In this talk I will discuss our efforts to develop robust and accurate many-body electronic structure methods that can withstand the challenges of modern and future classical computing platforms characterized by massive instruction and data parallelism, heterogeneity, and power constraints. Essential to further progress is the emerging many-body methdology with reduced computational complexity. At the core of such methods is block-sparse tensor algebra with irregular, data-dependent computation and data flow. I will highlight a performance-portable implementation of block-sparse tensor algebra utilizing emerging flow-based programming models.

Toward reduced scaling many-body electronic structure at exascale

255th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.