Computer simulations based on quantum chemistry and molecular dynamics
applied to challenging problems in chemistry, biochemistry and
materials science. Development of software and computer models are
targeted at problems of relevance to production of renewable fuels and
chemicals from biomass, health and drug design.
Building the Computational Tools for Scientific Discovieries
-
High performance computing (HPC) and large scale, data intensive
scientific computing with focus on (bio)chemistry and nanotechnology.
-
Highly parallel algorithms and software implementations
for predictive computational science through quantum mechanical (QM)
and classical atomistic numerical simulations on well established and
emerging HPC platforms including graphics processing units (GPUs).
-
Implementations into (mainly) the
ADF software package for density
functional theory (DFT) and
the AMBER software package for
classical MM and mixed QM/MM molecular dynamics (MD)
simulations, both used by a large number of academic and
industrial researchers world wide.
-
Education and training of the next generation of scientists in
software development and numerical simulation methods, quantum
chemistry and molecular dynamics.
Research Sponsors
|
|
My research on QM/MM simulation software featured on the cover
page of the Jan 2014 edition of the Journal of Computational
Chemistry.
|
|
Workshops and Symposia
Computational Chemistry Across Catalysis, ACS National Meeting San Diego (March 2016)
In March 2016 I organized a 4-day symposium at the spring 2016 ACS
national meeting that brought together researchers that share a
common interest in the computational modeling of catalytic
processes.
Presentations spanned all domains of catalysis including homogeneous
and heterogeneous catalysis, biocatalysis, photocatalysis,
and electrocatalysis.
A wide range of modeling methods were discussed, including ab initio
electronic structure theory, kinetics simulations (mean field,
KMC), molecular dynamics, non-adiabatic dynamics and free energy
perturbations. Co-organizers of the symposium were my collaborators
Dion Vlachos from University of Delaware and Carine Michel and
Philippe Sautet from the ENS Lyon in France.
GPU Computing Symposium and Workshop at SDSC (November 2013)
This two day GPU computing event
was organized by myself and Ross Walker (co-directors of the CUDA Teaching
Center at SDSC) with help from Jon Saposhnik (NVIDIA).
The symposium covered trends, tools and research discoveris using GPU
accelerated computing in areas ranging from pharmaceutical research to
geophysics
(see the program). The workshop
featured lectures on GPU programming using both CUDA and OpenAcc as well as
hands-on exercises.
AMBER workshop at ECNU in Shanghai (August 2011)
I am organizer and instructor of an
AMBER workshop
at East China Normal University
(ECNU) in Shanghai. This five day workshop (22-26 August, 2011) aims to
introduce researchers in the field of (bio)molecular simulations to the
broad collection of computational tools implemented in
the AMBER and AmberTools software packages
for molecular dynamics simulations.
ADF workshop at SDSC (March 2011)
Together with Dr. Matt Kundrat from SCM I
have organized and tutored
an ADF workshop
that was held on Thursday, 24 March 2011 at
the San Diego Supercomputer Center. The
workshop was geared both at begineers and expert users of ADF wishing to
learn about the new features of the ADF2010.02 release.
GPU Accelerated Molecular Dynamics for Biomolecular Simulations
Work done at SDSC since June 2009
Graphical processing units (GPUs) offer a tremendous amount of
computing power in a compact package (see graphics to the right). Both
the peak floating point operations per second and the memory bandwidth
of NVIDIA GPUs compare favorably to Intel CPUs.
However, this comes at the cost of reduced flexibility and increased
programming complexity as compared to CPUs.
I am involved in an effort to port the all-atom classical MD engine
PMEMD of the AMBER biomolecular
simulation package to run on GPUs [1-3].
The AMBER implementation
uses CUDA which is
a relatively simple extension of the standard C programming language
that allows one to code in an inherently parallel fashion and perform
all necessary operations to access and manipulate data on a GPU
device.
|
Peak floating point operations per second (Flop/s; left) and
memory bandwidth (right) for Intel CPUs and NVIDIA GPUs. (Taken
from our publication [1]
on GPU accelerated MD)
|
Throughput timings (ns/day) for AMBER GB simulations with a
time step of 2fs using the parallel CPU version and the parallel GPU
version with the SPDP precision model.
|
|
The implementation supports severl precision models in which the
contributions to the forces are calculated in single precision floating
point arithmetics but accumulated in double precision (SPDP),
accumulated in 64-bit fixed-point integers (SPFP, default model),
or everything is computed in single precision (SPSP, now deprecated),
or everything in double precision (DPDP).
The numerical noise due to rounding errors within the SPSP precision model
is sufficiently large to lead to an accumulation of errors which can result
in unphysical trajectories for long timescale simulations.
The mixed-precision SPFP model is currently recommended since the numerical
results obtained are comparable with those of the full double precision DPDP
model and the reference double precision CPU implementation but at
significantly reduced computational cost.
The AMBER GPU implementatin provides performance for GB MD simulations
on a single desktop machine that is on part with, and in some cases
exceeds, that of traditional supercomputers (see table on the left).
We achieve a similar performance advantage for PME based simulations
under periodic boundary conditions with up several hundred thousands
of atoms.
With GPUs becoming ubiquitous in workstations and also as acelerators
in HPC platforms, the impact of this implementation on the field of
molecular dynamics is broad and transformative.
|
[1] A. W. Götz, M. J. Williamson, D. Xu, D. Poole, S. Le Grand,
R. C. Walker,
J. Chem. Theory Comput. 8, 1542 (2012).
DOI: 10.1021/ct200909j
[2] S. Le Grand, A. W. Götz, R. C. Walker,
Comput. Phys. Commun. 184, 374-380 (2013).
DOI: 10.1016/j.cpc.2012.09.022
[3] R. Salomon-Ferrer, A. W. Götz, D. Poole, S. Le Grand, R. C. Walker,
J. Chem. Theory Comput. 9, 3878-3888 (2013).
DOI: 10.1021/ct400314y
(back to top)
QM/MM for Biomolecular Simulations
Work done at SDSC since June 2009
QM/MM approaches in AMBER
Hybrid quantum mechanical and molecular mechanical (QM/MM) approaches are
used extensively to study local electronic events in large molecular
systems with a diverse area of applications ranging from enzymatic catalysis
to properties of materials systems.
I have written an implementation that couples the MD software AMBER to
electronic structure software packages including ADF, GAMESS-US,
Gaussian, Orca, NWChem and TeraChem [1,2]. With this interface, ab
initio wave function theory and DFT methods become available for
QM/MM MD simulations with AMBER.
This represents a set of widely used programs, both commercial and
freely available, each with its own strengths for different electronic
structure methods and computing platforms ranging from desktop
workstations to supercomputers. In the case of TeraChem this also
includes accelerator hardware in form of graphics processing units
(GPUs).
Data exchange between the AMBER and the QM software is implemented by
means of files and system calls or the message passing interface (MPI)
standard.
This interface was used for example to study the absorption spectrum
of the photoactive yellow protein (PYP) (see picture on right). It was
shown that a large QM region (hundreds of atoms) is required to
achieve a converged spectrum [2].
[1]
A. W. Götz, M. A. Clark, R. C. Walker,
J. Comput. Chem. 35, 95-108 (2014).
DOI: 10.1002/jcc.23444
[2]
C. M. Isborn , A. W. Götz , M. A. Clark , R. C. Walker,
T. J. Martínez,
J. Chem. Theory Comput. 8, 5092-5106 (2012).
DOI: 10.1021/ct3006826
|
The picture shows a snapshot of PYP (photoactive yellow protein). The
chromophore (highlighted) and its environment are treated quantum
mechanically. See publication [2].
|
|
The picture shows the scheme used for adaptive QM/MM
simulations. Solvent can move via a transition region between
the active region (QM) and the Environment (MM).
|
|
Adaptive QM/MM approaches
I have implemented a parallelized adaptive QM/MM approach into the
AMBER MD software package [1]. It is based on a method developed by
Rosa Bulo [2].
Adaptive QM/MM enables the QM treatment of an active region of a
solvated molecular system including the solvent molecules in its
vicinity (see left). The environment is handled with MM to reach high
computational efficiency. During a simulation, solvent molecules can
move via a transition region between the active and environment
regions.
I was awarded a TRO (Triton Research Opportunity) grant by SDSC to
collaborate with Francesco Paesani and use my implementation to study
processes of relevance for atmospheric chemistry. Together with
Kyoyeon Park we have shown the potential of adaptive QM/MM for MD
simulations of aqueous systems [3].
Adaptive QM/MM can be applied to lots of exciting problems, for example
solvent effects on absorption spectra, binding free energies of ions
or drug-like molecules to proteins or enzymes.
[1]
A. W. Götz, K. Park, R. E. Bulo, F. Paesani, R. C. Walker,
publication in preparation (2015).
[2]
R. E. Bulo, B. Ensing, J. Sikkema, L. Visscher,
J. Chem. Theory Comput. 5, 2212-2221 (2009).
[3]
K. Park, A. W. Götz, R. C. Walker, F. Paesani,
J. Chem. Theory Comput. 8, 2868-2877 (2012).
DOI: 10.1021/ct300331f
|
(back to top)
Subsystem Density Functional Theory
Postdoc in Amsterdam with Dr. L. Visscher (November 2006 - April 2009)
I have been and still continue to work on the extension of the frozen density
embedding (FDE) method in density functional theory (DFT).
The main research topics are analytical gradients for FDE and nuclear
spin-spin coupling constants in the framework of FDE.
The research work is done with
Dr. L. Visscher
at the chair of theoretical chemistry at the
Vrije Universiteit Amsterdam
which is headed by Prof. Dr. E.-J. Baerends.
DFT is undoubtedly the most popular computational method for the investigation
of the electronic structure of molecules. It allows to obtain accurate
information on molecular properties at a moderate computational cost.
Studies of molecules of interest for classic organic or inorganic chemistry
are routine by now.
It is the time to find new approaches to be able to study also more complex
systems in fields of increasing importance such as life sciences and
nanotechnology.
The extension of FDE to a general subsystem DFT has a tremendous potential as
an accurate (in principle exact) multi-scale modelling method. Such methods
allow to focus the computational effort on those parts of the system which is
of importance for the property of interest, while still taking into account
the interaction with the remaining parts of the system.
Proper implementations will allow to tackle the investigation of properties of
chemical systems of unprecedented size and complexity.
The implementation is done in the Amsterdam Density Functional
(ADF)
program package.
(back to top)
Optimized Effective Potential Method and Time-dependent Density Functional Theory
Postdoc in Erlangen with Prof. Dr. A. Görling (October 2005 - October 2006)
Conventional density functional theory (DFT) methods like those based on the
local density approximation (LDA) or generalized gradient approximations
(GGAs) employ approximate functionals for the exchange-correlation (XC)
energy that are integrals of functions of the electron density and its
gradient. GGA methods are widely and successfully employed routine methods
to investigate electronic ground states and their properties in chemistry
and solid state physics. Despite their success, these conventional DFT
methods are not accurate enough for many questions of interest.
In recent years, a new generation of DFT methods emerged that uses
functionals that not only depend on the elctron density and its gradients
but also on Kohn-Sham orbitals.
These DFT methods require the solution of the optimized effective potential
(OEP) equations which are plagued by numerical stability problems if
localized basis sets are employed. Together with Dr. A. Heßelmann
I have developed a numerically stable approach which can be used for ground
state calculations with orbital-dependent functionals [1,2]. Our method has
been implemented and tested for exact-exchange (EXX) DFT which employs the
exact exchange functional instead of approximations to it.
Time-dependent density functional theory (TD-DFT) is a very efficient and in
many cases encouragingly accurate method to describe electronically excited
states of molecules.
A fundamental problem which has not yet been solved concerns excited states
with charge-transfer (CT) character. Such excited states, which are of
importance for a great deal of applications, cannot be described by
GGA XC functionals.
A very promising avenue to solve this problem is the extension of the
EXX approach to excited states [3].
Although the theory has been worked out already in 1998, no practical
implementation for molecules exists so far.
|
EXX and localized Hartree-Fock exchange potentials for pyridine
along the carbon nitrogen ring [2].
|
|
Based on our numerically stable OEP approach I have worked out algorithms
and implemented these into the TD-DFT module of the quantum chemical program
package TURBOMOLE.
Research along these lines is further pursued in Prof. Dr. A. Görling's
group.
[1] A. Heßelmann, A. W. Götz, F. Della Salla, A. Görling,
J. Chem. Phys. 127, 054102 (2007).
[2] A. Görling, A. Ipatov, A. W. Götz,
A. Heßelmann, Z. Phys. Chem. 224, 325-342 (2010).
[3] A. Görling, Int. J. Quant. Chem. 69, 265-277 (1998).
(back to top)
Density Fitting Approaches for Efficient Density Functional Calculations
PhD thesis, Erlangen, Germany (November 2001 - August 2005)
My PhD thesis in Theoretical Chemistry was accomplished at the
University of Erlangen under the
supervision of
Prof. Dr. B. A. Heß (University of Bonn, deceased)
and later under the supervision of
Prof. Dr. A. Görling.
I have been working on the LEDO-DFT [1] formalism and its implementation
into the quantum chemical program
package TURBOMOLE.
LEDO-DFT speeds up density functional theory (DFT) calculations and thus
enables simulations of larger molecules.
LEDO is an acronym for limited expansion of diatomic overlap densities
and is a novel density fitting approach to speed up molecular DFT
calculations.
Conventional density fitting methods (also called RI methods) as
generally employed in DFT expand the complete electron density into a
fit basis which is distributed over the whole system (molecule). As a
result one- to three-center two-electron repulsion integrals (ERIs)
have to be evaluated and the computational expense is O(N3)
with respect to the size N of the system under investigation as
compared to O(N4) for conventional DFT.
In the LEDO-DFT formalism each individual diatomic overlap density is
expanded into a fit basis which is restricted to the atoms of that
overlap density. Thus, two-center matrix elements of the Coulomb- and
Exchange-Correlation contribution to the Kohn-Sham matrix can be
expressed in terms of the one-center elements using the LEDO expansion
coefficients. Consequently only one- and two-center ERIs have to be
evaluated and the computational expense of the Kohn-Sham matrix
formation in a LEDO-DFT calculation is reduced to O(N2).
|
CPU time for the SCF of linear alkanes on a single core 2.0 GHz
Intel Xeon processor.
|
|
During my PhD I have implemented the calculation of analytical gradients [2]
within the LEDO-DFT formalism [1].
Furthermore, I have worked out criteria for the optimization of auxiliary
basis sets for the LEDO expansion and optimized auxiliary orbitals which
allow for LEDO-DFT calculations with sufficient accuracy [3].
My PhD thesis can be downloaded in
pdf
format from the opus server
of the University of Erlangen.
diploma thesis, Erlangen, Germany (April 2001 - October 2001)
My diploma thesis was accomplished during the time from April to October 2001
under the supervision of
Prof. Dr. B. A. Heß
(University of Bonn, deceased) at
the University of Erlangen.
It is entitled Implementierung eines vereinfachten
Dichtefunktionalverfahrens (Implementation of a Simplified Density
Functional Formalism) and deals with the implementation of the LEDO-DFT[1]
formalism (see PhD thesis) in the SCF part of the DFT programs of the quantum
chemical program package
TURBOMOLE.
My diploma thesis (available only in German) can be download here:
ps /
pdf
[1] C. Kollmar, B. A. Hess, Molec. Phys. 100, 1945-1955 (2002).
[2] A. W. Götz, C. Kollmar, B. A. Hess, Molec. Phys. 103, 175-182 (2005).
[3] A. W. Götz, C. Kollmar, B. A. Hess, J. Comput. Chem. 26, 1242-1253 (2005).
(back to top)
Renovating a Monastery in Northern Italy
(back to top)
Work camp in Àcqui Terme, Italy (September 1995)
In September 1995, after finishing school and before university started, I
participated in a workcamp organized by
SCI (Service Civil
International).
It took place in a small village near Àcqui Terme which is close to the
town of Asti in the Piemont region in northern Italy.
The philosohpy of the workcamps organized by SCI is to bring together the
working and/or creative power of people from different countries in order to
realize a project of social welfare. In general the participants have to pay
the trip to the workcamp by themselves, but board and lodging at the working
place are for free.
In my case the project of the work camp was to renovate the run-down
accomodations of a small protestant church for future use as a cost-free
excursion center for schools or other institutions or people in need.
We have been a dozen people from Great Britain, the Netherlands, Germany and
Italy. It was hard work, but also a lot of fun.
The duration of the workcamp was two weeks, rather short. Nevertheless it was a
unique and worthy experience. If you have time, I can just recommend to
participate in a workcamp organized by SCI.
(back to top)
last modification: 2014/03/19
|