Skip to first unread message
Chris
Apr 10, 2018, 1:37:25 PM4/10/18
to ipi-users
Hello,
I am relatively new to running i-PI, so I apologize if this is a trivial question. I am trying to run an i-PI simulation that uses the restart file from a previous i-PI run. The input files for both are identical, except that the second input.xml file (input2.xml) reads in the .restart file from the previous run's output. The issue I have is that, when looking at the output, the conserved quantity makes a significant jump from the end of the first run (2.01874173e+01) to the beginning of the second run (4.02728518e+01). This is a significantly larger increase of the conserved quantity than was exhibited in the first run's output (changes from step-to-step were on the order of 1e-3), so I feel like it is not an expected continuation from the first run. Is this an expected jump, or is there a problem in my input files somewhere?
Thanks,
Chris
I am relatively new to running i-PI, so I apologize if this is a trivial question. I am trying to run an i-PI simulation that uses the restart file from a previous i-PI run. The input files for both are identical, except that the second input.xml file (input2.xml) reads in the .restart file from the previous run's output. The issue I have is that, when looking at the output, the conserved quantity makes a significant jump from the end of the first run (2.01874173e+01) to the beginning of the second run (4.02728518e+01). This is a significantly larger increase of the conserved quantity than was exhibited in the first run's output (changes from step-to-step were on the order of 1e-3), so I feel like it is not an expected continuation from the first run. Is this an expected jump, or is there a problem in my input files somewhere?
Thanks,
Chris
venkat kapil
Apr 10, 2018, 2:05:28 PM4/10/18
to ipi-users
Hello Chris,
So the thing is that initialize belongs to the system class so it performs initialization ONLY for the system which doesn't not include the initialization for the thermostat. Because of that you see a discrepancy. If you do an NVE simulation you will not see any jumps. But this problem can be easily avoided by simply using the restart as an i-pi input. So you can run it like:
So the thing is that initialize belongs to the system class so it performs initialization ONLY for the system which doesn't not include the initialization for the thermostat. Because of that you see a discrepancy. If you do an NVE simulation you will not see any jumps. But this problem can be easily avoided by simply using the restart as an i-pi input. So you can run it like:
$PATH_TO_IPI ../old_run/qtip4p.restart
Chris
Apr 10, 2018, 2:30:53 PM4/10/18
to ipi-users
I was able to successfully restart the run with the information you provided. Thank you for your prompt and informative reply.
Michele Ceriotti
Apr 10, 2018, 5:03:59 PM4/10/18
to ipi-users
Alternatively, you can specify the conserved quantity restart in the appropriate field - can't remember off the top of my head, probably <thermostat>.
In any case, this is purely an "aestetic" problem, the trajectory is perfectly fine you just have a jump because you lose the counter of the accumulated energy balance between thermostat and system.
Reply all
Reply to author
Forward
0 new messages