We demonstrate a block-tensor based implementation of coupled-cluster valence-bond singles and doubles (CCVB-SD) [Small, D. W.; Head-Gordon M. J. Chem. Phys. 2012, 137, 114103] which is a simple modification to restricted CCSD (RCCSD) that provides a qualitatively correct description of valence correlations even in strongly correlated systems. We derive the Λ-equation of CCVB-SD and the corresponding unrelaxed density matrices. The resulting production-level implementation is applied to oligoacenes, correlating up to 318 electrons in 318 orbitals. CCVB-SD shows a qualitative agreement with exact methods for short acenes and reaches the bulk limit of oligoacenes in terms of natural orbital occupation numbers, whereas RCCSD shows nonvariational behavior even for relatively short acenes. A significant reduction in polyradicaloid character is found when correlating all valence electrons instead of only the π-electrons.

Linear-scaling algorithms must be developed in order to extend the domain of applicability of electronic structure theory to molecules of any desired size. However, the increasing complexity of modern linear-scaling methods makes code development and maintenance a significant challenge. A major contributor to this difficulty is the lack of robust software abstractions for handling block-sparse tensor operations. We therefore report the development of a highly efficient symbolic block-sparse tensor library in order to provide access to high-level software constructs to treat such problems. Our implementation supports arbitrary multi-dimensional sparsity in all input and output tensors. We avoid cumbersome machine-generated code by implementing all functionality as a high-level symbolic C++ language library and demonstrate that our implementation attains very high performance for linear-scaling sparse tensor contractions.

Coupled-cluster methods provide highly accurate models of molecular structure through explicit numerical calculation of tensors representing the correlation between electrons. These calculations are dominated by a sequence of tensor contractions, motivating the development of numerical libraries for such operations. While based on matrix–matrix multiplication, these libraries are specialized to exploit symmetries in the molecular structure and in electronic interactions, and thus reduce the size of the tensor representation and the complexity of contractions. The resulting algorithms are irregular and their parallelization has been previously achieved via the use of dynamic scheduling or specialized data decompositions. We introduce our efforts to extend the Libtensor framework to work in the distributed memory environment in a scalable and energy-efficient manner. We achieve up to 240× speedup compared with the optimized shared memory implementation of Libtensor. We attain scalability to hundreds of thousands of compute cores on three distributed-memory architectures (Cray XC30 and XC40, and IBM Blue Gene/Q), and on a heterogeneous GPU-CPU system (Cray XK7). As the bottlenecks shift from being compute-bound DGEMM's to communication-bound collectives as the size of the molecular system scales, we adopt two radically different parallelization approaches for handling load-imbalance, tasking and bulk synchronous models. Nevertheless, we preserve a unified interface to both programming models to maintain the productivity of computational quantum chemists.

An efficient new molecular orbital (MO) basis algorithm is reported implementing the pair atomic resolution of the identity approximation (PARI) to evaluate the exact exchange contribution (K) to self-consistent field methods, such as hybrid and range-separated hybrid density functionals. The PARI approximation, in which atomic orbital (AO) basis function pairs are expanded using auxiliary basis functions centered only on their two respective atoms, was recently investigated by Merlot et al. [J. Comput. Chem. 2013, 34, 1486]. Our algorithm is significantly faster than quartic scaling RI-K, with an asymptotic exchange speedup for hybrid functionals of (1 + X/N), where N and X are the AO and auxiliary basis dimensions. The asymptotic speedup is 2 + 2X/N for range separated hybrids such as CAM-B3LYP, ωB97X-D, and ωB97X-V which include short- and long-range exact exchange. The observed speedup for exchange in ωB97X-V for a C68 graphene fragment in the cc-pVTZ basis is 3.4 relative to RI-K. Like conventional RI-K, our method greatly outperforms conventional integral evaluation in large basis sets; a speedup of 19 is obtained in the cc-pVQZ basis on a C54 graphene fragment. Negligible loss of accuracy relative to exact integral evaluation is demonstrated on databases of bonded and nonbonded interactions. We also demonstrate both analytically and numerically that the PARI-K approximation is variationally stable.

The computational modeling of molecules under high pressure is a growing research area that augments experimental high-pressure chemistry. Here, a new electronic structure method for modeling atoms and molecules under pressure, Gaussians On Surface Tesserae Simulate HYdrostatic Pressure (GOSTSHYP) approach, is introduced. In this method, a set of Gaussian potentials is distributed evenly on the van der Waals surface of the investigated chemical system, leading to a compression of the electron density and the atomic scaffold. Since no parameters other than pressure need to be specified, GOSTSHYP allows straightforward geometry optimizations and ab initio molecular dynamics simulations of chemical systems under pressure for nonexpert users. Calculated energies, bond lengths, and dipole moments under pressure fall within the range of established computational methods for high-pressure chemistry. A Diels-Alder reaction and the cyclotrimerization of acetylene showcase the ability of GOSTSHYP to model pressure-induced chemical reactions. The connection to mechanochemistry is pointed out.

Libtensor is a framework designed to implement the tensor contractions arising form the coupled cluster and equations of motion computational quantum chemistry equations. It has been optimized for symmetry and sparsity to be memory efficient. This allows it to run efficiently on the ubiquitous and cost-effective SMP architectures. Unfortunately, movement of memory controllers on chip has endowed these SMP systems with strong NUMA properties. Moreover, the manycore trend in processor architecture demands that the implementation be extremely thread-scalable on node. To date, Libtensor has been generally agnostic of these effects. To that end, in this paper, we explore a number of optimization techniques including a thread-friendly and NUMA-aware memory allocator and garbage collector, tuning the tensor tiling factor, and tuning the scheduling quanta. In the end, our optimizations can improve the performance of contractions implemented in Libtensor by up to 2× on representative Ivy Bridge, Nehalem, and Opteron SMPs.

The method of complex basis functions is applied to molecular resonances at correlated levels of theory. Møller-Plesset perturbation theory at second order and equation-of-motion electron attachment coupled-cluster singles and doubles (EOM-EA-CCSD) methods based on a non-Hermitian self-consistent-field reference are used to compute accurate Siegert energies for shape resonances in small molecules including N₂^-, CO^-, CO₂^-, and CH₂O^-. Analytic continuation of complex 𝜃-trajectories is used to compute Siegert energies, and the 𝜃-trajectories of energy differences are found to yield more consistent results than those of total energies. The ability of such methods to accurately compute complex potential energy surfaces is investigated, and the possibility of using EOM-EA-CCSD for Feshbach resonances is explored in the context of e-helium scattering.

Simulating solids with quantum chemistry methods and Gaussian-type orbitals (GTOs) has been gaining popularity. Nonetheless, there are few systematic studies that assess the basis set incompleteness error (BSIE) in these GTO-based simulations over a variety of solids. In this work, we report a GTO-based implementation for solids and apply it to address the basis set convergence issue. We employ a simple strategy to generate large uncontracted (unc) GTO basis sets that we call the unc-def2-GTH sets. These basis sets exhibit systematic improvement toward the basis set limit as well as good transferability based on application to a total of 43 simple semiconductors. Most notably, we found the BSIE of unc-def2-QZVP-GTH to be smaller than 0.7 mE_h per atom in total energies and 20 meV in bandgaps for all systems considered here. Using unc-def2-QZVP-GTH, we report bandgap benchmarks of a combinatorially designed meta-generalized gradient approximation (mGGA) functional, B97M-rV, and show that B97M-rV performs similarly (a root-mean-square-deviation of 1.18 eV) to other modern mGGA functionals, M06-L (1.26 eV), MN15-L (1.29 eV), and Strongly Constrained and Appropriately Normed (SCAN) (1.20 eV). This represents a clear improvement over older pure functionals such as local density approximation (1.71 eV) and Perdew-Burke-Ernzerhof (PBE) (1.49 eV), although all these mGGAs are still far from being quantitatively accurate. We also provide several cautionary notes on the use of our uncontracted bases and on future research on GTO basis set development for solids.

In this work, we developed and showcased the occ-RI-K algorithm to compute the exact exchange contribution in density functional calculations of solids near the basis set limit. Within the Gaussian planewave (GPW) density fitting, our algorithm achieves a 1-2 orders of magnitude speedup compared to conventional GPW algorithms. Since our algorithm is well suited for simulations with large basis sets, we applied it to 12 hybrid density functionals with pseudopotentials and a large uncontracted basis set to assess their performance on band gaps of 25 simple solids near the basis set limit. The largest calculation performed in this work involves 16 electrons and 350 basis functions in the unit cell utilizing a 6 × 6 × 6 k-mesh. With 20-27% exact exchange, global hybrid functionals (B3LYP, PBE0, revPBE0, B97-3, SCAN0) perform similarly with a root-mean-square deviation (RMSD) of 0.61-0.77 eV, while other global hybrid functionals such as M06-2X (2.02 eV) and MN15 (1.05 eV) show higher RMSD due to their increased fraction of exact exchange. A short-range hybrid functional, HSE achieves a similar RMSD (0.76 eV) but shows a notable underestimation of band gaps due to the complete lack of long-range exchange. We found that two combinatorially optimized range-separated hybrid functionals, ωB97X-rV (3.94 eV) and ωB97M-rV (3.40 eV), and the two other range-separated hybrid functionals, CAM-B3LYP (2.41 eV) and CAM-QTP01 (4.16 eV), significantly overestimate the band gap because of their high fraction of long-range exact exchange. Given the failure of ωB97X-rV and ωB97M-rV, we have yet to find a density functional that offers consistent performance for both molecules and solids. Our algorithm development and density functional assessment will serve as a stepping stone toward developing more accurate hybrid functionals and applying them to practical applications.

Simulations of the n = 2 absorption spectra of He_N (N = 70, 150, 231, 300) clusters are reported, with nuclear configurations sampled by path integral molecular dynamics. The electronic structure is treated by a new approach, ALMO-CIS+CT, which is a formulation of configuration interaction singles (CIS) based on absolutely localized molecular orbitals (ALMOs). The method generalizes the previously reported ALMO-CIS model [K. D. Closser et al. J. Chem. Theory Comput. 11, 5791 (2015)] to include spatially localized charge transfer (CT) effects. It is designed to recover large numbers of excited states in atomic and molecular clusters, such as the entire n = 2 Rydberg band in helium clusters. ALMO-CIS+CT is shown to recover most of the error caused by neglecting charge transfer in ALMO-CIS and has comparable accuracy to standard CIS for helium clusters. For the n = 2 band, CT stabilizes states towards the blue edge by up to 0.5 eV. ALMO-CIS+CT retains the formal cubic scaling of ALMO-CIS with respect to system size. With improvements to the implementation over that originally reported for ALMO-CIS, ALMO-CIS+CT is able to treat helium clusters with hundreds of atoms using modest computing resources. A detailed simulation of the absorption spectra associated with the 2s and 2p bands of helium clusters up to 300 atoms is reported, using path integral molecular dynamics with a spherical boundary condition to generate atomic configurations at 3 K. The main features of experimentally reported fluorescence excitation spectra for helium clusters are reproduced.