Highly scalable molecular scale software development for environmental sciences
EMSL Project ID
35400
Abstract
Understanding the complex chemistry and physical properties of systems ranging from small gas-phase clusters to the solution phase is one of the major challenges for environmental sciences. Many previous efforts to examine the dynamical properties of these systems have involved the use of classical, empirical or semi-empirical potentials for the molecular interactions which have complications due to the fitting of the potentials for bulk properties as well as not being suitable for bond making and breaking processes. Ab initio methods combined with statistical methods, while quite expensive, offer the promise of providing a general framework for examining these systems. In this proposal, development and testing of highly scalable methods will be accomplished using two of the most generally available and used computational chemistry codes - NWChem and GAMESS. In particular, this research will focus on the development of new algorithms for Dynamical Nucleation Theory Monte Carlo (DNTMC), molecular dynamics using the fragment molecular orbital (FMO) method, and a new effective fragment method for long range effects in solvents. These methods will impact various environmental sciences including the formation dynamics of aerosol clusters in the atmosphere, the properties of molecular reactions and solution structure of ions (including actinides and heavy elements) in water and other solutions including ionic liquids used in "Green Chemistry". Multi-level parallelism and automatic dynamic load balancing as well as on the fly surface fitting and component technology will be used to enable a high level of scalability in these algorithms. The computational resources at the MSCF are ideal for this development. The architecture is similar to those which are typically available as a group level resource, but are available at a much larger scale than can be afforded by any researcher. This provides a useful migration path from development of initial software on local resources to large scale development and testing required for the actual production of results that will impact the science. In addition, the staff expertise required for trouble shooting issues is available at the MSCF.
Project Details
Project type
Capability Research
Start Date
2009-10-14
End Date
2012-09-30
Status
Closed
Released Data Link
Team
Principal Investigator
Team Members
Related Publications
Atoms, N. D.; Gordon, M. S.; Windus, T. L. Abstracts of Papers, 242nd ACS National Meeting, Denver, CO, August 28 - September 1, 2011, PHYS 209