Abstract:
Theoretical predictions of noncovalent interaction energies, important e.g. in drug-design or
design of hydrogen-storage materials, belong to grand-challenges of contemporary quantum
chemistry. In this respect, quantum Monte Carlo approaches based on the fixed-node
diffusion Monte Carlo, provide a promising alternative to the commonly used
coupled-cluster methods for their benchmark accuracy, massive parallelism, and more
favorable
scaling. The talk will cover accessible introduction and recent work of the author in this
field: new simple protocols based on trial wave functions with only one determinant leading
to very accurate results, range of their applicability, computational issues, examples, and
physics behind the success.