AMBER Archive (2009)

Subject: Re: [AMBER] AMBER job restart problem

From: Thomas Cheatham (tec3_at_utah.edu)
Date: Sat Aug 15 2009 - 10:49:10 CDT


> I'm currently trying to run my systems using AMBER, but got a problem
> when I'm trying to restart the simulation from previous run. What I did
> is simply copy the restrt file with coordinations and velocities, then
> restart the job. But when I check the output data in the .mdout file and
> found the data at the same time point are obviously different from
> previous run. I'm showing the output data below,

> NSTEP = 19100 TIME(PS) = 19.100 TEMP(K) = 310.64 PRESS = 108.3
> EELEC = -283272.0143 EHBOND = 0.0000 RESTRAINT = 5825.8596

vs.

> NSTEP = 100 TIME(PS) = 19.100 TEMP(K) = 2577.07 PRESS = 3038.9
> EELEC = -268207.1839 EHBOND = 0.0000 RESTRAINT = 595262.3848
...
> ntp=2,pres0=1.0,taup=2.5,
> ntr=1,

Couple of things to notice; temperature jumps, restraint energy is
extremely high and you are running constant (anisotropic) pressure
control. This suggests that the reference coordinates have changed and
this is an artifact of constant pressure simulation. What happens is that
during constant pressure, the positions of the restraint coordinates are
also scaled so that after the run is over, the restraint coordinates not
longer match the refc coordinates. If you restarted with the original
refc restraint coordinates, you get the behavior shown. Use the restrt as
the refc coordinates and it should work (although the restraint energy
will still be different).

> GRP1
> 100.0
> ATOM 1 13609

This restraint energy is extremely high; you could probably reduce this
value to 20.0 or even 5.0.

See also:

http://archive.ambermd.org/200807/0055.html

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber