How are NVE simulations done in PIMD?

[Shubhang Goswami]

Hello all,

I am new to PIMD and am curious how the NVE microcanoncial simulation is done for path integral molecular dynamics? I looked through the paper in the Journal of Chemical Physics 133, 124104 (2010) listed in the documentation that talks about thermostats. But I still do not understand whats going on with the path integral when the temperature is constantly changing in a microcanonical ensemble (beta changes every step hence the path length).

 

How does Ipi do microcanonical runs in its NVE mode and take care of this issue?

Sincerely,

Shubhang

[Michele Ceriotti]

The path integral formalism is inherently canonical and samples thermodynamic averages. 

In more general term "beta changes every step" does not make sense, because - even in a NVT 

simulation - the kinetic energy does not define the temperature *instantaneously* but only 

when averaged. 

The constant-energy version of PIMD (known as RPMD) is used as an approximation for quantum 

time correlation functions. If you want a primer about PIMD and RPMD you can follow the 

online course at https://tinyurl.com/pimd-mooc, otherwise http://doi.org/10.1146/annurev-physchem-040412-110122

is a good review on RPMD, and http://doi.org/10.1063/1.4901214 a good comparison of 

different approximate quantum dynamics methods.

Michele