Eritis sicut Deus, scientes

bonum et malum.

Lecture 3: methods and

terminology, part I
Molecular dynamics, internal coordinates, Z-matrices, force fields,
periodic boundary conditions, Hartree-Fock theory, basis sets, etc.

Dr Ilya Kuprov, University of Southampton, 2012

(for all lecture notes and video records see http://spindynamics.org)
Classical molecular dynamics
Classical MD is not a quantum mechanical method. It amounts to solving New-
ton’s equations for a collection of balls on springs and bendy surfaces:

gradient potential acceleration velocity momentum

   
force   dv  dr  
F   F  ma a v p  mv
dt dt

 dp  
 dt  F  p  t  The record of time dependence of all
     coordinates and momenta in the system is
 dr  p r  t  called the system trajectory.
 dt m

Advantages Drawbacks
Very fast (millions of atoms) Bond breaking and formation cannot be described.
Can predict statistical properties (density, melting Only mechanical, statistical, thermodynamic and
point, vapour pressure, etc.) some electrostatic properties can be computed.
Easy to understand and interpret. Ignores all quantum mechanical effects.

Software: AMBER, GROMACS, HyperChem, Chem3D

Classical molecular dynamics
Classical MD is used for systems and properties that do not exhibit strongly
quantum mechanical behaviour: large proteins, cellular membranes, DNA etc.

Properties that are calculated using mole-

cular dynamics simulations are:
1. Rotational and internal correlation
functions for large biomolecules.
2. Biomolecular geometries from NMR
distance and angle restraints.
3. Diffusion constants.
4. Binding constants and energies for
simple non-covalent interactions.
5. Mechanisms and pathways for simple
conformational transitions.
6. Ensemble thermodynamic properties
(enthalpy, entropy, Gibbs energy, etc.)
7. Energy minimum geometries, crystal
structures and phase diagrams.
Crystal structure of bacterial Photosystem I.

See AMBER and GROMACS manuals for more information.

Molecular dynamics force fields
The term force field, in the molecular dynamics context, refers to the equation
giving the forces on all atoms in the system as a function of internal coordinates.

bonds angles dihedrals

1 1 1
 
bonds 2
k b (l  l0 ) 2
 
angles 2
k a (   0 ) 2
 
dihedrals 2
kd [1  cos(n   )] 

N 1   r 
12
 r0ij   qi q j 
6

N

   i , j    2     F  
0 ij
,
j 1 i  j 1   rij   rij   4 0 rij 
 

short-range Van der Waals Coulomb

repulsion interaction interaction

The parameters of force fields are extracted from accurate quantum mechanical
simulations of a standard set of small molecules. They are assumed to be
transferable to similar structural motifs in larger molecules.
A more expensive method is to extract forces from a separate DFT simulation.

See AMBER and GROMACS manuals for more information.

Internal coordinates and Z-matrices
It is often convenient to
describe molecular geometry
in terms of bond lengths, bond
angles, dihedral angles, etc.,
rather than Cartesian coordi-
nates of atoms.
Internal coordinates are a
coordinate system chosen to
match the structure of the
Cartesian coordinates of methane:
molecule in some way. C 0.000000 0.000000 0.000000
Z-matrix is a standard set of H 0.000000 0.000000 1.089000
H 1.026719 0.000000 -0.363000
internal coordinates, specify- H -0.513360 -0.889165 -0.363000
ing the molecular geometry in H -0.513360 0.889165 -0.363000
terms of bond lengths, bond
Z-matrix of methane:
angles and dihedral angles. C
Coordinate optimizations and H 1 1.089000
H 1 1.089000 2 109.4710
transition state searches often H 1 1.089000 2 109.4710 3 120.0000
converge considerably faster in H 1 1.089000 2 109.4710 3 -120.0000
internal coordinates.

bonds angles dihedrals

User-friendly translation between Cartesians and Z-matrices is done by GaussView.

Redundant internal coordinates
Z-matrix coordinates encoun-
ter problems in linear, planar
and polycyclic systems, where
they are inferior to Cartesians.
A well-chosen coordinate sys-
tem should also minimize not
just the linear and quadratic,
but also higher order couplings
between coordinates.
Redundant internal coordinates
attempt to solve both problems
by starting with a redundant
set (e.g. all bonds, all angles
and all dihedrals) and using
SVD to separate out the best
independent linear combinati-
ons. The result is nearly always
superior to Cartesians, and
also allows freezing and scan-
ning arbitrary coordinates.

P. Pulay, G. Fogarasi, J. Chem. Phys. 96(4):2856-2860, 1992.

Periodic boundary conditions
Effects of the sample boundary (container walls or
vacuum) are often undesirable in bulk property
simulations. One way of getting around this
problem is to make atoms and interactions pro-
pagate cyclically across the container boundaries –
this is known as periodic boundary condition.

The container must be large enough to prevent

parts of the system from interacting with their own
images – at least 10 Å of padding is recommended.

A special case is slab boundary conditions, where

the system is periodic in two dimensions.

The total charge of the system must be zero to

avoid summing to an infinite charge. Angular
momentum is not conserved in PBC systems.

The cell geometry need not be rectangular – any

tiling of 2D or 3D space (e.g. with truncated
octahedra) may be used.

Periodic boundary systems are conveniently set up in HyperChem.

Hartree-Fock theory
Hartree-Fock theory gives approximate solutions to the following quantum
mechanical problem:

electron- inter- inter-

nuclear nuclear electron
electron momenta nuclear
momenta attraction repulsion repulsion
energy

 1 1  2
ZN ZN ZN 1  
  k         r   E  r 
2 N

 2 k 2 N mN Nk rNk N  N rNN  j  k rjk 


wavefunction

Solving this for a single electron in an

average electric field of all other electrons GUESS
is relatively easy, and HF method proce-
eds by treating each electron in turn in a Solve for Solve for
cycle (see the schematic on the right). electron 1 electron 2

The algorithm updates the wavefunctions

for each electron until the solution no lon-
Solve for Solve for
ger changes. This procedure is known as
electron N electron 3
the self-consistent field method.

Software: Gaussian, GAMESS, HyperChem, etc.

Hartree-Fock theory
Drawbacks Born-Oppenheimer approximation: the
Ignores two-electron correlation effects (moving electrons are assumed to react instan-
electrons avoid each other in ways that the average taneously to any rearrangement of the
field theory does not properly capture). nuclei. As a result, nuclear degrees of
freedom can be formally eliminated from
Relativistic effects are usually ignored completely. the equation in the previous slide.
Nuclei are assumed to be static point charges.

One of the Hartree-Fock orbitals in a fullerene derivative.

A single configuration ground state is assumed.

Slater determinants: a basic constraint on fer-

mionic systems requires that wavefunctions
change sign if the labels of any two particles are
permuted. For one-electron orbitals, the simplest
way of doing it is to put them into a determinant:

1 (r1 ) 2 (r1 )   N (r1 )

1 1 (r2 ) 2 (r2 )   N (r2 )
 (r1 , r2 ,, rN ) 
N!   
1 (rN ) 2 (rN )   N (rN )

A. Szabo, N.S. Ostlund, Modern Quantum Chemistry, Chapter 3.

Basis sets
A basis set is a way of converting integrodifferential equations into matrix
equations, which are much easier to solve on digital computers.

 1      
 2 k
  2
 U  r    r   E  r    r    ck k  r 
 k  k
basis
functions
wavefunction
Hamiltonian
coefficients

The equation for the wavefunction is replaced by a system of linear equations for
the expansion coefficients:

 1    
k k  2 k k
c   2
 U  r  
 k

 r   E 
k
c 
k k  r   H
k
c  Ecn
nk k

can be pre-computed simple linear

equations

The task of computing multi-dimensional integrals involving basis functions is

very formidable, but...

A. Szabo, N.S. Ostlund, Modern Quantum Chemistry, Chapter 3.

Gaussian basis sets
Property Benefits Atom-centred spherical coordinates are
A product of two All two-electron integrals usually used in atomic basis sets:
Gaussian functions is are analytical. Massive
another Gaussian at their time saving compared to Gn ,l ,m  r ,  ,    Yl ,m  ,   e  r 2 bn2
centre of gravity. numerical integration.
Distance based cut-offs spherical Gaussian
Gaussians fall off very
further improve perfor- harmonic
steeply with distance.
mance.
Linear combinations of Gaussians
(called contractions) are often used in
Property Drawbacks the radial part:
Incorrect asymptotic
Gaussians fall off very
behaviour at large
Gn ,l ,m  r ,  ,    Yl ,m  ,    ani e  r
steeply with distance. 2
bni2
distances.
Incorrect asymptotic i
Gaussians have zero
behaviour at small
derivative at the origin. contraction
distances.
coefficients
Gaussians forming the
Basis set superposition
basis are centred at the
error.
nuclei.

Gaussian basis sets are used by Gaussian03, GAMESS and similar packages.
Properties of atomic basis sets
Property Explanation
Basis functions intended to provide the description of the core electrons of an
Core functions
atom. Core electrons (e.g. 1s for carbon) do not participate in chemical bonding.

Basis functions intended to provide the description of the valence electrons of an

Valence functions
atom that are involved in chemical bonding. Difference to core is often notional.

Number of independent sets of radial basis functions in the valence part of the
Zeta number
basis (usually 2-4). They are introduced to improve basis flexibility.

Additional functions with a higher angular momentum number, introduced to

Polarization functions
allow atomic orbitals to be shifted (“polarized”) relative to their nucleus.

Additional functions with very broad profile, introduced to improve description of

Diffuse functions
the regions that are far away from the nuclei, usually used for anions.

Additional functions concentrated tightly around their nuclei to improve the

Tight functions
description of the nuclear region for NMR and ESR interaction calculations.

Effective Coulomb interaction potentials designed to replace inner orbitals of

Effective core potentials
heavy atoms that are not well described by non-relativistic methods.

Additional extremely diffuse basis functions required for accurate description of

Rydberg functions
Rydberg states.

A. Szabo, N.S. Ostlund, Modern Quantum Chemistry, Section 3.6.

Gaussian basis sets
Pople’s basis sets: Correlation-consistent basis sets:
6-311G++(3df,3pd) aug-cc-pVTZ
polarization
“correlation consistent
functions
polarized valence triple-
diffuse zeta basis set augmented
valence Gaussians
(for each zeta) Gaussians with diffuse functions”

core Gaussians
Specialized basis sets exist for almost any specific
purpose. Check the literature for any case-specific basis
What to look for: sets before using a general purpose one.

Polarization Diffuse Effective core Rydberg

Zeta number Tight functions
functions functions potentials functions
The more, the The more, the The more, the Only used for Used to Only used to
better/slower, better/slower, better/slower, computing approximate describe
usually 2-3. Usually one or usually a single HFCs and J- inner shells of Rydberg states
two l+1 sets. set for anions. couplings. heavy elements

Bigger and more flexible basis sets provide more accurate matrix representations for
the original differential equations and therefore yield more accurate results.

https://bse.pnl.gov/bse/portal (a collection of popular basis sets)

Plane-wave basis sets
Property Benefits Plane wave basis set use crystal lattice
All two-electron integrals coordinates:
Plane waves are complex
can be computed using
exponentials.   
fast Fourier transforms. G  r   e iG r

Plane-wave basis sets are

Plane waves are periodic. well adapted for periodic reciprocal lattice
boundary conditions. vectors

Uniform description
Plane waves do not decay. without the basis set In a given periodic box, only specific
superposition error. frequencies are allowed in specific
spatial directions:
Self-consistent field
Plane waves are
convergence is guaranteed
orthogonal.
in all situations.
Plane waves are
Simple accuracy criteria.
parameterized by energy.

Property Drawbacks
Plane waves are smooth Wavefunction cusps near
and non-local. nuclei are difficult.

See CASTEP documentation for further information.

Variational principle
Applied to quantum chemical calculations, the variational principle states that in a
given class of functions, the one with the lowest expectation value for the energy is
the best approximation to the ground state of a quantum system.

In practice, the variational princi- Basis set Energy / Hartree

ple guarantees that a bigger basis
STO-3G -0.49491
set would yield a better approxi-
mation, though the law of dimini- STO-4G -0.49848
shing returns certainly applies.
STO-5G -0.49951
STO-6G -0.49983

Because of error compensation effects, the

variational principle does not generally
apply to properties other than energy, and
even to energy differences, as the picture on
the left demonstrates.
For property calculations therefore, a bigger
basis set does not necessarily mean a more
accurate answer, but certainly means a
longer processing time.

P.W. Atkins, R.S. Friedman, Molecular Quantum Mechanics, Chapter 6.