Abstract
Quantum Monte Carlo methods have received considerable attention over the last
decades due to the great promise they have for the direct solution to the
many-body Schrodinger equation for electronic systems. Thanks to a low scaling
with number of particles, they present one of the best alternatives in the
accurate study of large systems and solid state calculations. In spite of such
promise, the method has not become popular in the quantum chemistry community,
mainly due to the lack of control over the fixed-node error which can be large
in many cases. In this article we present the application of large
multi-determinant expansions in quantum Monte Carlo, studying its performance
with first row dimers and the 55 molecules of the G1 test set. We demonstrate
the potential of the wave-function to systematically reduce the fixed-node
error in the calculations, achieving chemical accuracy in almost all cases
studied. When compared to traditional methods in quantum chemistry, the results
show a marked improvement over most methods including MP2, CCSD(T) and DFT with
various functionals; in fact the only method able to produce better results is
the explicitly-correlated CCSD(T) method with a large basis set. With recent
developments in trial wave functions and algorithmic improvements in Quantum
Monte Carlo, we are quickly approaching a time where the method can become the
standard in the study of large molecular systems and solids.