- We create modern interfaces (e.g. web APIs) between NWChem and KinBot and ATcT, fully taking advantage of the GPU and MIC capabilities of NWChem.
- Implement methods beyond DFT to treat heterogeneous catalytic systems.
NWChem is an open source computational chemistry code. In this this project,
Random Phase Approximation (RPA)
Predictive thermodynamics of heterogeneous catalysis surfaces, in particular for metal surfaces, has been challenge to computational chemistry modelers. Traditional many-body methods used in quantum chemistry, such as CCSD(T), were designed for molecular systems and are not well suited for metals as they become singular for electronic states containing large degeneracies. DFT methods would seem to be a more natural choice for modeling these systems; however, currently available exchange correlation functionals cannot describe surface energies and molecule-surface adsorption energies simultaneously. Clearly both of these energies need to be accurately balanced in order to describe the thermodynamics of catalytic systems in which the surface reconstructs during adsorption. The problem has been that exchange correlation functionals which contain strong gradient corrections, such as BLYP, produce reasonable adsorption energies but inaccurate surface energies. However, an opposite result is seen for functionals with weaker gradient correction terms such as PBEsol, which produces good surface energies but inaccurate adsorption energies.
Developments under ECC
An approach for estimating the correlation energy of metal systems is the random phase approximation (RPA). The RPA method seems to resolve many of the thermodynamic issues of surface adsorption on metals, e.g. the adsorption of CO and benzene on transition metal surfaces. These methods, while not nearly as computationally efficient as DFT methods, are considerably less expensive than Monte-Carlo methods and should be viable with the cheminformatics studies proposed in this project as long as an efficient implementations of these methods are available for DOE HPC computers.
A scientific service that uses NWChem and chemical computational databases to make materials and chemical modeling accessible via a broad spectrum of digital communications including posts to web APIs, social networks, and traditional email.
Developments under ECC
In the ECC project we are extending Arrows to include reaction pathway optimization, surface calculations, integrate with KinBot etc., as well as develop full featured web-based 3D molecular (e.g. https://arrows.emsl.pnnl.gov/api/3dbuilder), crystal, and surface builders.
- Eric Bylaska
All electronic structure developments are merged into NWChem.
Valiev, M.; Bylaska, E. J.; Govind, N.; Kowalski, K.; Straatsma, T. P.; Van Dam, H. J.; Wang, D.; Nieplocha, J.; Apra, E.; Windus, T. L., NWChem: A comprehensive and scalable open-source solution for large scale molecular simulations. Comput. Phys. Commun. 2010, 181, 1477-1489.
Jacquelin, M.; Jong, W. D.; Bylaska, E. In Towards Highly scalable Ab Initio Molecular Dynamics (AIMD) Simulations on the Intel Knights Landing Manycore Processor, 2017 IEEE International Parallel and Distributed Processing Symposium (IPDPS), 29 May-2 June 2017; 2017; pp 234-243.
Bylaska, E. J.; Jacquelin, M.; de Jong, W. A.; Hammond, J. R.; Klemm, M. In Performance Evaluation of NWChem Ab-Initio Molecular Dynamics (AIMD) Simulations on the Intel® Xeon Phi™ Processor, High Performance Computing, Cham, 2017//; Kunkel, J. M.; Yokota, R.; Taufer, M.; Shalf, J., Eds. Springer International Publishing: Cham, 2017; pp 404-418.
Bylaska EJ, Aprà E, Kowalski K, Jacquelin M, De Jong WA, Vishnu A, Palmer B, Daily J, Straatsma TP, Hammond JR, Klemm M. 8 Transitioning NWChem to the Next Generation of Manycore Machines. Exascale Scientific Applications: Scalability and Performance Portability. 2017 Nov 13:165.