imaginary-time PIMD

[Dongdong Kang]

Dear all,

I am going to use i-pi to do the standard imaginary-time PIMD simulations, not CMD and RPMD. Is there any example to show how to do that? 

Thanks

Dongdong

[Michele Ceriotti]

Ehm.... 80% of the examples are for imaginary time PIMD.

examples/pes/qtip4pf/h2o-piglet 

is a good starting point, check also the readme in 

example/pes 

for instructions.

[Dongdong Kang]

Hi Michele,

Thank you for your fast reply. It seems that the only difference of input files between RPMD and imaginary time PIMD is the thermostat mode. "nm_gle" is for imaginary time PIMD and "pile_g" is for RPMD. Am i right?

Thanks

Dongdong

[venkat kapil]

The "nm_gle" thermostat will give you an approximate imaginary time simulation within the PIGLET scheme [10.1103/PhysRevLett.109.100604] .  A "pile_g" thermostat (with a reasonable number of repliacs) will give you a non approximate imaginary time simulation and in addition a possibility of obtaining approximate quantum dynamics within the TRPMD approximation [10.1063/1.4883861]. The PIGLET scheme (that uses nm_gle) would be computationally cheaper but approximate.

[Michele Ceriotti]

I think it'd be good if you familiarise yourself with the underlying techniques. These two reviews are good starting points

http://doi.org/10.1146/annurev-physchem-040412-110122

http://doi.org/10.1038/s41570-017-0109