rpmd_nonlinear operator

[zhangcu...@gmail.com]

Dear All

    I am not very familiar with RPMD for calculating correlation function. According to some papers, nonlinear operator problemis one of limitaiton of RPMD, and it seems no good way to estimate the error. So, I am wondering, can we just say despite not very accurate, RPMD would still do better job for nonlinear operator than classical MD. Therefore, we can at least get closer to real correlation functions?

Thanks.

[Michele Ceriotti]

Hi, 

    There is no one-liner answer. Depends on many many aspects. For many mildly non-linear operators (e.g. dipole moment) TRPMD should be as good as it gets without getting into more sophisticated and expensive methods. If you have a strongly non-linear operator, it might be worse than classical MD. There is a considerable body of literature, so it's best that you do your reading and see for yourself for the specific problem you're interested in. I recommend you also check out how to use the imaginary time correlation function to assess the accuracy of the real-time approximation of it.

Cheers

[Mariana Rossi]

Hello,

It would help to know what sort of non-linear operator you are dealing with. As Michele says, looking at the imaginary time correlations to assess the quality of the real-time ones are useful, but there again sometimes it is hard to say what is a big difference and what is a small difference. 

It is certainly not true that RPMD would always give a better answer than classical MD.

All the best,

Mariana