NWChem: Recent and Ongoing Developments.

This paper summarizes developments in the NWChem computational chemistry suite since the last major release (NWChem 7.0.0). Specifically, we focus on functionality, along with input blocks, that is accessible in the current stable release (NWChem 7.2.0) and in the "master" development branch, interfaces to quantum computing simulators, interfaces to external libraries, the NWChem github repository, and containerization of NWChem executable images. Some ongoing developments that will be available in the near future are also discussed.

Enabling Fortran Standard Parallelism in GAMESS for Accelerated Quantum Chemistry Calculations.

The performance of Fortran 2008 DO CONCURRENT (DC) relative to OpenACC and OpenMP target offloading (OTO) with different compilers is studied for the GAMESS quantum chemistry application. Specifically, DC and OTO are used to offload the Fock build, which is a computational bottleneck in most quantum chemistry codes, to GPUs. The DC Fock build performance is studied on NVIDIA A100 and V100 accelerators and compared with the OTO versions compiled by the NVIDIA HPC, IBM XL, and Cray Fortran compilers. The results show that DC can speed up the Fock build by 3.0× compared with that of the OTO model. With similar offloading efforts, DC is a compelling programming model for offloading Fortran applications to GPUs.

Application Experiences on a GPU-Accelerated Arm-based HPC Testbed.

This paper assesses and reports the experience of ten teams working to port, validate, and benchmark several High Performance Computing applications on a novel GPU-accelerated Arm testbed system. The testbed consists of eight NVIDIA Arm HPC Developer Kit systems, each one equipped with a server-class Arm CPU from Ampere Computing and two data center GPUs from NVIDIA Corp. The systems are connected together using InfiniBand interconnect. The selected applications and mini-apps are written using several programming languages and use multiple accelerator-based programming models for GPUs such as CUDA, OpenACC, and OpenMP offloading. Working on application porting requires a robust and easy-to-access programming environment, including a variety of compilers and optimized scientific libraries. The goal of this work is to evaluate platform readiness and assess the effort required from developers to deploy well-established scientific workloads on current and future generation Arm-based GPU-accelerated HPC systems. The reported case studies demonstrate that the current level of maturity and diversity of software and tools is already adequate for large-scale production deployments.

Coupled Cluster Studies of Ionization Potentials and Electron Affinities of Single-Walled Carbon Nanotubes.

In this paper, we apply equation-of-motion coupled cluster (EOM-CC) methods in the studies of the vertical ionization potentials (IPs) and electron affinities (EAs) for a series of single-walled carbon nanotubes (SWCNT). The EOM-CC formulations for IPs and EAs employing excitation manifolds spanned by single and double excitations (IP/EA-EOM-CCSD) are used to study the IPs and EAs of the SWCNTs as a function of the nanotube length. Several armchair nanotubes corresponding to C20nH20 models with n = 2-6 have been used in benchmark calculations. In agreement with previous studies, we demonstrate that the electronegativity of C20nH20 systems remains, to a large extent, independent of the nanotube length. We also compare IP/EA-EOM-CCSD results with those obtained with coupled cluster models with single and double excitations corrected by perturbative triples, CCSD(T), and density functional theory (DFT) using global and range-separated hybrid exchange-correlation functionals.

Effect of intrinsic and extrinsic factors on the simulated D-band length of type I collagen.

A signature feature of collagen is its axial periodicity visible in TEM as alternating dark and light bands. In mature, type I collagen, this repeating unit, D, is 67 nm long. This periodicity reflects an underlying packing of constituent triple-helix polypeptide monomers wherein the dark bands represent gaps between axially adjacent monomers. This organization is visible distinctly in the microfibrillar model of collagen obtained from fiber diffraction. However, to date, no atomistic simulations of this diffraction model under zero-stress conditions have reported a preservation of this structural feature. Such a demonstration is important as it provides the baseline to infer response functions of physiological stimuli. In contrast, simulations predict a considerable shrinkage of the D-band (11-19%). Here we evaluate systemically the effect of several factors on D-band shrinkage. Using force fields employed in previous studies we find that irrespective of the temperature/pressure coupling algorithms, assumed salt concentration or hydration level, and whether or not the monomers are cross-linked, the D-band shrinks considerably. This shrinkage is associated with the bending and widening of individual monomers, but employing a force field whose backbone dihedral energy landscape matches more closely with our computed CCSD(T) values produces a small D-band shrinkage of < 3%. Since this force field also performs better against other experimental data, it appears that the large shrinkage observed in earlier simulations is a force-field artifact. The residual shrinkage could be due to the absence of certain atomic-level details, such as glycosylation sites, for which we do not yet have suitable data.

Parallel scalability of Hartree-Fock calculations.

Quantum chemistry is increasingly performed using large cluster computers consisting of multiple interconnected nodes. For a fixed molecular problem, the efficiency of a calculation usually decreases as more nodes are used, due to the cost of communication between the nodes. This paper empirically investigates the parallel scalability of Hartree-Fock calculations. The construction of the Fock matrix and the density matrix calculation are analyzed separately. For the former, we use a parallelization of Fock matrix construction based on a static partitioning of work followed by a work stealing phase. For the latter, we use density matrix purification from the linear scaling methods literature, but without using sparsity. When using large numbers of nodes for moderately sized problems, density matrix computations are network-bandwidth bound, making purification methods potentially faster than eigendecomposition methods.

Reconsidering Dispersion Potentials: Reduced Cutoffs in Mesh-Based Ewald Solvers Can Be Faster Than Truncation.

Long-range dispersion interactions have a critical influence on physical quantities in simulations of inhomogeneous systems. However, the perceived computational overhead of long-range solvers has until recently discouraged their implementation in molecular dynamics packages. Here, we demonstrate that reducing the cutoff radius for local interactions in the recently introduced particle-particle particle-mesh (PPPM) method for dispersion [Isele-Holder et al., J. Chem. Phys., 2012, 137, 174107] can actually often be faster than truncating dispersion interactions. In addition, because all long-range dispersion interactions are incorporated, physical inaccuracies that arise from truncating the potential can be avoided. Simulations using PPPM or other mesh Ewald solvers for dispersion can provide results more accurately and more efficiently than simulations that truncate dispersion interactions. The use of mesh-based approaches for dispersion is now a viable alternative for all simulations containing dispersion interactions and not merely those where inhomogeneities were motivating factors for their use. We provide a set of parameters for the dispersion PPPM method using either ik or analytic differentiation that we recommend for future use and demonstrate increased simulation efficiency by using the long-range dispersion solver in a series of performance tests on massively parallel computers.

Flow-Dependent Unfolding and Refolding of an RNA by Nonequilibrium Umbrella Sampling.

Nonequilibrium experiments of single biomolecules such as force-induced unfolding reveal details about a few degrees of freedom of a complex system. Molecular dynamics simulations can provide complementary information, but exploration of the space of possible configurations is often hindered by large barriers in phase space that separate metastable regions. To solve this problem, enhanced sampling methods have been developed that divide a phase space into regions and integrate trajectory segments in each region. These methods boost the probability of passage over barriers and facilitate parallelization since integration of the trajectory segments does not require communication, aside from their initialization and termination. Here, we present a parallel version of an enhanced sampling method suitable for systems driven far from equilibrium: nonequilibrium umbrella sampling (NEUS). We apply this method to a coarse-grained model of a 262-nucleotide RNA molecule that unfolds and refolds in an explicit flow field modeled with stochastic rotation dynamics. Using NEUS, we are able to observe extremely rare unfolding events that have mean first passage times as long as 45 s (1.1 × 10(15) dynamics steps). We examine the unfolding process for a range of flow rates of the medium, and we describe two competing pathways in which different intramolecular contacts are broken.

Coupled Cluster Theory on Graphics Processing Units I. The Coupled Cluster Doubles Method.

The coupled cluster (CC) ansatz is generally recognized as providing one of the best wave function-based descriptions of electronic correlation in small- and medium-sized molecules. The fact that the CC equations with double excitations (CCD) may be expressed as a handful of dense matrix-matrix multiplications makes it an ideal method to be ported to graphics processing units (GPUs). We present our implementation of the spin-free CCD equations in which the entire iterative procedure is evaluated on the GPU. The GPU-accelerated algorithm readily achieves a factor of 4-5 speedup relative to the multithreaded CPU algorithm on same-generation hardware. The GPU-accelerated algorithm is approximately 8-12 times faster than Molpro, 17-22 times faster than NWChem, and 21-29 times faster than GAMESS for each CC iteration. Single-precision GPU-accelerated computations are also performed, leading to an additional doubling of performance. Single-precision errors in the energy are typically on the order of 10(-6) hartrees and can be improved by about an order of magnitude by performing one additional iteration in double precision.

Computational studies of the thermochemistry for conversion of glucose to levulinic acid.

The thermochemistry of the conversion of glucose to levulinic acid through fructofuranosyl intermediates is investigated using the high-level ab initio methods G4 and G4MP2. The calculated gas phase reaction enthalpies indicate that the first two steps involving water molecule elimination are highly endothermic, while the other steps, including additional water elimination and rehydration to form levulinic acid, are exothermic. The calculated gas phase free energies indicate that inclusion of entropic effects makes the dehydration steps more favorable, although the elimination of the first water is still endothermic. Elevated temperatures and aqueous reaction environments are also predicted to make the dehydration reaction steps thermodynamically more favorable. On the basis of these enthalpy and free energy calculations, the first dehydration step in conversion of glucose to levulinic acid is likely a key step in controlling the overall progress of the reaction. An assessment of density functional theories and other theoretical methods for the calculation of the dehydration and hydration reactions in the decomposition of glucose is also presented.

Active-space completely-renormalized equation-of-motion coupled-cluster formalism: Excited-state studies of green fluorescent protein, free-base porphyrin, and oligoporphyrin dimer.

The completely renormalized equation-of-motion coupled-cluster approach with singles, doubles, and noniterative triples [CR-EOMCCSD(T)] has proven to be a reliable tool in describing vertical excitation energies in small and medium size molecules. In order to reduce the high numerical cost of the genuine CR-EOMCCSD(T) method and make noniterative CR-EOMCCSD(T) approaches applicable to large molecular systems, two active-space variants of this formalism [the CR-EOMCCSd(t)-II and CR-EOMCCSd(t)-III methods], based on two different choices of the subspace of triply excited configurations employed to construct noniterative correction, are introduced. In calculations for green fluorescent protein (GFP) and free-base porphyrin, where the CR-EOMCCSD(T) results are available, we show good agreement between the active-space CR-EOMCCSD(T) (variant II) and full CR-EOMCCSD(T) excitation energies. For the oligoporphyrin dimer (P(2)TA) active-space CR-EOMCCSD(T) results provide reasonable agreement with experimentally inferred data. For all systems considered we demonstrated that the active-space CR-EOMCCSD(T) corrections lower the EOMCCSD (iterative equation-of-motion coupled-cluster method with singles and doubles) excitation energies by 0.2 and 0.3 eV, which leads to a better agreement with experiment. We also discuss the quality of basis sets used and compare EOMCC excitation energies with excitation energies obtained with other methods. In particular, we demonstrate that for GFP and FBP Sadlej's TZP and cc-pVTZ basis sets lead to a similar quality of the EOMCC results. The performance of the CR-EOMCCSD(T) implementation is discussed from the point of view of timings of iterative parts and scalability of the most expensive, N(7), part of the calculation. In the latter case the scalability across 34 008 processors is reported.

Accurate dipole polarizabilities for water clusters n=2-12 at the coupled-cluster level of theory and benchmarking of various density functionals.

The static dipole polarizabilities of water clusters (2

Parallel computation of coupled-cluster hyperpolarizabilities.

Static hyperpolarizabilities of molecules (water, acetonitrile, chloroform, and para-nitroaniline) are calculated with large basis sets using coupled-cluster response theory and compared to four common density functional theory methods. These results reveal which methods and basis sets are appropriate for nonlinear optical studies for different types of molecules and provide a means for estimating errors from the quantum chemical approximation when including vibrational contributions or solvent effects at the QM/MM level. The largest calculation reported, which was for 72 electrons in 812 functions at C(2v) symmetry, took only a few hours on 256 nodes demonstrating that even larger calculations are quite feasible using modern supercomputers.

Reappraisal of cis effect in 1,2-dihaloethenes: an improved virtual orbital multireference approach.

Computed relative stabilities for isomers of 1,2-difluoroethene and 1,2-dichloroethene isomers are compared with predictions based on chemical hardness (eta) and electrophilicity (omega) using the principles of maximum hardness and minimum electrophilicity. The chemical hardness and electrophilicity deduced either from improved virtual orbital (IVO) energies or from correlated treatments correctly predict that cis 1,2-difluoroethene and 1,2-dichloroethene are energetically more stable than the corresponding trans isomers, and the ground state energies from multireference perturbation theory with IVO orbitals agree with these predictions. However, when the same quantities are computed using Hartree-Fock orbitals, serious inconsistencies between the two approaches emerge in predicting the stability of the isomers of the 1,2-dihaloethenes. The present study clearly demonstrates that the IVO energies are appropriate for the computation of hardness related parameters, notably the chemical hardness and electrophilicity. Moreover, the IVO methods also provide smooth potential energy curves for the cis-trans isomerization of the two 1,2-dihaloethenes.

Coupled-cluster dynamic polarizabilities including triple excitations.

Dynamic polarizabilities for open- and closed-shell molecules were obtained by using coupled-cluster (CC) linear response theory with full treatment of singles, doubles, and triples (CCSDT-LR) with large basis sets utilizing the NWChem software suite. By using four approximate CC methods in conjunction with augmented cc-pVNZ basis sets, we are able to evaluate the convergence in both many-electron and one-electron spaces. For systems with primarily dynamic correlation, the results for CC3 and CCSDT are almost indistinguishable. For systems with significant static correlation, the CC3 tends to overestimate the triples contribution, while the PS(T) approximation [J. Chem. Phys. 127, 164105 (2007)] produces mixed results that are heavily dependent on the accuracies provided by noniterative approaches used to correct the equation-of-motion CCSD excitation energies. Our results for open-shell systems show that the choice of reference (restricted open-shell Hartree-Fock versus unrestricted Hartree-Fock) can have a significant impact on the accuracy of polarizabilities. A simple extrapolation based on pentuple-zeta CCSD calculations and triple-zeta CCSDT calculations reproduces experimental results with good precision in most cases.

Linear response coupled cluster singles and doubles approach with modified spectral resolution of the similarity transformed Hamiltonian.

This paper discusses practical scheme for correcting the linear response coupled cluster with singles and doubles (CCSD) equations by shifting the poles corresponding to the equation-of-motion CCSD excitation energies by adding noniterative corrections due to triples. A simple criterion is derived for the excited states to be corrected in the spectral resolution of similarity transformed Hamiltonian on the CCSD level. Benchmark calculations were performed to compare the accuracies of static and dynamic polarizabilities obtained in this way with the CC3 and CCSDT counterparts.

Dynamic polarizabilities of polyaromatic hydrocarbons using coupled-cluster linear response theory.

Coupled-cluster theory with single and double excitations is applied to the calculation of optical properties of large polyaromatic hydrocarbons. Dipole polarizabilities are reported for benzene, pyrene, and the oligoacenes sequence n=2-6. Dynamic polarizabilities were calculated on polyacences as large as pentacene for a single frequency and for benzene and pyrene at many frequencies. The basis set effect was studied for benzene using a variety of basis sets in the Pople [Theor. Chim. Acta 28, 213 (1973)] and Dunning [J. Chem. Phys. 90, 1007 (1989)] families up to aug-cc-pVQZ and the Sadlej pVTZ basis [Collect. Czech. Chem. Commun. 53, 1995 (1998)], which was used exclusively for the largest molecules. Geometries were optimized using HF, B3LYP, PBE0, and MP2 and compared to experiment to measure method dependence and the possible role of bond-length alternation. Finally, the polarizability results were compared to four common density functionals (B3LYP, BLYP, PBE0, PBE).

MPW1K Performs Much Better than B3LYP in DFT Calculations on Reactions that Proceed by Proton-Coupled Electron Transfer (PCET).

DFT calculations have been performed with the B3LYP and MPW1K functional on the hydrogen atom abstraction reactions of ethenoxyl with ethenol and of phenoxyl with both phenol and alpha-naphthol. Comparison with the results of G3 calculations shows that B3LYP seriously underestimates the barrier heights for the reaction of ethenoxyl with ethenol by both proton-coupled electron transfer (PCET) and hydrogen atom transfer (HAT) mechanisms. The MPW1K functional also underestimates the barrier heights, but by much less than B3LYP. Similarly, comparison with the results of experiments on the reaction of phenoxyl radical with alpha-naphthol indicates that the barrier height for the preferred PCET mechanism is calculated more accurately by MPW1K than by B3LYP. These findings indicate that the MPW1K functional is much better suited than B3LYP for calculations on hydrogen abstraction reactions by both HAT and PCET mechanisms.