Utilizing high performance computing for chemistry: parallel computational chemistry

Jong, Wibe A. de; Bylaska, Eric J.; Govind, Niranjan; Janssen, Curtis L.; Kowalski, Karol; Müller, Thomas; Nielsen, Ida M. B.; Dam, Hubertus J. J. van; Veryazov, Valera; Lindh, Roland

doi:10.1039/c002859b

Cited by 87 publications

(78 citation statements)

References 240 publications

(262 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We shall use the shorthand 5) and, o, p, and q to denote the whole N c -tuples, as usual. Now, in order to obtain the Lagrange multipliers λ J , we just need to solve…”

Section: Calculation Of the Lagrange Multipliersmentioning

confidence: 99%

“…The atoms are ordered in the same way, as in fig. 3.1, and the trick to generate a banded matrix R with minimal bandwidth is to alternatively index bond length constraints with odd numbers, 5) and bond angle constraints with even ones, 6) where the regular pattern involving the atom indicies that participate of the same constraints has allowed again to use a lighter notation. The constraints equations in this case are…”

Section: Open Single-branch Chain With Constrained Bond Lengths and mentioning

confidence: 99%

“…In addition to this, since biologically interesting molecules (like proteins [1] and DNA [2]) consist of thousands of atoms, their trajectories in configuration space are esentially chaotic, and therefore reliable quantities can be obtained from the simulation only after statistical analysis [3]. In order to cope with these two requirements, which force the computation of a large number of dynamical steps if predictions want to be made, great efforts are being done both in hardware [4,5] and in software [6,7] solutions. In fact, only in very recent times, simulations for interesting systems of hundreds of thousand of atoms in the millisecond scale are starting to become affordable, being still, as we mentioned, the main limitation of these computational techniques the large difference between the elemental time step used to integrate the equations of motion and the total time span needed to obtain useful information.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Exact and efficient calculation of lagrange multipliers in biological polymers with constrained bond lengths and bond angles: Proteins and nucleic acids as example cases

2011

View full text Add to dashboard Cite

In order to accelerate molecular dynamics simulations, it is very common to impose holonomic constraints on their hardest degrees of freedom. In this way, the time step used to integrate the equations of motion can be increased, thus allowing, in principle, to reach longer total simulation times. The imposition of such constraints results in an aditional set of N c equations (the equations of constraint) and unknowns (their associated Lagrange multipliers), that must be solved in one way or another at each time step of the dynamics. In this work it is shown that, due to the essentially linear structure of typical biological polymers, such as nucleic acids or proteins, the algebraic equations that need to be solved involve a matrix which is banded if the constraints are indexed in a clever way. This allows to obtain the Lagrange multipliers through a non-iterative procedure, which can be considered exact up to machine precision, and which takes O(N c ) operations, instead of the usual O(N 3 c ) for generic molecular systems. We develop the formalism, and describe the appropriate indexing for a number of model molecules and also for alkanes, proteins and DNA. Finally, we provide a numerical example of * Email: garcia.risueno@gmail.com the technique in a series of polyalanine peptides of different lengths using the AMBER molecular dynamics package.

show abstract

“…We shall use the shorthand 5) and, o, p, and q to denote the whole N c -tuples, as usual. Now, in order to obtain the Lagrange multipliers λ J , we just need to solve…”

Section: Calculation Of the Lagrange Multipliersmentioning

confidence: 99%

Section: Open Single-branch Chain With Constrained Bond Lengths and mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Exact and efficient calculation of lagrange multipliers in biological polymers with constrained bond lengths and bond angles: Proteins and nucleic acids as example cases

2011

View full text Add to dashboard Cite

show abstract

“…Even though Eq. (19) suggests that our parallel in time algorithms should not be expected to greatly reduce the cost of a MD simulation with these simple two-and three-body potentials, respectable speedups of 5.2 (VP, t = 5.0, M = 96, smaller time steps will be needed if H atoms are included in the simulation) and 2.7 (VP, t = 40.0, M = 96) were obtained by parallelizing over time alone. We note that this speedup was obtained without using a sophisticated coarse model preconditioner.…”

Section: A Stillinger-weber 1000 Si Atom MD Simulationsmentioning

confidence: 99%

“…A key to these remarkable improvements has been the development of advanced a) Electronic mail: Eric.Bylaska@pnnl.gov b) Electronic mail: weare@uchicago.edu c) Electronic mail: jweare@ucsd.edu parallel programs for MD 6,7,17 and AIMD. 18,19 These programs, which efficiently distribute the work of computing the atomic forces, have been very successful in simulating large numbers of atoms, particularly for MD. Recently, there has also been significant progress in extending the time scales of MD simulations.…”

Section: Introductionmentioning

confidence: 99%

Extending molecular simulation time scales: Parallel in time integrations for high-level quantum chemistry and complex force representations

Bylaska

Weare

2013

The Journal of Chemical Physics

View full text Add to dashboard Cite

Structure and dynamics of the hydration shells of the ion from ab initio molecular dynamics and combined ab initio and classical molecular dynamics simulations The Journal of Chemical Physics 132, 194502 (2010) Parallel in time simulation algorithms are presented and applied to conventional molecular dynamics (MD) and ab initio molecular dynamics (AIMD) models of realistic complexity. Assuming that a forward time integrator, f (e.g., Verlet algorithm), is available to propagate the system from time t i (trajectory positions and velocities, the dynamics problem spanning an interval from t 0 . . . t M can be transformed into a root finding problem,for the trajectory variables. The root finding problem is solved using a variety of root finding techniques, including quasi-Newton and preconditioned quasi-Newton schemes that are all unconditionally convergent. The algorithms are parallelized by assigning a processor to each time-step entry in the columns of F(X). The relation of this approach to other recently proposed parallel in time methods is discussed, and the effectiveness of various approaches to solving the root finding problem is tested. We demonstrate that more efficient dynamical models based on simplified interactions or coarsening time-steps provide preconditioners for the root finding problem. However, for MD and AIMD simulations, such preconditioners are not required to obtain reasonable convergence and their cost must be considered in the performance of the algorithm. The parallel in time algorithms developed are tested by applying them to MD and AIMD simulations of size and complexity similar to those encountered in present day applications. These include a 1000 Si atom MD simulation using Stillinger-Weber potentials, and a HCl + 4H 2 O AIMD simulation at the MP2 level. The maximum speedup ( serial execution time parallel execution time ) obtained by parallelizing the Stillinger-Weber MD simulation was nearly 3.0. For the AIMD MP2 simulations, the algorithms achieved speedups of up to 14.3. The parallel in time algorithms can be implemented in a distributed computing environment using very slow transmission control protocol/Internet protocol networks. Scripts written in Python that make calls to a precompiled quantum chemistry package (NWChem) are demonstrated to provide an actual speedup of 8.2 for a 2.5 ps AIMD simulation of HCl + 4H 2 O at the MP2/6-31G* level. Implemented in this way these algorithms can be used for long time high-level AIMD simulations at a modest cost using machines connected by very slow networks such as WiFi, or in different time zones connected by the Internet. The algorithms can also be used with programs that are already parallel. Using these algorithms, we are able to reduce the cost of a MP2/6-311++G(2d,2p) simulation that had reached its maximum possible speedup in the parallelization of the electronic structure calculation from 32 s/time step to 6.9 s/time step.

show abstract

Perturbative Coupled‐Cluster Methods on Graphics Processing Units: Single‐ and Multi‐Reference Formulations

Ma¹,

Bhaskaran‐Nair²,

Villa³

et al. 2016

Electronic Structure Calculations on Graphics Processing Units

View full text Add to dashboard Cite

Utilizing high performance computing for chemistry: parallel computational chemistry

Cited by 87 publications

References 240 publications

Exact and efficient calculation of lagrange multipliers in biological polymers with constrained bond lengths and bond angles: Proteins and nucleic acids as example cases

Exact and efficient calculation of lagrange multipliers in biological polymers with constrained bond lengths and bond angles: Proteins and nucleic acids as example cases

Extending molecular simulation time scales: Parallel in time integrations for high-level quantum chemistry and complex force representations

Perturbative Coupled‐Cluster Methods on Graphics Processing Units: Single‐ and Multi‐Reference Formulations

Contact Info

Product

Resources

About