PluginPlay: Enabling exascale scientific software one module at a time.

For many computational chemistry packages, being able to efficiently and effectively scale across an exascale cluster is a heroic feat. Collective experience from the Department of Energy's Exascale Computing Project suggests that achieving exascale performance requires far more planning, design, and optimization than scaling to petascale. In many cases, entire rewrites of software are necessary to address fundamental algorithmic bottlenecks. This in turn requires a tremendous amount of resources and development time, resources that cannot reasonably be afforded by every computational science project. It thus becomes imperative that computational science transition to a more sustainable paradigm. Key to such a paradigm is modular software. While the importance of modular software is widely recognized, what is perhaps not so widely appreciated is the effort still required to leverage modular software in a sustainable manner. The present manuscript introduces PluginPlay, https://github.com/NWChemEx-Project/PluginPlay, an inversion-of-control framework designed to facilitate developing, maintaining, and sustaining modular scientific software packages. This manuscript focuses on the design aspects of PluginPlay and how they specifically influence the performance of the resulting package. Although, PluginPlay serves as the framework for the NWChemEx package, PluginPlay is not tied to NWChemEx or even computational chemistry. We thus anticipate PluginPlay to prove to be a generally useful tool for a number of computational science packages looking to transition to the exascale.

Harnessing the Power of Multi-GPU Acceleration into the Quantum Interaction Computational Kernel Program.

We report a new multi-GPU capable ab initio Hartree-Fock/density functional theory implementation integrated into the open source QUantum Interaction Computational Kernel (QUICK) program. Details on the load balancing algorithms for electron repulsion integrals and exchange correlation quadrature across multiple GPUs are described. Benchmarking studies carried out on up to four GPU nodes, each containing four NVIDIA V100-SXM2 type GPUs demonstrate that our implementation is capable of achieving excellent load balancing and high parallel efficiency. For representative medium to large size protein/organic molecular systems, the observed parallel efficiencies remained above 82% for the Kohn-Sham matrix formation and above 90% for nuclear gradient calculations. The accelerations on NVIDIA A100, P100, and K80 platforms also have realized parallel efficiencies higher than 68% in all tested cases, paving the way for large-scale ab initio electronic structure calculations with QUICK.

From NWChem to NWChemEx: Evolving with the Computational Chemistry Landscape.

Since the advent of the first computers, chemists have been at the forefront of using computers to understand and solve complex chemical problems. As the hardware and software have evolved, so have the theoretical and computational chemistry methods and algorithms. Parallel computers clearly changed the common computing paradigm in the late 1970s and 80s, and the field has again seen a paradigm shift with the advent of graphical processing units. This review explores the challenges and some of the solutions in transforming software from the terascale to the petascale and now to the upcoming exascale computers. While discussing the field in general, NWChem and its redesign, NWChemEx, will be highlighted as one of the early codesign projects to take advantage of massively parallel computers and emerging software standards to enable large scientific challenges to be tackled.

Ab Initio Multiple Spawning Method for Intersystem Crossing Dynamics: Spin-Forbidden Transitions between (3)B1 and (1)A1 States of GeH2.

Dynamics at intersystem crossings are fundamental to many processes in chemistry, physics, and biology. The ab initio multiple spawning (AIMS) method was originally developed to describe internal conversion dynamics at conical intersections where derivative coupling is responsible for nonadiabatic transitions between electronic states with the same spin multiplicity. Here, the applicability of the AIMS method is extended to intersystem crossing dynamics in which transitions between electronic states with different spin multiplicities are mediated by relativistic spin-orbit coupling. In the direct AIMS dynamics, the nuclear wave function is expanded in the basis of frozen multidimensional Gaussians propagating on the coupled electronic potential energy surfaces calculated on the fly. The AIMS method for intersystem crossing is used to describe the nonadiabatic transitions between the (3)B1 and (1)A1 states of GeH2. The potential energies and gradients were obtained at the CASSCF(6,6)/6-31G(d) level of theory. The spin-orbit coupling matrix elements were calculated with the configuration interaction method using the two-electron Breit-Pauli Hamiltonian. The excited (3)B1 state lifetime and intersystem crossing rate constants were estimated by fitting the AIMS state population with the first-order kinetics equation for a reversible unimolecular reaction. The obtained rate constants are compared with the values predicted by the statistical nonadiabatic transition state theory with transition probabilities calculated using the Landau-Zener and weak coupling formulas.

Nontotally symmetric trifurcation of an SN 2 reaction pathway.

A new type of reaction pathway which involves a nontotally symmetric trifurcation was found and investigated for a typical SN 2-type reaction, NC(-) + CH3 X → NC-CH3 + X(-) (X = F, Cl). A nontotally symmetric valley-ridge inflection (VRI) point was located along the C3 v reaction path. For X = F, the minimum energy path (MEP) starting from the transition state (TS) leads to a second-order saddle point with C3v symmetry, which connects three product minima of Cs symmetry. For X = Cl, four product minima have been observed, of which three belong to Cs symmetry and one to C3v symmetry. The branching path from the VRI point to the lower symmetry minima was determined by a linear interpolation technique. The branching mechanism is discussed based on the reaction path curvature and net atomic charges, and the possibility of a nonotally symmetric n-furcation is discussed.

Energy-Efficient Computational Chemistry: Comparison of x86 and ARM Systems.

The computational efficiency and energy-to-solution of several applications using the GAMESS quantum chemistry suite of codes is evaluated for 32-bit and 64-bit ARM-based computers, and compared to an x86 machine. The x86 system completes all benchmark computations more quickly than either ARM system and is the best choice to minimize time to solution. The ARM64 and ARM32 computational performances are similar to each other for Hartree-Fock and density functional theory energy calculations. However, for memory-intensive second-order perturbation theory energy and gradient computations the lower ARM32 read/write memory bandwidth results in computation times as much as 86% longer than on the ARM64 system. The ARM32 system is more energy efficient than the x86 and ARM64 CPUs for all benchmarked methods, while the ARM64 CPU is more energy efficient than the x86 CPU for some core counts and molecular sizes.

Dynamics simulations with spin-flip time-dependent density functional theory: photoisomerization and photocyclization mechanisms of cis-stilbene in ππ* states.

On-the-fly dynamics simulations were carried out using spin-flip time dependent density functional theory (SF-TDDFT) to examine the photoisomerization and photocyclization mechanisms of cis-stilbene following excitation to the ππ* state. A state tracking method was devised to follow the target state among nearly degenerate electronic states during the dynamics simulations. The steepest descent path from the Franck-Condon structure of cis-stilbene in the ππ* state is shown to reach the S1-minimum of 4,4-dihydrophenanthrene (DHP) via a cis-stilbene-like structure (referred to as (S1)cis-min) on a very flat region of the S1-potential energy surface. From the dynamics simulations, the branching ratio of the photoisomerization is calculated as trans:DHP = 35:13, in very good agreement with the experimental data, trans:DHP = 35:10. The discrepancy between the steepest descent pathway and the significant trans-stilbene presence in the branching ratio observed experimentally and herein computationally is clarified from an analysis of geometrical features along the reaction pathway, as well as the low barrier of 0.1 eV for the pathway from (S1)cis-min to the twisted pyramidal structure on the S1-potential energy surface. It is concluded that ππ*-excited cis-stilbene propagates primarily toward the twisted structural region due to dynamic effects, with partial branching to the DHP structural region via the flat-surface region around (S1)cis-min.