Temperature

arXiv:2507.08010v1 [cond-mat.stat-mech] 4 Jul 2025

Miroslav Grmela ∗
École Polytechnique Montréal, C.P.6079 suc. Centre-ville,
Montréal, H3C 3A7, Québec, Canada

Abstract
Conversations about weather, environment, health, cuisine, and even
politics, all involve the word ”temperature”. It was an attempt to un-
derstand the working of heat engines that gave the temperature a clear
definition. In this Note we put the equilibrium-thermodynamics definition
into a larger context of the multiscale thermodynamics, the multiscale rate
thermodynamics, and nonphysical environments.

1 Equilibrium temperatures
The fundamental stone on which the equilibrium thermodynamics stays is the
observation of the existence of the approach of externally unforced macroscopic
systems to equilibrium states at which no time evolution takes place (called
0-th law of thermodynamics). All systems considered in this section are either
at equilibrium or are approaching equilibrium. More general situations are
considered in Sections 2 and 3.

1.1 Macroscopic equilibrium temperature

Macroscopic systems are characterized in the classical equilibrium thermody-
namics by an internal energy per unit volume ϵ and the number of moles per
unit volume ν. In this Note we limit ourselves only to the energy. The number
of moles remains unchanged and we omit it in our notation.
Macroscopic systems are surrounded by two types of walls. One allows free
passing of ϵ (Wall ∥) and the other prevents it (Wall ♯). Let us consider a
particular situation in which a macroscopic system that is surrounded by Wall
♯ is divided into two subsystems (S1 , S2 ) by Wall ∥. The system is not at
equilibrium at the time t = 0. The internal energies of the two subsystems
are ϵ1 and ϵ2 . According to 0-th law of thermodynamics the system starts to
evolve in time and reaches eventually an equilibrium state. The equilibrium
thermodynamics addresses only the final outcome of the time evolution but not
the time evolution itself (see 1.3 where such time evolution is addressed). The
∗ e-mail: [email protected]

1
state reached as t → ∞ is the state at which entropy s(ϵ) reaches its maximum
(MaxEnt principle). The individual nature of the macroscopic system under
investigation is expressed in equilibrium thermodynamics in the relation s = s(ϵ)
(called a fundamental thermodynamic relation). All relations s = s(ϵ) are
required to be real valued, sufficiently regular, and concave function. Moreover,
the entropy s as well as the internal energy ϵ are assume to be extensive state
variables (i.e. ϵ = ϵ1 + ϵ2 and s(ϵ1 , ϵ2 ) = s(ϵ1 ) + s(ϵ2 )).
With these requirements the MaxEnt principle implies
ds ds
= (1)
dϵ1 dϵ2
This means that at the equilibrium state the temperatures T1 and T2 of the two
subsystems S1 and S2 , defined by

1 ds(ϵ)
= = ϵ∗ > 0 (2)
T dϵ
are equal. The equilibration of temperatures in the approach to equilibrium
is also the basis of their measurement. Let S2 be a system in which states are
characterized by a collection ξ of state variables. The state variables in ξ include
the internal energy ϵ but they also include other state variables. For example
if S2 is a human body then ξ includes many biological state variables like for
instance the variables characterizing states of nerves. Their coupling to the
internal energy then brings about the feeling of the temperature. This biological
thermometer has always existed, has always been known, and moreover, it has
been absolutely essential for survival of humans and all living creatures. An
objective measurement of the temperature is made by thermometers in which
ξ includes state variables, like for instance volume, that can be objectively
quantified.
We mention two other important consequences of the above requirements on
the fundamental thermodynamic relation s = s(ϵ) and on the internal energy
ϵ. First, we note that the extensibility of both ϵ and s implies that s is a 1-
homogeneous function of ϵ, i.e. s = λ1 s(λϵ) for λ ∈ R. Consequently, s = s(ϵ)
can be extended (by Euler relation) to S = S(E, V ); S = E ∗ E + V ∗ V , where
V is the volume, S and E is the entropy and the internal energy of the whole
system (i.e. S = V s; E = V ϵ; E ∗ = ϵ∗ = T1 ; V ∗ = − PT is the conjugate of V ; P
has the physical interpretation of the pressure. The second consequence of the
requirements on s and ϵ, namely that ds(ϵ) ∗
dϵ = ϵ > 0 is that there is a one-to-one
relation between s and ϵ. We can consider s to be a state variable and ϵ = ϵ(s)
as the fundamental thermodynamic relation.
Summing up, the concept of temperature in the classical equilib-
rium thermodynamics requires: (i) a substance having an internal
energy, (ii) walls allowing to pass or to stop passing it, (iii) the
process of equilibration driven by the gradient of a potential called
entropy (MaxEnt). The multiplicative inverse of the temperature is
the conjugate ϵ∗ (with respect to the entropy) of the internal energy;

2
ϵ∗ = T1 . The MaxEnt principle makes the group of Legendre transformations a
fundamental group of thermodynamics and consequently the contact geometry
in the three-dimensional space with coordinates (s, ϵ, ϵ∗ ) is the natural geom-
etry of thermodynamics. The Legendre transformations preserve the contact
1-form ds − ϵ∗ dϵ. The fundamental thermodynamic relation s = s(ϵ) makes (in
the contact geometry formulation of the classical equilibrium thermodynam-
ics) appearance as the Legendre submanifold (i.e. the two-dimensional man-
ifold on which ds − ϵ∗ dϵ = 0). The Legendre transformation of s = s(ϵ) is
s∗ = s∗ (ϵ∗ ) = ϕ(ϵ̂(ϵ∗ ), ϵ∗ ) where

ϕ(ϵ, ϵ∗ ) = −s(ϵ) + ϵ∗ ϵ (3)

∂ϕ
and ϵ̂(ϵ∗ ) is a solution of ∂ϵ = 0.

1.2 Microscopic equilibrium temperature

In this section we see macroscopic systems as composed of n ∼ 1023 particles
obeying (we assume) the classical mechanics. State variables of the particles are
(r, v) ∈ R6n where r are position coordinates and v momenta of the particles.
In the Liouville lift [1] the state variables are x = f (r, v), f : R6n → R+ . We
shall denote the Liouville state space with the symbol M ; f (r, v) ∈ M .
As in the classical thermodynamics we do not specify the time evolution
describing the approach to equilibrium (0-th law of thermodynamics) but ad-
dress only its final outcome by the MaxEnt principle. The entropy in this
completelyR microscopic
R viewpoint of macroscopic systems is the Gibbs entropy
s(x) = − dr dvf (r, v) ln f (r, v) (we put the Boltzmann constant kB = 1) that
is universal for all macroscopic systems. RThe Rindividual nature of macroscopic
systems is expressed in the energy e(x) = dr dvf (r, v)h(r, v), where h(r, v) is
the microscopic Hamiltonian of the particles. There is no internal energy ϵ since
all the details of the dynamics are expressed in terms of x. There is no substance
of unknown origin that contributes to the total energy e. The MaxEnt passage
from (s(x), e(x)) to the fundamental thermodynamic relation (s∗ = s∗ (ϵ∗ )) is
made by the Legendre transformation: s∗ = s∗ (ϵ∗ ) = ϕ(x̂(ϵ∗ ), ϵ∗ ), where

ϕ(x, ϵ∗ ) = −s(x) + ϵ∗ e(x) (4)

and x̂(ϵ∗ ) is a solution to ∂ϕ ˆ ∗

∂x = 0: f (r, v) = exp −ϵ h(r, v).
We note that the temperature enters the Gibbs microscopic thermodynamics
in the MaxEnt passage to the equilibrium (as the Lagrange multiplier in the
thermodynamic potential (4)) and then, when the passage is completed, in the
equilibrium submanifold

Meq = {f ∈ M |f (r, v) = fˆ(r, v) = exp −ϵ∗ h(r, v)} (5)

representing in the Gibbs theory the classical equilibrium thermodynamics.

The equilibrium manifold (5) is in fact a family of manifolds parametrized by

3
the temperature. We could now systematically look at all possible geometri-
cal characterization of the manifolds (5) and investigate their possible physi-
cal (equilibrium-thermodynamic) interpretations. Results of such investigation
would give physical interpretations of the temperature and offer also possible
ways to measure it. For instance, the most well known is the relation between
v2 ˆ
R
the temperature and the average kinetic energy ( dv 2m f (r, v)) in the case when
2
v
h(r, v) = 2m (i.e. in the case when the macroscopic system under investigation
is an ideal gas, m is the mass of one particle).
Summing up, the temperature on the completely microscopic level of de-
scription does not exists. It emerges in the MaxEnt passage to the equilibrium
state. It is a parameter in the equilibrium manifolds (5) expressing the classical
equilibrium thermodynamics in the Gibbs microscopic thermodynamics.

1.3 Mesoscopic equilibrium temperatures

In this section we take a mesoscopic view of macroscopic systems. The state
variable x does not take into account all the microscopic details as the n-particle
distribution function f (r, v) does. For instance x may be 1-particle distribution
function. With such state variable there are obviously many microscopic details
that cannot be expressed. The mesoscopic thermodynamics with 1-particle dis-
tribution function as state variable, called kinetic theory, has been pioneered
by Ludwig Boltzmann [2]. The approach to equilibrium is made in his theory
again by MaxEnt but the MaxEnt passage is not postulated but is made by
following solutions of the Boltzmann equation. This means that the 0-th postu-
late of the equilibrium thermodynamics and the MaxEnt principle (postulated
in both the classical equilibrium thermodynamics and the Gibbs microscopic
equilibrium thermodynamics) appear in solutions of the Boltzmann equation.
The only problem with the Boltzmann theory is its limited applicability. It
addresses only one macroscopic system, namely the ideal gas.
How can the Boltzmann theory be extended to general macroscopic systems?
One way to achieve it is to extract from the theory an abstract mathematical
structure that is common also to other well established (i.e. tested with ex-
perimental observations) mesoscopic dynamical theories (in particular hydro-
dynamics). The original Boltzmann theory could be then seen as a particular
realization of such abstract Boltzmann theory and other mesoscopic dynamical
theories addressing general macroscopic systems as its other particular realiza-
tions. This path has been taken in [3], [4], [5], [6]. A similar path, but started
with the plasma dynamics rather than the Boltzmann dynamics, has been taken
in [7], [8]. Both investigations have been motivated by Arnold’s formulation
[9] of the Euler hydrodynamic equation as a noncanonical Hamilton equation.
Arnold’s investigation introduced geometry into mesoscopic dynamics.
The time evolution governed by the abstract Boltzmann equation

ẋ = [L(x)x∗ − Ξx∗ (x, x∗ )]x∗ =ϕx (x,ϵ∗ ) (6)

describes indeed the MaxEnt approach to equilibrium states x̂(ϵ∗ ) (that are

4
solutions to ϕx (x, ϵ∗ ) - (see [10]). We use in (6) the following notation: ẋ = ∂x∂t ;
∂ϕ
ϕx is a shorthand notation for ∂x ; ϕ(x) is the thermodynamic potential (4); L(x)
is a Poisson bivector expressing kinematics of the mesoscopic state variable x;
Ξ(x, x∗ ) is a dissipation potential. We refer the reader to [10] where all properties
that L and Ξ are required to satisfy and many their particular realizations and
physical interpretations can be found. The Boltzmann kinetic equation as well
as for instance the Navier-Stokes-Fourier hydrodynamic equations are indeed
particular realizations of (6).
Another feature of Eq.(6) shows its belonging to thermodynamics. Not only
the passage from an initial state to the asymptotically (i.e. t → ∞) reached state
(i.e. to the equilibrium state x̂(ϵ∗ )) is made by a Legendre transformation (by
MaxEnt)) but the complete time evolution leading to x̂(ϵ∗ ) can be displayed as a
sequence of infinitesimal Legendre transformations (transformations preserving
the contact structure). In the lift of (6) (see [11], [12], [13]), [14]) from the space
with coordinates x to the space with coordinates (x, x∗ , ϕ) that is equipped
with the contact 1-form dϕ − x∗ dx the time evolution governed by (6) takes
place on a Legendre submanifold (the submanifold on which dϕ − x∗ dx = 0 )
and preserves dϕ − x∗ dx. As we noted already in Section 1.1, the fundamental
group of thermodynamics on all levels is the group of Legendre transformations.
The lift (see [11], [12], [13]), [14]) carries dynamics from mechanics (in which
the fundamental group is the group of canonical transformation preserving the
symplectic structure) to thermodynamics (in which the fundamental group is
the group of Legendre transformations preserving the contact structure).
Still another way to see (6) is as a reversal of the extension of equilibrium
thermodynamics to thermodynamic with fluctuations (see e.g. [15]). The Max-
Ent reduction made by following solutions of (6) is a top-down passage from
levels that take into account more details to the level of equilibrium thermo-
dynamics. The extension of equilibrium thermodynamics to thermodynamic
with fluctuations is a bottom-up extension from the level of equilibrium ther-
modynamics to a level involving more details (involving fluctuations). Both the
top-down and the bottom-up passages are driven by the entropy.
The state variable x in this section is a mesoscopic state variable. This
means that there are details in dynamics that cannot be expressed with it. The
contribution of the ignored details to the total energy is an internal energy ϵ.
The way we take it into account in mesoscopic dynamical theories gives the
physical interpretation to the temperature. In the rest of this section we shall
discuss two ways to include the internal energy ϵ: In Section 1.3.1 by modifying
the thermodynamic potential ϕ(x), and in Section 1.3.2 by adopting ϵ into the
set of state variables.

1.3.1 Modifications of the energy and the entropy

The classical equilibrium thermodynamics (Section 1.1) and the Gibbs micro-
scopic thermodynamics (Section 1.2) are two extreme answers to the question
of how to take into account the microscopic nature of the macroscopic systems.
In the former in a single state variable (in the internal energy x = ϵ), in the

5
latter by taking into account all the mechanical details in the n-particle distri-
bution function x = f (r, v). Taking the mesoscopic point of view of macroscopic
systems we have to attempt to combine the two extremes.
First we do not modify the chosen mesoscopic state variable x but turn
to modifications of the energy e(x). Let x be 1-particle distribution function
f (r, v). If the macroscopic system is an ideal gas (as in the Boltzmann investiga-
v2
R R
tion) then e(x) = dr dv 2m is the energy that can be completely expressed in
terms of x. Any particle-particle interactions need at least a 2-particle distribu-
tion function to express it. In the case of long range interactions we can approx-
imate the two particle distribution function f (r1 , v1 , r2 , v2 ) by f (r1 , v1 )f (r2 , v2 ).
The energy e(x) with such approximation is then the Vlasov mean field energy.
There are two other ingenious ways to express complex interactions with
the 1-particle distribution functions. Both extend the 1-particle distribution
function f (r, v) to f (r, v, ζ) and both address macroscopic systems composed of
complex molecules.
Let the molecules be long chain-like macromolecules. Due to their snake
like shapes they move, in the presence of other macromolecules, like snakes. De
Gennes [16] expressed this type motion by choosing ζ = (R, χ), where χ ∈ R is
a coordinate along the backbone of the chain and R ∈ R3 is the tangent vector
at the backbone coordinate χ. The snake like motion, called reptation by de
Gennes, is mathematically expressed as diffusion along the backbone.
A similar type of extension has been introduced independently by Ruggeri
and Sugyiama [17] for fluid composed of complex macromolecules with an in-
ternal energy inside them. In their extension f (r, v) → f (r, v, ζ), the new coor-
dinate ζ ∈ R is the internal energy of a single macromolecule.
Still another way to introduce the influence of ignored mechanical details is
to modify the entropy s(x). We have seen in Section 1.2 that on the completely
microscopic level where all mechanical details are taken into account and the
entropy s(x) is universal (the Gibbs entropy) for all systems. Its role is to sweep
away unimportant details and reveal the level of the equilibrium thermodynam-
ics. The same role plays the entropy (the Boltzmann entropy) in the Boltzmann
kinetic theory (except that the sweeping is made by following solutions of the
Botzmann equation) since the kinetic energy of the particles composing an ideal
gas composed of point particles is the complete energy.
Let us replace the ideal gas with the van der Waals gas. The particles com-
posing it interact via long range attractive forces and short range hard-core
repulsive forces. Following van Kampen [18] we still choose the 1-particle dis-
stribution function f (r, v) as the mesoscopic state variable. The long range
attractive force is expressed in van Kampen’s theory as the Vlasov mean field
force, but the hard-core repulsive force is expressed in the modification of the
entropy s(x). From the point of view of the classical equilibrium thermodynam-
ics, the hard R core
R interactions are thus expressed asR a heat. R The Boltzmann
entropy − dr dvf (r, v) ln f (r, v) is replaced by − dr dvf (r, v) ln f (r, v)−
where b is the volume of the particle. In this modification of the Boltzmann
theory the entropy ceases thus its universality. It is taking the role of contribut-

6
ing to the specification of the individual nature of the van der Waals fluid. We
recall that on the level of the equilibrium thermodynamics the individual nature
of macroscopic systems is expressed only in the entropy, the internal energy is
universal (it serves as a state variable).
All the ways to express the influence of details (that cannot be expressed
in terms of the chosen mesoscopic state variable x) that we discussed above
modify the equilibrium manifold (5). Consequently, they modify the physical
interpretation of the temperature ϵ∗ = T1 parametrizing them.

1.3.2 Internal energy as an extra state variable

The second family of attempts to take into account effect of details that cannot
be expressed in terms of the chosen mesoscopic state variable x follow the ex-
ample of fluid mechanics. We recall the formulation of fluid mechanics. Macro-
scopic systems (fluids) are seen as a continuum ([19]). Its states are character-
ized by fields (real valued functions of the position coordinate r) of mass density
ρ(r) and the momentum u(r). The time evolution equation is the Euler equa-
tion (that is the Newton equation for the continuum) supplemented by the local
mass conservation equation. How shall we express the internal energy? Follow-
ing the example of the equilibrium thermodynamics we simply adopt it as an
additional state variable except now it is a field rather than a real number as in
2
the equilibrium thermodynamics. The energy e(x) = dr u2ρ(r) (r)
R R
+ drϵ(r) is the
complete energy. This setting supplemented by the local energy conservation
equation and the Navier-Stokes-Fourier dissipation is the first and still the most
used and the most successful mesoscopic dynamical theory. Since that overall
2
kinetic energy dr u2ρ(r) (r)
R
is an energy expressed in terms of the state variables
(ρ(r), u(r)) the complete internal energy is as in the equilibrium thermodynam-
ics except that the equilibrium thermodynamics, including the entropy s and
the temperature ϵ∗ = T1 are local. All these quantities are functions of the
position coordinate r.
A need for an extension of the classical fluid mechanics arises for instance
in fluids involving long range (on the macroscopic scale) interactions [20], [21],
in the heat conduction on the microscopic scale [22], [23], [24] and in complex
fluids (as for example polymeric fluids and suspensions [10]). All extensions are
made by adopting extra fields into the set of the state variables.
We illustrate the changes in the physical interpretation of the temperature
caused by enlarging the set of the extra fields in the fluid mechanics on the ex-
ample of a polymeric fluid composed of large macromolecular chains. The state
variables of simple fluids (ρ(r), u(r), ϵ(r)) is replaced by (ρ(r), u(r), ϵpol , e
ϵ(r)),
where ϵ(r) = ϵpol + e ϵ(r), where ϵ(pol) is the energy of the macromolecular chains
on the scale of the macromolecules. If we interact with the polymeric fluid then
the energy that we can directly use is the one that can be expressed in terms
of (ρ(r), u(r)) and the energy ϵ(r) is the internal energy determining the tem-
perature ϵ∗ (r) ∂ϵ(r)
∂s
= T (r)1
. Let us now imagine robots on the macromolecular
scale (MacroMolecular Scale robots (MMS robots)). When they interact with

7
the polymeric fluids then the energy that they directly control and can directly
use is the energy expressed in terms of (ρ(r), u(r), ϵ(pol) (r)). The temperature
(pol)
that they feel is ϵ(pol)∗ (r) ∂ϵ∂s 1 1
(pol) (r) = T (pol) (r) that is different from T (r) . The

MMS robots are meant to represent in our imagination the complex processes
that take place in biological systems.

2 Rate temperatures
The classical equilibrium thermodynamics and the classical fluid mechanics are
well established theories. This means that solutions to governing equations of
a more microscopic well established dynamical theory (as for example a kinetic
theory) has to display an approach to them. Returning to Boltzmann’s pio-
neering investigation of the mesoscopic dynamics, we have already shown that
the approach to the classical equilibrium thermodynamics is indeed displayed
in the original as well as the abstract Boltzmann equation (6). We now turn
to the approach to the classical fluid mechanics. Two paths to investigate this
problem have been suggested and explored. The first, taken by Chapman and
Enskog [25], search an invariant (or approximately invariant) manifold on which
the 1-particle distribution function f (r, v) settles after some time. The second,
introduced by Grad [26], cast first the Boltzmann equation to a hierarchy (Grad
hierarchy) and then closes it. The second path has more the spirit of thermo-
dynamic, we shall therefore follow it.
The 0-th law of equilibrium thermodynamics gave us the physical basis of
the equilibrium thermodynamics. We shall call the existence of the approach
of a well established and autonomous mesoscopic dynamical theory to another
well established and autonomous dynamical theory that takes into account less
details a 0-th law of rate thermodynamics. We use the name ”rate thermo-
dynamics” since in order to formulate thermodynamics we need an approach
to a fixed point (to a state at which no time evolution takes place). This is
indeed the case in the approach to equilibrium states but it is not the case in
the approach to another dynamics. Such approach however turns out to be an
approach to fixed points if we consider the time evolution in the space of vector
fields rather than the time evolution in the state space. The finally reached
fixed points are the vector fields governing the time evolution in the reduced
dynamical theory. The time evolution in the space of vector fields is called a rate
time evolution. We thus call the thermodynamics arising in its investigation a
rate thermodynamics.
The equation governing the time evolution in the space of vector fields is
obtained by lifting Eq.(6) governing the time evolution in the state space to
cotangent (or tangent) bundles. The geometrical formulation of such lifts as
well as their physical interpretation is discussed in detail in [12], [13],[14]. The
essence of the lifts is to consider in (6) the variable x∗ as an independent variable.
We rewrite Eq.(6) into the form

x˙∗ = GRx∗ (x, x∗ ) (7)

8
where
R(x, x∗ ) = e† < x∗ , L(x)ex (x) > −Ξ(x, x∗ ) (8)
is a Rayleighian, G = (ϕxx )−1 , e† = T (rate)
1
, T (rate) is the rate temperature.
The Poisson bivector L as well as the dissipation potential Ξ in (8) are assumed
to be degenerate in the sense that Lsx = 0 and Ξ depends on x∗ only through
its dependence on Kx∗ , where K is a linear operator satisfying Kex = 0.
The Rayleighian plays in the rate thermodynamics the same role as the
thermodynamic potential ϕ (see (4)) plays in the equilibrium thermodynamics.
The first term < x∗ , L(x)ex (x) > is the rate energy (replacing the energy e
in (4)) and the second term Ξ(x, x∗ ) is the dissipation potential (replacing the
entropy s in (4)). Note that the dissipation potential is closely related to the
rate entropy (the rate entropy equals < x∗ , Ξx∗ >, note that in the particular
case when Ξ is a quadratic function of x∗ the entropy production equals twice
the dissipation potential).
A few comments are in order. First, we note that the abstract Boltzmann
equation (6) in its reinterpretation as an equation (7) governing the time evo-
lution of x∗ rather than x recovers and extends the Onsager principle. The
reduced time evolution is obtained as an extremum (it can be shown a mini-
mum, see [12], [13],[14]) of the Rayleighian. This is the statement of the Onsager
principle [27],[28], [29],[30]. Equation (7) shows it and in addition it also shows
how the reduced vector field is reached. Moreover, Eq.(7) also provides the
physical foundation of the Onsager principle (it is the 0-th law of the rate ther-
modynamics).
The second comment is about the presence of external forces. In the equilib-
rium thermodynamics the external forces prevent approach to equilibrium are
thus excluded. They are not excluded in the rate thermodynamics. Dynamical
theories reduce to dynamical theories involving less details also in the presence
of external forces. The Rayleighian can include external forces.
In the third comment we note that the temperature T (rate) in the Rayleighian
is, in general, different from the equilibrium temperature T . The multiplicative
inverse of the rate temperature is the derivative of the rate entropy with respect
to the rate energy. Direct measurements of the rate temperature would need
Wall ∥ freely passing the rate energy and and Wall ♯ stopping its passing. We do
not see how to make such walls. More promising would be to observe fluctuations
of a mesoscopic dynamical theory. We have noted in Section 1.1 that in the
equilibrium thermodynamics the bottom-up extension of the equilibrium theory
to include fluctuations is a reversal process of the top-down reduction governed
by (6). Both the extension and the reduction are driven by the entropy and
both provide ways to measure the temperature. We conjecture that similarly in
the context of the rate thermodynamics a bottom-up extension of a mesoscopic
dynamical theory (for instance hydrodynamics) to include fluctuations that is
reversal process of the top-down reduction governed by (7). Both the extension
and the reduction we expect to be driven by the Rayleighian and both we expect
to offer a way to measure the rate temperatures. We hope to investigate this
conjecture more in the future and we also hope that we shall be joined by

9
interested readers.

3 Nonstandard temperatures
Everyday dealing with hot and cold and with the temperature characterizing
them led to use the temperature also as a measure of, for instance, the inten-
sity of social interactions. In this section we discuss briefly two nonstandard
temperatures.

3.1 Granular temperature

Granular media, as for instance sand, are macroscopic systems that can be
regarded in the same way as we regarded macroscopic systems in the previous
sections. However, their particular structure and particular behavior turn our
attention to features that do not arise and are not studied in other macroscopic
systems. For instance we are interested what triggers an avalanche (i.e. what
triggers granular systems with motionless granular particles to start to behave
as fluids). With such interests in mind Sam Edwards [31] has introduced a
”granular temperature” that he calls fluffiness or also a compactivity.
As we pointed out in Section 1.1 a quantity that plays the role of the temper-
ature in the classical thermodynamics needs (i) a medium having ”an internal
energy”, (ii) walls through which the medium can freely pass or be completely
stopped, (iii) a process of equilibration. Edwards has chosen the medium to be
the space in the granular system that is not occupied by the grains. ”Internal
energy” is its volume, and the equilibration the process is shaking. The problem
of developing this elegant idea into a multiscale theory of granular media has
been found challenging.

3.2 Rumor temperature

In this example we turn to politics. We indeed hear and read the word tem-
perature in this context. For example we read: ”meeting of the two leaders
lowered the temperature”. We suggest the following definition. The medium is
an information related to politics, its ”internal energy” is its seriousness (in the
sense of possible consequences), and the process of equilibration is the passage
of the information by gossip.

References
[1] J. Liouville, Sur la théorie de la variation des constantes arbitraires. Journal
de mathématiques pures et appliquées, 3:342–349 (1838)
[2] Boltzmann, L. Vorlesungen Über Gastheorie, I Teil; R. Barth: Leipzig, Ger-
many, (1896)

10
[3] Grmela, M. Particle and bracket formulations of kinetic equations. Contemp.
Math. 28, 125–132 (1984)
[4] Grmela, M. Bracket formulation of diffusion-convection equations. Physica
D, 21, 179–212 (1986)

[5] Grmela, M.; Öttinger, H.C. Dynamics and thermodynamics of complex flu-
ids: General formulation. Phys. Rev. E, 56, 6620 (1997)
[6] Öttinger, H.C.; Grmela, M. Dynamics and thermodynamics of complex flu-
ids: Illustration of the general formalism. Phys. Rev. E 56, 6633 (1997)
[7] Morrison, P.J. Bracket formulation for irreversible classical fields. Phys. Lett.
A, 100, 423–427 (1984)
[8] Kaufman, A.N. Dissipative Hamiltonian systems: A unifying principle.
Phys. Lett. A, 100, 419–422 (1984)
[9] Arnold, V.I. Sur la géometrie différentielle des groupes de Lie de dimension
infini et ses applications dans l’hydrodynamique des fluides parfaits. Ann.
Inst. Fourier, 16, 319–361 (1966)
[10] Pavelka, M.; Klika, V.; Grmela, M. Multiscale Thermo-Dynamics; De
Gruyter: Berlin, Germany, (2018)
[11] Grmela, M. Thermodynamical lift of the nonlinear Onsager-Casimir vector
field. In Proceedings of the Workshop on Hamiltonian Systems, Transforma-
tion Groups and Spectral Transform Methods, Montreal, QC, Canada, 20 - 26
October 1989; Harnad, J., Marsden, E.J., Eds.; CRM Publisher: Montreal,
QC, USA, 199 - 207 (1990)
[12] O. Esen, M. Grmela, M. Pavelka, On the Role of Geometry in Statistical
Mechanics and Thermodynamics I: Geometric Perspective, J. Math Phys. 63,
(12) 12902, 1-32 (2022)
[13] O. Esen, M. Grmela, M. Pavelka, On the Role of Geometry in Statistical
Mechanics and Thermodynamics II: Thermodynamic Perspective, J. Math.
Phys. 63, (12) 123305 1-21 (2022)

[14] M.Grmela, Thermodynamics and Rate Thermodynamics, J.Stat.Phys.

191:75 (2024)
[15] H. Callen, Thermodynamics: an introduction to the physical theories of
equilibrium thermostatics and irreversible thermodynamics. Wiley, (1960)

[16] P. de Gennes, Reptation of a polymer chain in the pres- ence of fixed

obstacles J Chem Phys. 55, 572–9 (1971)
[17] Ruggeri, T.; Sugiyama, M. Rational Extended Thermodynamics Beyond
the Monoatomic Gas; Springer: Berlin, Germany, 2015.

11
[18] van Kampen, N, Condensation of classical gas with long-range attraction,
Phys. Rev. 135, A362, (1964)
[19] Euler, L. Principes Généraux du Mouvement des Fluides; Académie Royale
des Sciences et des Belles-Lettres de Berlin, Mémories: Berlin, Germany,
(1755); p. 11; Phys. D 237, 1825 (2008)

[20] Hill, T.L. Thermodynamics of small systems. J. Chem. Phys. 1962, 16,
3182–3197.
[21] R.V. Chamberlin, M.R. Clark, V. Mujica, G.H. Wolf, Multiscale Thermo-
dynamics: Energy, Entropy, and Symmetry, Symmetry 13, 721 (2021)

[22] Jou D, Casas-Vàzquez J and Lebon G, Extended Irreversible Thermody-

namics 4th edn Berlin, Gremany: Springer (2010)
[23] P. Ván, Theories and heat pulse experiments of non-Fourier heat conduc-
tion, Commun. Appl. Ind. Math. 7 (2) 150–166 (2016)

[24] R. Kovács, Heat equations beyond Fourier: From heat waves to thermal
metamaterials, Phys. Rep. 1048, 1-75 (2024)
[25] Chapman, S.C.; Cowling, T.C. The Marhemafical Theory of Non-Uniform
Gases; Cambridge University Press: Cambridge, UK, (1961)
[26] H. Grad, Principles of kinetic theory of gases, Handbuch der Physik 12,
Springer, Berlin, (1958).
[27] Rayleigh, L. Proc. Math. Soc. London, 4, 357 (1873)
[28] L. Onsager. Reciprocal relations in irreversible processes I, II. Physical
Review, 37(4):405, 38(12):2265, 1931.

[29] L.Onsager, S. Machlup, Fluctuations and Irreversible Processes Physical

Review. 91 (6): 1505–1512 (1953)
[30] M.Doi, Onsager’s variational principle in soft matter, J.Phys. Condensed
matter, 23, 284118 (2011)

[31] S.F. Edwards, R.B.S. Oakeshott, Physica A 157 (1989) 1080

12