Printed temperature in constrained dynamics with i-PI

[Richa Rashmi]

Hi everyone!

I am running constrained dynamics simulations (rigid bonds/angles) in the NVT ensemble using i-PI. However, the printed temperature appears lower than expected. I computed the temperature manually from the printed kinetic energy using the correct number of degrees of freedom (accounting for the constraints), and the result seemed reasonable. 

I just wanted to confirm: is it expected that i-PI doesn't account for the reduced degrees of freedom when printing the temperature in constrained dynamics simulations?

Thanks for your help.
Best,
Richa

[Michele Ceriotti]

Hopefully @Venkat Kapil can help more, but the short answer is that the constrained dynamics code has never been tested very thoroughly, and I'm pretty confident that the constrained degrees of freedom are not taken into account. Besides, it's far from trivial to get a decent kinetic temperature estimator when using rigid bodies. What are you trying to use this for?

M

[Richa Rashmi]

Dear Prof. Ceriotti,

Thank you for your reply. I was trying to see if I could replicate the SHAKE algorithm (as implemented in LAMMPS) within i-PI for a bulk water simulation.

Best,
Richa