On Thu, Aug 06, 2015, Khabiri, Morteza wrote:
>
> I am running thermodynamic integration with amber.
Which version of Amber? Are you using sander or pmemd?
>
> when I am using this input as "lambda=1" the system runs well. But
> when I switch to "lambda = 0" the system crashes with the following
> coordinates:
>
>
> -2.2263528 0.4948044 1.3902758 -9.8532522 -20.0000000 4.7261966
> 6.1548176 20.0000000 3.7616364 4.3712168 0.3085792 -0.6998316
> 11.9788399 -20.0000000 -20.0000000 -20.0000000 20.0000000 20.0000000
> 3.6102063 2.5685121 -2.9439387 -2.4273888 -20.0000000 -20.0000000
> 20.0000000 10.4990634 20.0000000 -0.6574321 4.3513134 -1.9494082
> 15.8827264 -20.0000000 20.0000000 10.8980516 20.0000000 -20.0000000
> -0.2922468 3.1186614 -4.0249280 -9.9457371 4.3672343 -20.0000000
> 12.1206371 -14.9312673 20.0000000 3.3266217 -1.5028986 1.1370716
> -20.0000000 -12.9537867 -20.0000000 20.0000000 20.0000000 16.7487781
> 87.1893599 101.2862751 136.4590540 90.0000000 90.0000000 90.0000000
This looks(?) like the tail end of a (rather large) coordinate file. All
the values of "20.000" suggest that you must be seeing lots of "vmax errors"
in the output. Is this the case?
Are you starting the restart simulation from a previous run with some
different value of lambda? That might lead to this behavior. You need to
prepare and equilibrate each value of lambda separately, which might require
initial minimization, for example.
Have you used the "checkValidity" command in parmed to look for problems? The
other usual debugging strategy is to set ntpr=1 and nstlim to some small value
(say 100), to see if you can learn anything about what is going wrong. Do
such runs on a single cpu (assuming you are using pmemd here.)
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Aug 06 2015 - 06:00:02 PDT
