a small chemical potential, Phys.Rev. D66
(2002) 074507, [hep-lat/0204010].
119. S. Ejiri, F. Karsch, E. Laermann, and C. Schmidt, The Isentropic equation of state
of 2-flavor QCD, Phys.Rev. D73 (2006) 054506, [hep-lat/0512040].
120. S. Borsanyi, Z. Fodor, S. D. Katz, S. Krieg, C. Ratti, et al., Fluctuations of
conserved charges at finite temperature from lattice QCD, JHEP 1201 (2012) 138,
[arXiv:1112.4416].
121. HotQCD Collaboration, A. Bazavov et al., Fluctuations and Correlations of net
baryon number, electric charge, and strangeness: A comparison of lattice QCD
results with the hadron resonance gas model, Phys.Rev. D86 (2012) 034509,
[arXiv:1203.0784].
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 65

122. S. Borsanyi, G. Endrodi, Z. Fodor, S. Katz, S. Krieg, et al., QCD equation of state
at nonzero chemical potential: continuum results with physical quark masses at
order mu2 , JHEP 1208 (2012) 053, [arXiv:1204.6710].
123. S. Borsanyi, Z. Fodor, S. Katz, S. Krieg, C. Ratti, et al., Freeze-out parameters:
lattice meets experiment, Phys.Rev.Lett. 111 (2013) 062005, [arXiv:1305.5161].
124. for the BNL-Bielefeld-CCNU Collaboration, P. Hegde, The QCD equation of
state to O(4B ), arXiv:1412.6727.
125. M. Laine and Y. Schroder, Quark mass thresholds in QCD thermodynamics,
Phys.Rev. D73 (2006) 085009, [hep-ph/0603048].
126. M. Stephanov, Non-Gaussian fluctuations near the QCD critical point,
Phys.Rev.Lett. 102 (2009) 032301, [arXiv:0809.3450].
127. M. Asakawa, S. Ejiri, and M. Kitazawa, Third moments of conserved charges as
probes of QCD phase structure, Phys.Rev.Lett. 103 (2009) 262301,
[arXiv:0904.2089].
128. B. Friman, F. Karsch, K. Redlich, and V. Skokov, Fluctuations as probe of the
QCD phase transition and freeze-out in heavy ion collisions at LHC and RHIC,
Eur.Phys.J. C71 (2011) 1694, [arXiv:1103.3511].
129. M. Stephanov, On the sign of kurtosis near the QCD critical point, Phys.Rev.Lett.
107 (2011) 052301, [arXiv:1104.1627].
130. M. A. Stephanov, K. Rajagopal, and E. V. Shuryak, Signatures of the tricritical
point in QCD, Phys.Rev.Lett. 81 (1998) 48164819, [hep-ph/9806219].
131. M. A. Stephanov, K. Rajagopal, and E. V. Shuryak, Event-by-event fluctuations in
heavy ion collisions and the QCD critical point, Phys.Rev. D60 (1999) 114028,
[hep-ph/9903292].
132. S. Jeon and V. Koch, Event by event fluctuations, hep-ph/0304012.
133. Y. Hatta and M. Stephanov, Proton number fluctuation as a signal of the QCD
critical endpoint, Phys.Rev.Lett. 91 (2003) 102003, [hep-ph/0302002].
134. C. Athanasiou, K. Rajagopal, and M. Stephanov, Using Higher Moments of
Fluctuations and their Ratios in the Search for the QCD Critical Point, Phys.Rev.
D82 (2010) 074008, [arXiv:1006.4636].
135. Studying the phase diagram of qcd matter at rhic, 2014.
https://drupal.star.bnl.gov/STAR/files/BES_WPII_ver6.9_Cover.pdf.
136. STAR Collaboration, M. Aggarwal et al., Higher Moments of Net-proton
Multiplicity Distributions at RHIC, Phys.Rev.Lett. 105 (2010) 022302,
[arXiv:1004.4959].
137. STAR Collaboration, L. Adamczyk et al., Energy Dependence of Moments of
Net-proton Multiplicity Distributions at RHIC, Phys.Rev.Lett. 112 (2014) 032302,
[arXiv:1309.5681].
138. STAR Collaboration, L. Adamczyk et al., Beam energy dependence of moments of
the net-charge multiplicity distributions in Au+Au collisions at RHIC,
Phys.Rev.Lett. 113 (2014) 092301, [arXiv:1402.1558].
139. STAR Collaboration, N. R. Sahoo, Recent results on event-by-event fluctuations
from the RHIC Beam Energy Scan program in the STAR experiment,
J.Phys.Conf.Ser. 535 (2014) 012007, [arXiv:1407.1554].
140. STAR Collaboration, X. Luo, Energy Dependence of Moments of Net-Proton and
Net-Charge Multiplicity Distributions at STAR, PoS CPOD2014 (2014) 019,
[arXiv:1503.0255].
141. PHENIX Collaboration, J. T. Mitchell, The RHIC Beam Energy Scan Program:
Results from the PHENIX Experiment, Nucl.Phys. A904-905 (2013) 903c906c,
[arXiv:1211.6139].
66 H.-T. Ding, F. Karsch and S. Mukherjee

142. A. Bazavov, H. Ding, P. Hegde, O. Kaczmarek, F. Karsch, et al., Freeze-out

Conditions in Heavy Ion Collisions from QCD Thermodynamics, Phys.Rev.Lett.
109 (2012) 192302, [arXiv:1208.1220].
143. M. Kitazawa and M. Asakawa, Relation between baryon number fluctuations and
experimentally observed proton number fluctuations in relativistic heavy ion
collisions, Phys.Rev. C86 (2012) 024904, [arXiv:1205.3292].
144. A. Bzdak and V. Koch, Acceptance corrections to net baryon and net charge
cumulants, Phys.Rev. C86 (2012) 044904, [arXiv:1206.4286].
145. A. Bzdak and V. Koch, Local Efficiency Corrections to Higher Order Cumulants,
Phys.Rev. C91 (2015) 027901, [arXiv:1312.4574].
146. F. Karsch, Determination of Freeze-out Conditions from Lattice QCD Calculations,
Central Eur.J.Phys. 10 (2012) 12341237, [arXiv:1202.4173].
147. S. Mukherjee and M. Wagner, Deconfinement of strangeness and freeze-out from
charge fluctuations, PoS CPOD2013 (2013) 039, [arXiv:1307.6255].
148. S. Borsanyi, Z. Fodor, S. Katz, S. Krieg, C. Ratti, et al., Freeze-out parameters
from electric charge and baryon number fluctuations: is there consistency?,
Phys.Rev.Lett. 113 (2014) 052301, [arXiv:1403.4576].
149. S. Mukherjee, freeze-out condition from lattice qcd and the role of additional
strange hadrons, PoS CPOD2014 (2014) 005.
150. S. Capstick and N. Isgur, Baryons in a Relativized Quark Model with
Chromodynamics, Phys.Rev. D34 (1986) 2809.
151. D. Ebert, R. Faustov, and V. Galkin, Mass spectra and Regge trajectories of light
mesons in the relativistic quark model, Phys.Rev. D79 (2009) 114029,
[arXiv:0903.5183].
152. A. Bazavov, H.-T. Ding, P. Hegde, O. Kaczmarek, F. Karsch, et al., The melting
and abundance of open charm hadrons, Phys.Lett. B737 (2014) 210215,
[arXiv:1404.4043].
153. M. Asakawa, T. Hatsuda, and Y. Nakahara, Maximum entropy analysis of the
spectral functions in lattice QCD, Prog.Part.Nucl.Phys. 46 (2001) 459508,
[hep-lat/0011040].
154. R. Rapp and H. van Hees, Heavy Quark Diffusion as a Probe of the Quark-Gluon
Plasma, arXiv:0803.0901.
155. H.-T. Ding, Hard and thermal probes of QGP from the perspective of Lattice QCD,
arXiv:1404.5134.
156. H. B. Meyer, Transport Properties of the Quark-Gluon Plasma: A Lattice QCD
Perspective, Eur.Phys.J. A47 (2011) 86, [arXiv:1104.3708].
157. F. Karsch, E. Laermann, P. Petreczky, S. Stickan, and I. Wetzorke, A Lattice
calculation of thermal dilepton rates, Phys.Lett. B530 (2002) 147152,
[hep-lat/0110208].
158. Y. Nakahara, M. Asakawa, and T. Hatsuda, Hadronic spectral functions in lattice
QCD, Phys.Rev. D60 (1999) 091503, [hep-lat/9905034].
159. G. Aarts, C. Allton, J. Foley, S. Hands, and S. Kim, Spectral functions at small
energies and the electrical conductivity in hot, quenched lattice QCD, Phys.Rev.Lett.
99 (2007) 022002, [hep-lat/0703008].
160. H.-T. Ding, A. Francis, O. Kaczmarek, F. Karsch, E. Laermann, et al., Thermal
dilepton rate and electrical conductivity: An analysis of vector current correlation
functions in quenched lattice QCD, Phys.Rev. D83 (2011) 034504,
[arXiv:1012.4963].
161. O. Kaczmarek, E. Laermann, M. Muller, F. Karsch, H. Ding, et al., Thermal
dilepton rates from quenched lattice QCD, PoS ConfinementX (2012) 185,
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 67

[arXiv:1301.7436].
162. O. Kaczmarek and M. Muller, Temperature dependence of electrical conductivity
and dilepton rates from hot quenched lattice QCD, PoS LATTICE2013 (2014)
175, [arXiv:1312.5609].
163. H.-T. Ding, O. Kaczmarek, and F. Meyer, Vector spectral functions and transport
properties in quenched QCD, arXiv:1412.5869.
164. S. Gupta, The Electrical conductivity and soft photon emissivity of the QCD
plasma, Phys.Lett. B597 (2004) 5762, [hep-lat/0301006].
165. B. B. Brandt, A. Francis, H. B. Meyer, and H. Wittig, Thermal Correlators in the
channel of two-flavor QCD, JHEP 1303 (2013) 100, [arXiv:1212.4200].
166. G. Aarts, C. Allton, A. Amato, P. Giudice, S. Hands, et al., Electrical conductivity
and charge diffusion in thermal QCD from the lattice, arXiv:1412.6411.
167. A. Amato, G. Aarts, C. Allton, P. Giudice, S. Hands, et al., Electrical conductivity
of the quark-gluon plasma across the deconfinement transition, Phys.Rev.Lett. 111
(2013) 172001, [arXiv:1307.6763].
168. D. Fernandez-Fraile and A. Gomez Nicola, The Electrical conductivity of a pion
gas, Phys.Rev. D73 (2006) 045025, [hep-ph/0512283].
169. PHENIX Collaboration, A. Adare et al., Energy Loss and Flow of Heavy Quarks
in Au+Au Collisions at s(NN)**(1/2) = 200-GeV, Phys.Rev.Lett. 98 (2007)
172301, [nucl-ex/0611018].
170. STAR Collaboration, B. Abelev et al., Erratum: Transverse momentum and
centrality dependence of high-pT non-photonic electron suppression in Au+Au

collisions at sN N = 200 GeV, Phys.Rev.Lett. 98 (2007) 192301,
[nucl-ex/0607012].
171. ALICE Collaboration, B. Abelev et al., Suppression of high transverse momentum

D mesons in central Pb-Pb collisions at sN N = 2.76 TeV, JHEP 1209 (2012) 112,
[arXiv:1203.2160].
172. M. He, R. J. Fries, and R. Rapp, Modifications of Heavy-Flavor Spectra in

sNN = 62.4 GeV Au-Au Collisions, arXiv:1409.4539.
173. S. Cao, Y. Huang, G.-Y. Qin, and S. A. Bass, The Influence of Initial State
Fluctuations on Heavy Quark Energy Loss in Relativistic Heavy-ion Collisions,
arXiv:1404.3139.
174. H. Ding, A. Francis, O. Kaczmarek, F. Karsch, H. Satz, et al., Heavy Quark
diffusion from lattice QCD spectral functions, J.Phys. G38 (2011) 124070,
[arXiv:1107.0311].
175. S. Caron-Huot and G. D. Moore, Heavy quark diffusion in perturbative QCD at
next-to-leading order, Phys.Rev.Lett. 100 (2008) 052301, [arXiv:0708.4232].
176. P. Kovtun, D. T. Son, and A. O. Starinets, Holography and hydrodynamics:
Diffusion on stretched horizons, JHEP 0310 (2003) 064, [hep-th/0309213].
177. O. Kaczmarek, Continuum estimate of the heavy quark momentum diffusion
coefficient , arXiv:1409.3724.
178. D. Banerjee, S. Datta, R. Gavai, and P. Majumdar, Heavy Quark Momentum
Diffusion Coefficient from Lattice QCD, Phys.Rev. D85 (2012) 014510,
[arXiv:1109.5738].
179. S. Caron-Huot, M. Laine, and G. D. Moore, A Way to estimate the heavy quark
thermalization rate from the lattice, JHEP 0904 (2009) 053, [arXiv:0901.1195].
180. J. Casalderrey-Solana and D. Teaney, Heavy quark diffusion in strongly coupled
N=4 Yang-Mills, Phys.Rev. D74 (2006) 085012, [hep-ph/0605199].
181. H. B. Meyer, A Calculation of the shear viscosity in SU(3) gluodynamics, Phys.Rev.
D76 (2007) 101701, [arXiv:0704.1801].
68 H.-T. Ding, F. Karsch and S. Mukherjee

182. M. Luscher and P. Weisz, Locality and exponential error reduction in numerical
lattice gauge theory, JHEP 0109 (2001) 010, [hep-lat/0108014].
183. G. Parisi, R. Petronzio, and F. Rapuano, A Measurement of the String Tension
Near the Continuum Limit, Phys.Lett. B128 (1983) 418.
184. P. De Forcrand and C. Roiesnel, REFINED METHODS FOR MEASURING
LARGE DISTANCE CORRELATIONS, Phys.Lett. B151 (1985) 7780.
185. G. D. Moore and D. Teaney, How much do heavy quarks thermalize in a heavy ion
collision?, Phys.Rev. C71 (2005) 064904, [hep-ph/0412346].
186. F. Karsch and H. Wyld, Thermal Greens Functions and Transport Coefficients on
the Lattice, Phys.Rev. D35 (1987) 2518.
187. H. B. Meyer, The Yang-Mills spectrum from a two level algorithm, JHEP 0401
(2004) 030, [hep-lat/0312034].
188. A. Nakamura and S. Sakai, Transport coefficients of gluon plasma, Phys.Rev.Lett.
94 (2005) 072305, [hep-lat/0406009].
189. H. B. Meyer, A Calculation of the bulk viscosity in SU(3) gluodynamics,
Phys.Rev.Lett. 100 (2008) 162001, [arXiv:0710.3717].
190. F. Karsch, D. Kharzeev, and K. Tuchin, Universal properties of bulk viscosity near
the QCD phase transition, Phys.Lett. B663 (2008) 217221, [arXiv:0711.0914].
191. D. Kharzeev and K. Tuchin, Bulk viscosity of QCD matter near the critical
temperature, JHEP 0809 (2008) 093, [arXiv:0705.4280].
192. P. Romatschke and D. T. Son, Spectral sum rules for the quark-gluon plasma,
Phys.Rev. D80 (2009) 065021, [arXiv:0903.3946].
193. P. Romatschke, Relativistic Viscous Fluid Dynamics and Non-Equilibrium Entropy,
Class.Quant.Grav. 27 (2010) 025006, [arXiv:0906.4787].
194. G. D. Moore and K. A. Sohrabi, Kubo Formulae for Second-Order Hydrodynamic
Coefficients, Phys.Rev.Lett. 106 (2011) 122302, [arXiv:1007.5333].
195. G. D. Moore and K. A. Sohrabi, Thermodynamical second-order hydrodynamic
coefficients, JHEP 1211 (2012) 148, [arXiv:1210.3340].
196. G. Denicol, H. Niemi, E. Molnar, and D. Rischke, Derivation of transient
relativistic fluid dynamics from the Boltzmann equation, Phys.Rev. D85 (2012)
114047, [arXiv:1202.4551].
197. R. Baier, P. Romatschke, D. T. Son, A. O. Starinets, and M. A. Stephanov,
Relativistic viscous hydrodynamics, conformal invariance, and holography, JHEP
0804 (2008) 100, [arXiv:0712.2451].
198. Y. Kohno, M. Asakawa, and M. Kitazawa, Shear viscosity to relaxation time ratio
in SU(3) lattice gauge theory, Phys.Rev. D89 (2014) 054508, [arXiv:1112.1508].
199. O. Philipsen and C. Schafer, The second order hydrodynamic transport coefficient
for the gluon plasma from the lattice, JHEP 1402 (2014) 003, [arXiv:1311.6618].
200. P. Romatschke, New Developments in Relativistic Viscous Hydrodynamics,
Int.J.Mod.Phys. E19 (2010) 153, [arXiv:0902.3663].
201. A. Adil and I. Vitev, Collisional dissociation of heavy mesons in dense QCD
matter, Phys.Lett. B649 (2007) 139146, [hep-ph/0611109].
202. R. Sharma, I. Vitev, and B.-W. Zhang, Light-cone wave function approach to open
heavy flavor dynamics in QCD matter, Phys.Rev. C80 (2009) 054902,
[arXiv:0904.0032].
203. M. He, R. J. Fries, and R. Rapp, Heavy-Quark Diffusion and Hadronization in
Quark-Gluon Plasma, Phys.Rev. C86 (2012) 014903, [arXiv:1106.6006].
204. D. Ebert, R. Faustov, and V. Galkin, Heavy-light meson spectroscopy and Regge
trajectories in the relativistic quark model, Eur.Phys.J. C66 (2010) 197206,
[arXiv:0910.5612].
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 69

205. D. Ebert, R. Faustov, and V. Galkin, Spectroscopy and Regge trajectories of heavy
baryons in the relativistic quark-diquark picture, Phys.Rev. D84 (2011) 014025,
[arXiv:1105.0583].
206. A. Mocsy, P. Petreczky, and M. Strickland, Quarkonia in the Quark Gluon Plasma,
Int.J.Mod.Phys. A28 (2013) 1340012, [arXiv:1302.2180].
207. A. Bazavov, P. Petreczky, and A. Velytsky, Quarkonium at Finite Temperature,
arXiv:0904.1748.
208. N. Brambilla, S. Eidelman, B. Heltsley, R. Vogt, G. Bodwin, et al., Heavy
quarkonium: progress, puzzles, and opportunities, Eur.Phys.J. C71 (2011) 1534,
[arXiv:1010.5827].
209. E. Eichten, K. Gottfried, T. Kinoshita, K. Lane, and T.-M. Yan, Charmonium: The
Model, Phys.Rev. D17 (1978) 3090.
210. E. Eichten, K. Gottfried, T. Kinoshita, K. Lane, and T.-M. Yan, Charmonium:
Comparison with Experiment, Phys.Rev. D21 (1980) 203.
211. A. Mocsy, Potential Models for Quarkonia, Eur.Phys.J. C61 (2009) 705710,
[arXiv:0811.0337].
212. M. Laine, O. Philipsen, P. Romatschke, and M. Tassler, Real-time static potential
in hot QCD, JHEP 0703 (2007) 054, [hep-ph/0611300].
213. A. Beraudo, J.-P. Blaizot, and C. Ratti, Real and imaginary-time Q anti-Q
correlators in a thermal medium, Nucl.Phys. A806 (2008) 312338,
[arXiv:0712.4394].
214. N. Brambilla, J. Ghiglieri, A. Vairo, and P. Petreczky, Static quark-antiquark pairs
at finite temperature, Phys.Rev. D78 (2008) 014017, [arXiv:0804.0993].
215. W. Caswell and G. Lepage, Effective Lagrangians for Bound State Problems in
QED, QCD, and Other Field Theories, Phys.Lett. B167 (1986) 437.
216. G. T. Bodwin, E. Braaten, and G. P. Lepage, Rigorous QCD analysis of inclusive
annihilation and production of heavy quarkonium, Phys.Rev. D51 (1995)
11251171, [hep-ph/9407339].
217. N. Brambilla, A. Pineda, J. Soto, and A. Vairo, Potential NRQCD: An Effective
theory for heavy quarkonium, Nucl.Phys. B566 (2000) 275, [hep-ph/9907240].
218. A. Rothkopf, T. Hatsuda, and S. Sasaki, Complex Heavy-Quark Potential at Finite
Temperature from Lattice QCD, Phys.Rev.Lett. 108 (2012) 162001,
[arXiv:1108.1579].
219. Y. Burnier and A. Rothkopf, Disentangling the timescales behind the
non-perturbative heavy quark potential, Phys.Rev. D86 (2012) 051503,
[arXiv:1208.1899].
220. Y. Burnier and A. Rothkopf, A hard thermal loop benchmark for the extraction of
the nonperturbative QQ potential, Phys.Rev. D87 (2013) 114019,
[arXiv:1304.4154].
221. Y. Burnier, O. Kaczmarek, and A. Rothkopf, Static quark-antiquark potential in the
quark-gluon plasma from lattice QCD, arXiv:1410.2546.
222. S. Nadkarni, Nonabelian Debye Screening. 2. The Singlet Potential, Phys.Rev. D34
(1986) 3904.
223. A. Bazavov, Y. Burnier, and P. Petreczky, Lattice calculation of the heavy quark
potential at non-zero temperature, arXiv:1404.4267.
224. A. Rothkopf, Improved Maximum Entropy Analysis with an Extended Search Space,
J.Comput.Phys. 238 (2013) 106114, [arXiv:1110.6285].
225. A. Rothkopf, Improved Maximum Entropy Method with an Extended Search Space,
PoS LATTICE2012 (2012) 100, [arXiv:1208.5162].
226. Y. Burnier and A. Rothkopf, Bayesian Approach to Spectral Function
70 H.-T. Ding, F. Karsch and S. Mukherjee

Reconstruction for Euclidean Quantum Field Theories, Phys.Rev.Lett. 111 (2013)

182003, [arXiv:1307.6106].
227. H. Ding, A. Francis, O. Kaczmarek, F. Karsch, H. Satz, et al., Charmonium
properties in hot quenched lattice QCD, Phys.Rev. D86 (2012) 014509,
[arXiv:1204.4945].
228. M. Asakawa and T. Hatsuda, J / psi and eta(c) in the deconfined plasma from
lattice QCD, Phys.Rev.Lett. 92 (2004) 012001, [hep-lat/0308034].
229. S. Datta, F. Karsch, P. Petreczky, and I. Wetzorke, Behavior of charmonium
systems after deconfinement, Phys.Rev. D69 (2004) 094507, [hep-lat/0312037].
230. A. Jakovac, P. Petreczky, K. Petrov, and A. Velytsky, Quarkonium correlators and
spectral functions at zero and finite temperature, Phys.Rev. D75 (2007) 014506,
[hep-lat/0611017].
231. H. Iida, T. Doi, N. Ishii, H. Suganuma, and K. Tsumura, Charmonium properties
in deconfinement phase in anisotropic lattice QCD, Phys.Rev. D74 (2006) 074502,
[hep-lat/0602008].
232. WHOT-QCD Collaboration, H. Ohno et al., Charmonium spectral functions with
the variational method in zero and finite temperature lattice QCD, Phys.Rev. D84
(2011) 094504, [arXiv:1104.3384].
233. G. Aarts, C. Allton, M. B. Oktay, M. Peardon, and J.-I. Skullerud, Charmonium at
high temperature in two-flavor QCD, Phys.Rev. D76 (2007) 094513,
[arXiv:0705.2198].
234. S. Borsanyi, S. Durr, Z. Fodor, C. Hoelbling, S. D. Katz, et al., Charmonium
spectral functions from 2+1 flavour lattice QCD, JHEP 1404 (2014) 132,
[arXiv:1401.5940].
235. G. Aarts, C. Allton, T. Harris, S. Kim, M. P. Lombardo, et al., The bottomonium
spectrum at finite temperature from Nf = 2 + 1 lattice QCD, JHEP 1407 (2014)
097, [arXiv:1402.6210].
236. H. Ohno, H. T. Ding, and O. Kaczmarek, Quark mass dependence of quarkonium
properties at finite temperature, arXiv:1412.6594.
237. B. Thacker and G. P. Lepage, Heavy quark bound states in lattice QCD, Phys.Rev.
D43 (1991) 196208.
238. G. P. Lepage, L. Magnea, C. Nakhleh, U. Magnea, and K. Hornbostel, Improved
nonrelativistic QCD for heavy quark physics, Phys.Rev. D46 (1992) 40524067,
[hep-lat/9205007].
239. G. Aarts, S. Kim, M. Lombardo, M. Oktay, S. Ryan, et al., Bottomonium above
deconfinement in lattice nonrelativistic QCD, Phys.Rev.Lett. 106 (2011) 061602,
[arXiv:1010.3725].
240. G. Aarts, C. Allton, S. Kim, M. Lombardo, S. Ryan, et al., Melting of P wave
bottomonium states in the quark-gluon plasma from lattice NRQCD, JHEP 1312
(2013) 064, [arXiv:1310.5467].
241. G. Aarts, C. Allton, S. Kim, M. P. Lombardo, M. B. Oktay, et al., S wave
bottomonium states moving in a quark-gluon plasma from lattice NRQCD, JHEP
1303 (2013) 084, [arXiv:1210.2903].
242. G. Aarts, C. Allton, S. Kim, M. Lombardo, M. Oktay, et al., What happens to the
and b in the quark-gluon plasma? Bottomonium spectral functions from lattice
QCD, JHEP 1111 (2011) 103, [arXiv:1109.4496].
243. S. Kim, P. Petreczky, and A. Rothkopf, Lattice NRQCD study of S- and P-wave
bottomonium states in a thermal medium with Nf = 2 + 1 light flavors,
arXiv:1409.3630.
244. W. Florkowski and B. L. Friman, Spatial dependence of the finite temperature
 THERMODYNAMICS OF STRONG-INTERACTION MATTER FROM LATTICE QCD 71

meson correlation function, Z.Phys. A347 (1994) 271276.

245. A. Bazavov, F. Karsch, Y. Maezawa, S. Mukherjee, and P. Petreczky, In-medium
modifications of open and hidden strange-charm mesons from spatial correlation
functions, arXiv:1411.3018.
246. F. Karsch, E. Laermann, S. Mukherjee, and P. Petreczky, Signatures of
charmonium modification in spatial correlation functions, Phys.Rev. D85 (2012)
114501, [arXiv:1203.3770].
247. M. DElia, Lattice QCD Simulations in External Background Fields, Lect.Notes
Phys. (2013) [arXiv:1209.0374].
248. A. Yamamoto, Chiral Magnetic Effect on the Lattice, Lect.Notes Phys. 871 (2013)
387397, [arXiv:1207.0375].
249. G. Endrodi, QCD in magnetic fields: from Hofstadters butterfly to the phase
diagram, arXiv:1410.8028.
250. M. DElia, Lattice QCD with purely imaginary sources at zero and non-zero
temperature, arXiv:1502.0604.
251. M. DElia, S. Mukherjee, and F. Sanfilippo, QCD Phase Transition in a Strong
Magnetic Background, Phys.Rev. D82 (2010) 051501, [arXiv:1005.5365].
252. E.-M. Ilgenfritz, M. Kalinowski, M. Muller-Preussker, B. Petersson, and
A. Schreiber, Two-color QCD with staggered fermions at finite temperature under
the influence of a magnetic field, Phys.Rev. D85 (2012) 114504, [arXiv:1203.3360].
253. G. Bali, F. Bruckmann, G. Endrodi, Z. Fodor, S. Katz, et al., The QCD phase
diagram for external magnetic fields, JHEP 1202 (2012) 044, [arXiv:1111.4956].
254. V. Bornyakov, P. Buividovich, N. Cundy, O. Kochetkov, and A. Schafer,
Deconfinement transition in two-flavor lattice QCD with dynamical overlap fermions
in an external magnetic field, Phys.Rev. D90 (2014) 034501, [arXiv:1312.5628].
255. E. M. Ilgenfritz, M. Muller-Preussker, B. Petersson, and A. Schreiber, Magnetic
catalysis (and inverse catalysis) at finite temperature in two-color lattice QCD,
Phys.Rev. D89 (2014) 054512, [arXiv:1310.7876].
256. M. DElia and F. Negro, Chiral Properties of Strong Interactions in a Magnetic
Background, Phys.Rev. D83 (2011) 114028, [arXiv:1103.2080].
257. G. Bali, F. Bruckmann, G. Endrodi, Z. Fodor, S. Katz, et al., QCD quark
condensate in external magnetic fields, Phys.Rev. D86 (2012) 071502,
[arXiv:1206.4205].
258. G. Bali, F. Bruckmann, G. Endrodi, S. Katz, and A. Schafer, The QCD equation of
state in background magnetic fields, JHEP 1408 (2014) 177, [arXiv:1406.0269].
259. C. Bonati, M. DElia, M. Mariti, F. Negro, and F. Sanfilippo, Magnetic
Susceptibility of Strongly Interacting Matter across the Deconfinement Transition,
Phys.Rev.Lett. 111 (2013) 182001, [arXiv:1307.8063].
260. C. Bonati, M. DElia, M. Mariti, F. Negro, and F. Sanfilippo, Magnetic
susceptibility and equation of state of Nf = 2 + 1 QCD with physical quark masses,
Phys.Rev. D89 (2014) 054506, [arXiv:1310.8656].
261. G. Bali, F. Bruckmann, G. Endrodi, F. Gruber, and A. Schaefer, Magnetic
field-induced gluonic (inverse) catalysis and pressure (an)isotropy in QCD, JHEP
1304 (2013) 130, [arXiv:1303.1328].
262. G. Bali, F. Bruckmann, G. Endrodi, and A. Schafer, Paramagnetic squeezing of
QCD matter, Phys.Rev.Lett. 112 (2014) 042301, [arXiv:1311.2559].
263. L. Levkova and C. DeTar, Quark-gluon plasma in an external magnetic field,
Phys.Rev.Lett. 112 (2014) 012002, [arXiv:1309.1142].
264. G. Endrodi, QCD equation of state at nonzero magnetic fields in the Hadron
Resonance Gas model, JHEP 1304 (2013) 023, [arXiv:1301.1307].
72 H.-T. Ding, F. Karsch and S. Mukherjee

265. M. Abramczyk, T. Blum, G. Petropoulos, and R. Zhou, Chiral magnetic effect in

2+1 flavor QCD+QED, PoS LAT2009 (2009) 181, [arXiv:0911.1348].
266. P. Buividovich, M. Chernodub, E. Luschevskaya, and M. Polikarpov, Numerical
evidence of chiral magnetic effect in lattice gauge theory, Phys.Rev. D80 (2009)
054503, [arXiv:0907.0494].
267. G. Bali, F. Bruckmann, G. Endrodi, Z. Fodor, S. Katz, et al., Local CP-violation
and electric charge separation by magnetic fields from lattice QCD, JHEP 1404
(2014) 129, [arXiv:1401.4141].