Opportunities in Scientific Computing with
Emphasis on Atomistic Simulations
Texas Tech University, February 5, 2008
Scientific Computing/Computational Science
CDSSIM – Cyberinfrastructure for Chemical Dynamics
Simulations; Sailesh Baidya, U. Lourderaj, Yu Zhuang
Chemical Dynamics Simulations
● Cl- + CH3I → ClCH3 + I- SN2 nucleophilic
substitution reaction; Jiaxu Zhang, Upakarasamy
Lourderaj
● Protonated peptide ion surface-induced
dissociation; Bipasha Deb, Kyoyeon Park,
Wenfang Hu,Kihyung Song
Acknowledgements
Research Group
S. V. Addepalli
Sailesh Baidya
Wenfang Hu
Kyoyeon Park
Khatuna Kakhiani
U. Lourderaj
Mingying Xue
Li Yang
Jiaxu Zhang
Simulation Collaborations
Kihyung Song
NSF-PIRE Researchers
Phil Smith (HPCC)
Yu Zhuang
Funding
National Science Foundation
Welch Foundation
Computer Programs
VENUS
VENUS/MOPAC
VENUS/NWChem
VENUS/GAMESS
WEBSERVICES FOR CHEMICAL DYNAMICS
RESEARCH AND EDUCATION
• monte.chem.ttu.edu – animations of chemical dynamics
simulations; tutorials for chemical dynamics instruction (high
school, undergraduate, graduate), current tutorial for gas-phase
SN2 reactions and more are planned.
• cdssim.chem.ttu.edu – library of chemical dynamics
computer programs and simulation models; user interface for
modifying simulation models; resources for performing a
chemical dynamics simulation and animating the result.
• pire-europe.chem.ttu.edu – NSF Partnership in International
Research and Education (PIRE), “Simulations of Electronic
Non-Adiabatice for Reactions with Hydrocarbon Liquids,
Macromolecules and Surfaces”
COMPUTATIONAL
SCIENCE/SCIENTIFIC COMPUTING
The term “computational science” was first used by
Ken Wilson (awarded a Nobel Prize in physics) to
refer to those activities in science and engineering
(and now also medicine!) that exploit computing as
their main tool.
In the spring of 1994, the IEEE began publishing a
magazine called “IEEE Computational Science and
Engineering”
COMPUTATIONAL
SCIENCE/SCIENTIFIC COMPUTING
Applied Math
Computer
Science
Engineering/
Science/
Medicine
Scientific
Computing
ATOMISTIC SIMULATION RESEARCH AT TTU
ALGORITHM DEVELOPMENT FOR SCIENTIFIC COMPUTING
Edward J. Allen (Mathematics) - development and analysis of numerical
methods to solve nuclear engineering problems such as neutron transport
and reactor kinetics.
Thomas L. Gibson (Physics) - parallelization of sequential Monte Carlo code for
modelling lipids in cell membranes; development of the distributed Positron Model.
William L. Hase (Chemistry) – efficient numerical integration algorithms for
solving coupled, non-linear differential equations.
Rajesh Khare (Chemical Engineering) – Efficient methods for saddle point
determination on potential energy surfaces, multi-scale simulations.
Jorge A. Morales (Chemistry) – efficient numerical integration algorithms for
solving coupled, non-linear differential equations.
Bill Poirier (Chemistry) – efficient linear solvers and eigensolvers for terascale
parallel computing, linear partial differential equations, sparse linear algebra.
COMPUTATIONAL BIOLOGY
William L. Hase (Chemistry) – dissociation of peptide ions, dynamics of enzyme
catalysis.
Rajesh Khare (Chemical Engineering) – molecular simulations of DNA dynamics,
lubrication in human joints.
Jorge A. Morales (Chemistry) – time-dependent, coherent-states dynamics of
inter- and intra- molecular electron transfers coupled to nuclear motion in
prototypical organic and biochemical molecules (e.g. small bridged donoracceptor systems, small peptides)
Mark Vaughn (Chemical Engineering) – 1. Molecular dynamics of lipid bilayer
membranes with and without embedded protein. Single, binary and ternary lipid
compositions. Particularly interested in mechanically strained membranes.
2. Molecular dynamics of solid-tethered DNA oligomers.
CHEMICAL REACTION DYNAMICS
David Birney (Chemistry) – potential energy surfaces of organic reactions,
pseudopericyclic, and pericyclic reactions.
William L. Hase (Chemistry) – dynamics of organic reactions, gas-surface
collisions, dynamics of unimolecular dissociation.
Jorge A. Morales (Chemistry) – Development of a unifying quantum/classical
coherent states (CS) dynamics for all molecular particles (nuclei and electrons)
and for all the molecular degrees of freedom (i. e. translational, rotational,
vibrational, and electronic CS) along with its computational implementation.
Theory applied to reactive collisions and charge transfer processes.
Bill Poirier (Chemistry) – exact quantum dynamics of gas phase reactions,
thermal rate constants relevant to atmospheric and combustion chemistry,
rovibrational dynamics of rare gas clusters.
MATERIALS SCIENCE
Stefan K. Estreicher (Physics) – properties of defects in group IV and III-N
semiconductors, free energies, vibrational dynamics, vibrational lifetimes, MD
simulations.
William L. Hase (Chemistry) – tribology dynamics, heat transfer and structures
at interfaces.
Rajesh Khare – molecular simulations of nanofluidic devices, lubrication,
phase equilibria, properties of supercooled liquids and glassy polymers, rates of
activated processes.
Charles W. Myles (Physics) - theoretical and computational materials physics,
with emphasis on semiconductor materials. High electric field transport.
clathrates and other exotic materials. Electronic properties of defects, electronic
bandstructures, properties of semiconductor alloys. Molecular Dynamics and
Monte Carlo computer simulations.
Mark Vaughn (Chemical Engineering) - probabilistic potential theory (a
stochastic method for solving elliptic PDEs) to compute averaged properties of
dense, reactive suspensions.
SOFTWARE DEVELOPMENT
Thomas L. Gibson (Physics) - MPI computer programs (PATMOL) to calculate
the positron-molecule interaction potential for use in quantum scattering
calculations.
William L. Hase (Chemistry) – computer programs for chemical dynamics
simulations, VENUS and VENUS/NWChem; scientific computing website
“cdsism.chem.ttu.edu”.
Rajesh Khare (Chemical Engineering) – Molecular dynamics simulation
codes (in FORTRAN) for nanofluidics and interfacial heat transfer.
Jorge A. Morales (Chemistry) – development of the code CSTech (“Coherent
States at Tech”) to implement the above-mentioned coherent states dynamics.
This involves compute grid development and code parallelization inter alia.
Bill Poirier (Chemistry) - development of the ScalIT package for performing
sparse iterative linear algebra on massively parallel computers.
ENERGY OF MOLECULES
E  T ( p )  V (q )
Molecular Mechanics (MM) Potential Energy Function
bond stretch
V  f (r  r0 ) 2
torsion angle bend
V  V0 sin n
r

bond angle
V  f (   0 )
2
non-bonded
V  a / r12  b / r 6
r
ENERGY OF MOLECULES
Time-independent Schrödinger Equation from
Quantum Mechanics
Hˆ (r; q)(r; q)  Er (r; q)
r - coordinates of electrons, q - coordinates of nuclei
Ĥ - Hamiltonian operator,  - wave function
Er - energy of electrons
V (q)  Er (q)  VN (q)
VN – nuclear-nuclear repulsion
ENERGY OF MOLECULES
Solving Hˆ (r; q)(r; q)  Er (r; q), (q are fixed)
2

Hˆ (r )  Tˆ (r )  Vˆ (r )    
i
 2mi
E (q)  (r ) Hˆ (r )
 2
i  V (r )

2
2
2
  2 2 2
xi yi zi
2
i
Expand ( r ) in a basis set, developed from experience and
chemical intuition:
( r )   ci i ( r )
E ( q )
0
Variational Theorem: Minimize E(q) with respect to ci :
ci
Leads to a set of secular equations which are solved by linear
algebra (i.e. matrix operations)
QUANTUM THEORY OF
MATTER
“The underlying physical laws necessary for the
mathematical theory of a large part of physics and
the whole of chemistry are thus completely known,
and the difficulty is only that the exact applications
of these laws leads to equations much too
complicated to be soluble.”
P. A. M. Dirac, 1929
COMPUTATIONAL AND
THEORETICAL CHEMISTRY
Development of approximate models that are
instructive and, after careful testing and further
improvement, often give results in agreement with
experiment and are predictive.
Nobel Prize in Quantum Chemistry
 Walter Kohn – Density Functional Theory
 John Pople – Method of Quantum Chemistry
Computations (embodied in the Gaussian
computer program)
ATOMIC-LEVEL MOTION
 Classical Mechanics (approximate)
Newton’s equations of motion
d 2 qi
V ( q )
Fi  
m 2
qi
dt
i = 1,..3N for N - atoms
 Quantum Mechanics (exact)
 ( q , t ) ˆ
i
 H ( q, t )
t
CHEMICAL DYNAMICS SOFTWARE AND
SIMULATION (CDSSIM) SYSTEM
(http://cdssim.chem.ttu.edu)
Library of open source computer programs, documentation, and
simulation models (input files) for distribution.
Software Tools
•To upload, modify, and build simulation models
•Animate chemical dynamics simulations.
Computational Resources for performing chemical dynamics
simulations. Goal is to have grid computing resources.
Cyberinfrastructure for sharing chemical dynamics software.
COMPUTER PROGRAMS
Current
•RRKM – microcanonical RRKM calculations.
•VENUS96 – classical trajectory simulations for a variety of
initial conditions, and assortment of analytic potentials.
Under Development
•VENUS06 – gas/surface scattering, new integration algorithms,
electronic non-adiabatic transitions (?), and other options.
•VENUS/MOPAC – semiempirical QM and QM/MM direct
dynamics.
•VENUS/NWChem and VENUS/GAMESS – interfaces for QM
and QM/MM direct dynamics.
•Links – to other websites like POTLIB (Ron Duchovic) a library
of analytic potentials to link to VENUS.
CLASSICAL TRAJECTORY SIMULATION
Solve Hamilton’s classical equations of motion
Energy = E = H = T(p) + V(q)
∂pi/∂t = -∂H/∂qi and ∂qi/∂t = ∂H/∂qi
or Newton’s classical equations of motion, - ∂V/∂qi = mi ∂2V/∂t2
General classical trajectory computer program VENUS
I. Potential Energy Surface V
A. Analytic potential energy function – ab initio calculations and/or experimental
data such as equilibrium geometries, vibrational frequencies, activation energies,
potential well depths, and heats of reaction.
B. Direct dynamics simulations – obtain the gradient ∂V/∂qi directly from electronic
structure theory; i.e. time-independent quantum mechanics.
II. Initial Conditions – select random initial conditions, for an ensemble of trajectories
to model experiment.
III. Numerically Integrate the Classical Equations of Motion
IV. Trajectory Results – transform final momenta and coordinates into product energies,
structures, etc.
DIRECT DYNAMICS COMPUTER CODES
The general chemical dynamics computer program VENUS is interfaced with different
electronic structure computer programs
VENUS has:
 Many different analytic potential energy functions, including reactive and nonreactive MM functions.
 A suite of numerical integrators.
 A variety of options for choosing trajectory initial conditions to model different
kinds of experiments.
 Algorithms for analyses of trajectory results.
VENUS/Electronic Structure Packages for Direct Dynamics:
 VENUS/NWChem and VENUS/GAMESS under development.
 VENUS/MOPAC, developed but needs documentation.
 VENUS/MNDO99 ( with updates) under development.
SIMULATION MODELS FOR VENUS06
Select Type of Simulation Model
 Bimolecular Collisions
Collision Between Two Gas-Phase Molecule
 Unimolecular Decomposition
Decomposition of a Vibrationally Excited Molecule
 Intramolecular Dynamics
Dynamics of Vibrational Energy Flow Within a Molecule
 Reaction Path Following and VTST
Following the Reaction Path and Calculating the Variational Transition State
Theory (VTST) Rate Constant for association of Two Particles
 Normal Mode Analysis
Vibrational Frequencies for a Molecule, Surface, Cluster, etc.
 SN2 Dynamics
Dynamics of Gas-phase X- + CH3Y SN2 Nucleophilic Substitution Reactions
 Minimum Energy Geometry Optimization
Finding a Minimum Energy Geometry for a Molecule, Surface, Cluster, etc.
 Neutral Atom Small Molecule Collision with a Surface
 Projectile Ion Collision with a Surface
Ar + H2O Collision
Initial Conditions for H2O
Movie Maker
Diglycine-H+ + Diamond {111}
Animation
CDSSIM: WEB-BASED
CHEMICAL DYNAMICS SIMULATIONS
A Tool for On-line Chemical Dynamics Simulations from
Your Desk Computer
•Chemical dynamics computer programs and documentation.
•Tools for using existing, modifying, and building simulation
models (input files).
•Computational resources for performing simulations.
•Tools to animate the simulations.
Goal: Facilitate access to high-level chemical dynamics simulation
tools and software, and motivate collaborative science.
Dynamics of the Cl- + CH3I → ClCH3 + ISN2 Nucleophilic Substitution Reaction
Collaboration: between Roland Wester Research Group,
University of Freiburg, Germany (Experiments); and Bill Hase
Research Group, Texas Tech University (Chemical Dynamics
Simulations).
Traditional Atomic-Level Reaction Mechanism:
Cl- + CH3I → Cl----CH3I → ClCH3---I- → ClCH3 + I-
Science 319, 183 (2008)
Direct mechanism
Roundabout mechanism
SURFACE-INDUCED DISSOCIATION (SID)
(I+)*
I+
(I+)* → Product
Mass spectrometry technique to determine ion (I+)
structure and energetics. Need to know energy transfer
efficiencies to interpret experiment.
f
i
Etrans
 Etrans
 Esurface  Eint
Trajectory Simulations
Ions: Si(Me)3+, Cr(CO)6+, protonated glycine and alanine
peptide ions
Surfaces: CH3(CH2)nS & CF3(CF2)nS/Au{111}, diamond{111}
Experimental Collaborators
Luke Hanley, Julia Laskin and Jean Futrell, Vicki Wysocki
Surface: CF3(CF2)7S-Au
Protonated peptide
+
TOP
VIEW
SIDE
VIEW
Energy Transfer Dynamics
Cr+(CO)6, Ei = 30 eV, Θi = 45o.
Samy Meroueh
PCCP 3, 2306 (2001)
Average per-cent transfer to
∆Eint for the H-SAM.
Simulation: 10 %.
Experiment: 11-12%; Cooks
and co-workers.
Energy Transfer Dynamics
Cr+(CO)6, Ei = 30 eV, Θi = 45o.
Samy Meroueh
PCCP 3, 2306 (2001)
Average per-cent transfer to
∆Eint for the H-SAM.
Simulation: 10 %.
Experiment: 11-12%; Cooks
and co-workers.
Fragmentation Dynamics
Diglycine + Diamond {111}, Ei = 70 eV, θi = 0o
QM+MM, Peptide described by AM1 (VENUS/MOPAC)
Diglycine + Diamond {111}, Ei = 70 eV, θi = 0o
QM+MM, Peptide described by AM1 (VENUS/MOPAC)
SHATTERING FRACTION VERSUS COLLISION ENERGY
FOR (gly)2-H+ + DIAMOND {111}a
Y. Wang and K. Song, J. Am. Soc. Mass Spectrom. 2003, 14, 1402.
Collision Energy (eV)
Shattering Fractionb
30
50
70
100
0.08
0.13
0.44
0.71
a. The collision angle is 0 degrees, perpendicular to the surface.
The trajectories are QM+MM, with QM AM1 for the peptide.
b. Fraction of the trajectories which shatter.
Thanks!
Questions?