Developing Coupled Cluster Methodologies for GPUs

• Code: NWCHEM
• Science Domain: Computational Chemistry
• PI: Karol Kowalski, Pacific Northwest National Laboratory

Description

Vital in myriad interactions with proteins, glycans (macromolecules with carbohydrate functionality) encompass and protect the cell, playing an important role in the recognition of pathogens. Plant glycans such as cellulose, starch, and chitin offer the promise for creating materials with new physical and chemical properties. Researchers from Pacific Northwest National Laboratory and IBM Almaden Research Center aim to scale the NWChem application to utilize GPU accelerators to provide benchmark energies to allow for accurate parameterization of force fields for glycans as well as develop and disseminate an open-source database of accurate glycan conformational energies. New implementations of high-accuracy methods capable of taking advantage of Summit computational resources will significantly shift the system-size limit tractable by very accurate yet expensive methods accounting for the inter-electron correlation effects. The team’s results promise to allow for the computational investigation of new carbohydrate based materials from bulk polymers to nanoparticles with useful functionality, particularly in the areas of medicine, energy generation and advanced materials.