Browsing by Author "Valeev, Edward F."

Now showing 1 - 6 of 6
  • Thumbnail Image
    CoNST: Code Generator for Sparse Tensor Networks
    Raje, Saurabh; Xu, Yufan; Rountev, Atanas; Valeev, Edward F.; Sadayappan, P. (ACM, 2024-08)
    Sparse tensor networks represent contractions over multiple sparse tensors. Tensor contractions are higher-order analogs of matrix multiplication. Tensor networks arise commonly in many domains of scientific computing and data science. Such networks are typically computed using a tree of binary contractions. Several critical inter-dependent aspects must be considered in the generation of efficient code for a contraction tree, including sparse tensor layout mode order, loop fusion to reduce intermediate tensors, and the mutual dependence of loop order, mode order, and contraction order. We propose CoNST, a novel approach that considers these factors in an integrated manner using a single formulation. Our approach creates a constraint system that encodes these decisions and their interdependence, while aiming to produce reduced-order intermediate tensors via fusion. The constraint system is solved by the Z3 SMT solver and the result is used to create the desired fused loop structure and tensor mode layouts for the entire contraction tree. This structure is lowered to the IR of the TACO compiler, which is then used to generate executable code. Our experimental evaluation demonstrates significant performance improvements over current state-of-the-art sparse tensor compiler/library alternatives.
  • Thumbnail Image
    Developing a Computational Chemistry Framework for the Exascale Era
    Richard, Ryan M.; Bertoni, Colleen; Boschen, Jeffery S.; Keipert, Kristopher; Pritchard, Benjamin; Valeev, Edward F.; Harrison, Robert J.; de Jong, Wibe A.; Windus, Theresa L. (2019-03)
    Within computational chemistry, the NWChem package has arguably been the de facto standard for running high-accuracy numerical simulations on the most powerful supercomputers. In order to better address the challenges presented by emerging exascale architectures, the decision has been made to rewrite NWChem. Design of the resulting package, NWChemEx, has been driven by exascale computing; however, significant additional design considerations have arisen from the team's involvement with the Molecular Sciences Software Institute (MolSSI). MolSSI is a National Science Foundation initiative focused on establishing coding and data standards for the computational chemistry community. As a result, NWChemEx is built upon a general computational chemistry framework called the simulation development environment (SDE) that is designed with a focus on extensibility and interoperability. The present manuscript describes the modular approach of the SDE and how it has been used to implement the self-consistent field algorithm within NWChemEx.
  • Thumbnail Image
    SparseMaps-A systematic infrastructure for reduced-scaling electronic structure methods. VI. Linear-scaling explicitly correlated N-electron valence state perturbation theory with pair natural orbital
    Guo, Yang; Pavosevic, Fabijan; Sivalingam, Kantharuban; Becker, Ute; Valeev, Edward F.; Neese, Frank (AIP Publishing, 2023-03)
    In this work, a linear scaling explicitly correlated N-electron valence state perturbation theory (NEVPT2-F12) is presented. By using the idea of a domain-based local pair natural orbital (DLPNO), computational scaling of the conventional NEVPT2-F12 is reduced to near-linear scaling. For low-lying excited states of organic molecules, the excitation energies predicted by DLPNO-NEVPT2-F12 are as accurate as the exact NEVPT2-F12 results. Some cluster models of rhodopsin are studied using the new algorithm. Our new method is able to study systems with more than 3300 basis functions and an active space containing 12 p-electrons and 12 p-orbitals. However, even larger calculations or active spaces would still be feasible.
  • Thumbnail Image
    Theoretical Prediction of Nuclear Magnetic Shielding Constants of Acetonitrile
    Adam, Ahmad Yahia (Virginia Tech, 2012-05-10)
    Gauge invariant shielding constants calculations of ?H, ?C, and ??N were calculated for acetonitrile in the gas and liquid phases. Dierent basis sets as well as dierent ab initio and DFT methods were tested to select a time-ecient level of theory with reasonable accuracy. The eect of nuclear motion on the shielding constants was also explored. To investigate solvent eects on the shielding constants of acetonitrile, dierent clusters were extracted from molecular dynamics simulations. Convergence to the experimental values varied for the dierent clusters. The geometry of the central molecule in a cluster played an important factor in reaching convergence.
  • Thumbnail Image
    Ultrahigh Vacuum Studies of the Fundamental Interactions of Chemical Warfare Agents and Their Simulants with Amorphous Silica
    Wilmsmeyer, Amanda Rose (Virginia Polytechnic Institute and State University, 2012-08-03)
    Developing a fundamental understanding of the interactions of chemical warfare agents (CWAs) with surfaces is essential for the rational design of new sorbents, sensors, and decontamination strategies. The interactions of chemical warfare agent simulants, molecules which retain many of the same chemical or physical properties of the agent without the toxic effects, with amorphous silica were conducted to investigate how small changes in chemical structure affect the overall chemistry. Experiments investigating the surface chemistry of two classes of CWAs, nerve and blister agents, were performed in ultrahigh vacuum to provide a well-characterized system in the absence of background gases. Transmission infrared spectroscopy and temperature-programmed desorption techniques were used to learn about the adsorption mechanism and to measure the activation energy for desorption for each of the simulant studied. In the organophosphate series, the simulants diisopropyl methylphosphonate (DIMP), dimethyl methylphosphonate (DMMP), trimethyl phosphate (TMP), dimethyl chlorophosphate (DMCP), and methyl dichlorophosphate (MDCP) were all observed to interact with the silica surface through the formation of a hydrogen bond between the phosphoryl oxygen of the simulant and an isolated hydroxyl group on the surface. In the limit of zero coverage, and after defect effects were excluded, the activation energies for desorption were measured to be 57.9 ± 1, 54.5 ± 0.3, 52.4 ± 0.6, 48.4 ± 1, and 43.0 ± 0.8 kJ/mol for DIMP. DMMP, TMP, DMCP, and MDCP respectively. The adsorption strength was linearly correlated to the magnitude of the frequency shift of the ν(SiO-H) mode upon simulant adsorption. The interaction strength was also linearly correlated to the calculated negative charge on the phosphoryl oxygen, which is affected by the combined inductive effects of the simulants’ different substituents. From the structure-function relationship provided by the simulant studies, the CWA, Sarin is predicted to adsorb to isolated hydroxyl groups of the silica surface via the phosphoryl oxygen with a strength of 53 kJ/mol. The interactions of two common mustard simulants, 2-chloroethyl ethyl sulfide (2-CEES) and methyl salicylate (MeS), with amorphous silica were also studied. 2-CEES was observed to adsorb to form two different types of hydrogen bonds with isolated hydroxyl groups, one via the S moiety and another via the Cl moiety. The desorption energy depends strongly on the simulant coverage, suggesting that each 2-CEES adsorbate forms two hydrogen bonds. MeS interacts with the surface via a single hydrogen bond through either its hydroxyl or carbonyl functionality. While the simulant work has allowed us to make predictions agent-surface interactions, actual experiments with the live agents need to be conducted to fully understand this chemistry. To this end, a new surface science instrument specifically designed for agent-surface experiments has been developed, constructed, and tested. The instrument, located at Edgewood Chemical Biological Center, now makes it possible to make direct comparisons between simulants and agents that will aid in choosing which simulants best model live agent chemistry for a given system. These fundamental studies will also contribute to the development of new agent detection and decontamination strategies.
  • Thumbnail Image
    Understanding Postpartum Depression from a Structural Family Theory Perspective: Examining Risk and Protective Factors
    Banker, Jamie Elizabeth (Virginia Polytechnic Institute and State University, 2010-08-24)
    This study examined pregnancy risk and protective factors for developing postpartum depression from a structural family theory lens. The purpose of this study was to (1) examine previously identified pregnancy stressors to learn which stressors put women more at risk for postpartum depression and (2) to identify possible buffers for women who are at risk for developing postpartum depression. In this paper, two analyses were proposed. Analysis I, uses a hierarchal regression analysis to examine the impact of couple related stress on postpartum depression. Analysis II uses moderated multiple regression to test factors during pregnancy which may protect at-risk women from postpartum depression symptoms. Three post-hoc exploratory analyses were conducted following the originally proposed analyses. Secondary data was used in this study. The data was collected in four large urban hospitals in Utah from 2005-2007 and included 1568 women. The results of these analyses illustrate the importance of conceptualizing postpartum depression from a family systems perceptive. Specifically, this study shows that a couple’s relationship, depending on the stress level experienced in the relationship, can be both a risk and protective factor for pregnant women.