Small Energy Discontinuity on i-PI restart

[Ioannis Pagiazitis]

Hello all,

When I restart a long i-PI run using the built-in RESTART file, I notice that the total energy at the first step of the new run does not exactly match the last step before writing the restart. The difference is small but consistent.

1. Is this energy "jump" expected from the way i-PI writes and reads restart files?

2. Should I be concerned about energy conservation when continuing a trajectory this way?

Any insights would be greatly appreciated!

[Mariana Rossi]

Hi,

Could you share inputs and outputs, perhaps also with a plot, in order to characterize this energy "jump"? I don't recall ever having seen it. Maybe best as an issue in the github repo?
(https://github.com/i-pi/i-pi/issues)

All the best,

Mariana

[Ioannis Pagiazitis]

Hi Mariana (and i-PI users),

Thanks so much for offering your help. I'm attaching the following files:
1. "rpmd.xml" - my full i-PI input (10 beads, NVT/PILE/G)
2. "tcp.in" - the LAMMPS input script
3. Energy continuity plot - total conserved energy vs MD step, showing the small jump at restart.

I have confirmed that the only change between the two runs is loading the RESTART file (no other parameter tweaks).
Please let me know if you spot anything odd in the inputs or if you would like any additional outputs. 
I really appreciate your time and any insight you might have into whether this behavior is expected (or due to something I overlooked).
If you'd prefer I open a formal issue on the GitHub repo instead, just let me know and I'll do so.

Thanks again,

Ioannis

[Venkat Kapil]

Apologies. I ended up replying to you by email. Here is my response for the benefit of everyone else. 

Hi Ioannis, 

Thanks for sharing the scripts - they are very helpful. 

The change in the conserved quantity is likely due to the velocities being reinitialized in the 0-th step due to the following:

<velocities mode="thermal" units="kelvin">
5.8023e6 <!--Set Temperature-->
</velocities>


Instead of using RESTART as a checkpoint file. Try directly adding to your  bash script 
`i-pi RESTART > log.i-pi &` 
(RESTART is formatted as an XML). 

Kind regards, 

Venkat