1.

Meshless methods

1. Why mesh free: No re-meshing, Cannot model discontinuities, material breakage.

2. Properties of mesh free shape function
a. Shape function must satisfy the partition of unity and must be stable.
b. The shape function constructed should satisfy a certain order of consistency
(capable of reproducing lowest orders of complete polynomials)
c. The domain for eld variable approximation/interpolation (termed the support
domain or inuence domain or smoothing domain) should be small compared
with the entire problem domain (compact support). Banded Matrix
d. The algorithm should be computationally efcient. It should be of the same
order
of complexity as that of FEM (efciency).
e. Ideally, the shape function should possess the Kronecker delta function property
essential boundary condition implementation
f. Ideally, the eld approximation using the shape function should be compatible
throughout the problem domain. Compatibility refers to the continuity of the
approximation on the boundaries between subdomains, based on which shape
the functions are constructed. Both consistency and compatibility affect the
accuracy and convergence of the numerical results.
3. Comparison with FEM shape function: Do not satisfy kronekar delta. SPH shape
function depends only on the weight function (assume the uniform particle
distribution). Because of this special property of the SPH shape function, the true value
of the eld variable should be retrieved again, after obtaining the nodal parameters at
all the eld nodes (particles).
4. Shape function generation methods
a. Finite integral schemes used in SPH. The function is represented using its
information in a local domain (smoothing domain or inuence domain) via an
integral form. Consistency is achieved by properly choosing the weight
function.
b. Finite series representation methods have a long history of development.
Consistency is ensured by the use of the basis functions. The inclusion of special
terms in the basis can also improve the accuracy of the results for certain classes
of problems. ( ) = ( ) ( ) . The finite integral representation is
called as the kernel approximation.
c. This expression is valid when the weighting function is positive (meaningful
presentation of physical quantities), compact (value of u at a particular node
only depends on nodes within the smoothing domain), and its integral over the
domain is unity (consistency).
5. Choice of weight function
a. In solving any partial differential equation (PDE) based on the weak form
formulation, such as the Galerkin approach, there is a minimum consistency
requirement for ensuring the convergence of the solution from the discretized
equation system. The minimum consistency requirement depends on the order

of the PDE. For a PDE of order 2k, the minimum requirement of the consistency
is
for Galerkin formulation. This is equivalent to the requirement of
representing the polynomial of all orders up to the kth order. The reason is that
an approximation that can exactly represent the polynomial of all the orders up
to the kth order can represent any smooth function and all its derivatives up to
the kth order with arbitrary accuracy as the nodal distance approaches zero.
b. Limitations of meshfree methods: Construction of shape functions time
consuming. Mesh is still required in weak form methods. Larger band width and
non-symmetric matrix. Essential boundary conditions implementation a
problem
c. Basis function A set of linearly independent that are used in a linear
combination for the approximation of any function.
2. Mathematics

1. Root finding methods Newton Raphson and secant method (approximates derivative
with Taylor series).
2. System of equations
a. Cholesky factorization mainly used form symmetric matrices =
. Then
follow LU decomposition
b. Condition no low well-conditioned matrix, if condition no is 10 . K digits of
accuracy would be lost in computation. Denotes how much the solution will
change for a small change in RHS in solving AX = B.
c. Eigen value is the value by which a vector is scaled when multiplied by a
matrix.(A I)I = 0.
d. ODE can be solved using Runge kutta methods; do not require computation
of higher order derivatives.
3. PDE types: Hyperbolic not evolving towards steady state-wave like; parabolic evolving
towards steady state-diffusion; elliptic- Reached steady state.
3. Molecular dynamics

1. Continuum viewpoint no underlying inhomogeneous microstructure, that is, matter can

be divided infinitely without change of material properties (numerical implementation:
FEM) Material properties buried in Youngs modulus E
2. Atomistic viewpoint - consider the discreteness of matter for example, the discreteness
of an atomic lattice in a metal, where atoms are glued to their position. No spatial
discretization necessary given by atomic distances, e.g. lattice.
3. MD is not only suitable for elasticity problems; MD can also be used to solve plastic or
fracture problems (dissipative), naturally, and also capable of solving the dynamical
evolution of non-equilibrium processes.
4. In atomistic simulations, the goal is to model, analyze, and understand the motion of each
atom in the material. The collective behavior of the atoms allows to understand how the
material undergoes deformation, phase changes or other phenomena, providing links
between the atomic scales to meso or macro-scale phenomena.

5. MD generates the dynamical trajectories of a system of N particles by integrating Newtons

equations of motion, with suitable initial and boundary conditions, and proper interatomic
potentials, while satisfying thermos-dynamical (macroscopic) constraints.
6. The atomic displacement field consists of a low-frequency (coarse) and high frequency
part (fine). Time step t needs to be small enough to model the vibrations of atomic
bonds correctly. Vibration frequencies may be extremely high, in particular for light atoms
Thus: Time step on the order of femto seconds. Need 1,000,000 integration steps to
calculate trajectory over 1 nanosecond: Significant computational burden. Time step can
(typically) not varied during simulation; it is fixed. Total time scale O(ns).
7. MD allow to model materials with dimensions of several hundred nanometers and beyond:
Allows to study deformation and properties, mechanisms etc. with a very detailed
computational microscope, thus bridging through various scales from nano to macro.
With the definition of the interatomic potentials (how atoms interact) all materials
properties are defined.
8. Using MD one can obtain thermodynamic, structural, mechanical, dynamic and transport
properties of a system of particles which can be a solid, liquid, or gas. It provides a direct
link between potential model and physical properties. This is really useful from the
standpoint of fundamental understanding of physical matter. It is also very relevant to the
structure-property correlation paradigm in materials science. Complete control over input,
initial and boundary conditions. This is what gives physical insight into complex system
behavior. This is also what makes simulation so useful when combined with experiment
and theory. Detailed atomic trajectories. This is what one can get from MD, or other
atomistic simulation techniques, that experiment often cannot provide. This point alone
makes it compelling for the experimentalist to have access to simulation.
9. The conversion of this microscopic information to macroscopic observables such as
pressure, stress tensor, strain tensor, energy heat capacities, etc., requires theories and
strategies developed in the realm of statistical mechanics. Statistical mechanics is
fundamental to the study of many different atomistic systems, by providing averaging
procedure or links between microscopic system states of the many-particle system and
macroscopic thermo-dynamical properties, such as temperature, pressure, heat capacity etc.
10. Statistical mechanics relates microscopic state of a system to macroscopic observables. The
state of a system is defined through some thermodynamic parameters (T, P, N)
11. Phase space is the space where all possible states of a system are represented all possible
values of position and momentum.
12. Ensemble is a collection of points in a phase space corresponding to a given thermodynamic
state. Different combination of material trajectories/initial conditions and velocities can
lead to the same observed phenomena at macroscale. This collection of points is the
ensemble.
a. NVE Isolated system microcannonical
b. NVT- Cannonical It is also sometimes called constant temperature molecular
dynamics (CTMD). In NVT, the energy of endothermic and exothermic
processes is exchanged with a thermostat. A variety of thermostat methods is
available to add and remove energy from the boundaries of an MD system to
control temperature include velocity rescaling, the Nos-Hoover thermostat.

13. Ensemble average is the average over all points in the phase space in a single time. Dynamic
average is the average over a single point in phase space at all times.
14. Ergodic hypothesis states that over a long period of time the ensemble average is equal to
time average.
15. The Lennard-Jones potential is a mathematically simple model that approximates the
interaction between a pair of neutral atoms or molecules. LJ potential - Sigma denotes the
distance at which attractive and repulsive forces cancel are in equilibrium and well depth
is the strength of the potential.
16. Initial position of the atoms in MD simulation is assigned to reach the required density (and
lattice structure) and the configuration has the lowest energy. Velocities are chosen such
that the net momentum is zero.
17. MD results check
a. Resultant velocity follows a Maxwellian distribution.
b. Energy is conserved in the case of NVE ensemble or NVT with velocity
rescaling turned off.
c. In the case NVT the temperature and volume are conserved.
18. Equilibration The phase in which the velocities are scaled up to reach a given temperature,
after which properties are monitored.
19. The interaction between the particles is either described by a "force field" (classical MD),
a quantum chemical model, all called the potential energy or just the "potential".
20. MD is a special case of DEM where particles are spherical in shape and DEM the rotation
of the particles are considered.
21. A molecular dynamics simulation requires the definition of a potential function/force field,
or a description of the terms by which the particles in the simulation will interact. A force
field (or interatomic potentials) refers to the functional form and parameter sets used to
calculate the potential energy of a system of atoms or coarse-grained particles MD
simulations. The parameters of the energy functions can be derived from experimental work
and quantum mechanical calculations. Force fields are used by chemist while engineers use
inter-atomic potentials. Particle-particle interactions can reproduce structural and
conformational changes but not reproduce chemical reactions. The reduction from a fully
quantum description to a classical potential entails two main approximations.
a. BornOppenheimer approximation, which states that the dynamics of electrons
is so fast that they can be considered to react instantaneously to the motion of
their nuclei. As a consequence, they may be treated separately.
b. The second one treats the nuclei, which are much heavier than electrons, as point
particles that follow classical Newtonian dynamics.
22. The potential functions representing the non-bonded energy are formulated as a sum over
interactions between the particles of the system. The simplest case the total PE can be
calculated from the sum of energy contributions between pairs of atoms. In many-body
potentials, the potential energy includes the effects of three or more particles interacting
with each other. The forces are obtained as the derivative of the Potential energy from these
potentials.
23. In simulations with pairwise potentials, global interactions in the system also exist, but
they occur only through pairwise terms. In many-body potentials, the potential energy