If the intention is to use variance minimisation and then VMC as an initial guide to performing more accurate DMC calculations then the situation is not so clear cut. Only the variance of the expectation value of an operator is dependent on the quality of the guiding wavefunction, not the expectation value itself. It is therefore not necessary to perform the optimisation to the same level of accuracy as for the VMC calculation. Indeed, the optimisation process can be more computationally expensive than the DMC run itself. This balance in effort spent on optimising the initial trial/guiding wavefunction versus time spent performing actual calculations will be returned to in chapter , where DMC calculations are performed on a series of systems describing excited states without the need for re-optimising the wavefunction for each system.

Tue Nov 19 17:11:34 GMT 1996