Quantum Scenery: um novo software de química quântica semiempírico para sistemas de memória distribuída

Abstract

Efficient quantum molecular models are of highly importance when treating biomolecules and complex molecular systems. Programming efforts in optimizing less demanding computational models were lacking for a very long time. However, adopting less accurate quantum models for calculating properties is fundamental when modeling chemical and biological processes. This works focuses on introducing a new quantum chemistry software based on a distributed, SPMD paradigm for quantum properties, using the linear algebra library ScaLAPACK. Early results show that this introduced software is accurate when predicting the semiempirical total energy of molecular systems. A module for optimizing molecular geometries is also introduced based on gradients from the total energy. This module is capable of predicting energy minima geometries for smaller systems.