Abstract by Jeremy Jorgensen
Physics and Astronomy
Accelerating Materials Discovery
Computational materials databases are the key to materials predictions. Calculating these databases is an expensive process that requires significant investments in computing resources. Most of the computing effort is exhausted converging two coupled, numerical integrals. We show that convergence of these integrals is an unpredictable process and that legacy techniques at improving convergence are ineffective. Advanced numerical techniques, adapted to this problem, have the potential to substantially decrease computation time.