Skip to main content

Development of accurate force fields for aqueous biological environments using a massively parallel multiscale approach


EMSL Project ID
30794

Abstract

The ability to calculate accurate energies and properties of large molecules with massively-parallel computers presents an extraordinary opportunity to develop new models for bonded and non-bonded interactions used in classical force fields by directly calculating such parameters from first principles electronic structure calculations. Since those force fields usually contain a large set of parameters, especially those that describe polarization, it becomes increasingly difficult to determine these parameters in a rigorous way using only experimental data. In contrast, quantum chemical calculations accurately determine force constants, charge distributions, and polarizabilities for the relevant moieties with well-known accuracy. The coupled-cluster method is universally accepted as the most accurate method for molecules near their equilibrium geometry, and the functionality to calculate all relevant properties for determining force fields is available within NWChem. Using NWChem on the new EMSL supercomputer, polarizable force field parameters will be determined using first principles for use in both classical and nuclear quantum statistical mechanical simulations. The use of those state-of-the-art quantum chemical methods for force field development will increase the accuracy and generality of existing models and contribute significantly to the development of a hierarchy of increasingly accurate models. The emergence of a Jacobs ladder of models for molecular simulations, in analogy to that in density functional theory [J. P. Perdew and K. Schmidt, in "Density Functional Theory and Its Application to Materials, edited by V. Van Doren, C. Van Alsenoy, and P. Geerlings (AIP, Melville, NY, 2001)], promises to make significant contributions to molecular simulations relevant to all four of EMSLs science themes. The open (non-proprietary) nature of this project ensures that it will have maximum impact on the user community for the respective models developed.

Project Details

Project type
Capability Research
Start Date
2008-10-11
End Date
2011-09-30
Status
Closed

Team

Principal Investigator

Jeffrey Hammond
Institution
Intel Corporation

Team Members

Daniel Lambrecht
Institution
University of Pittsburgh

Soohaeng Yoo
Institution
Pacific Northwest National Laboratory

Jay Bardhan
Institution
Environmental Molecular Sciences Laboratory

William Farone
Institution
Applied Power Concepts, Inc.

Larkin Scott
Institution
University of Chicago

Benoît Roux
Institution
University of Chicago

Karl Freed
Institution
University of Chicago

Jasper Werhahn
Institution
Technische Universität München

Karol Kowalski
Institution
Pacific Northwest National Laboratory

Sotiris Xantheas
Institution
Pacific Northwest National Laboratory

Marat Valiev
Institution
Environmental Molecular Sciences Laboratory