Re: Replica exchange and colvars - Missing Stage from restart

From: Jérôme Hénin (jerome.henin_at_ibpc.fr)
Date: Sun Sep 15 2013 - 15:41:54 CDT

Hi Andrei,

This is a bug that was reported and fixed a while ago:
http://www.ks.uiuc.edu/Research/namd/mailing_list/namd-l.2011-2012/3431.html
Unfortunately, that was shortly after NAMD 2.9 came out, and it is still the current version, so the "official" binaries still have the bug.

If you can recompile the binary you use, you can fix this easily, either by applying the patch I am attaching, or by updating the colvars module altogether from the repository: https://github.com/colvars/colvars

If you cannot recompile, there should be a way to work around this by pretending you are running a moving restraint calculation:
(in the harmonic block)
targetCenters <the same as centers>
targetNumSteps 1
targetNumStages 0

I can't try it right away, but either that or a variant of it should rid you of this annoying error.

Let me know if you need more help.
Jerome

----- Original Message -----
>
> Hello,
>
> I am running a replica exchange simulation with a harmonic restraint
> and everything was running very smoothly for a few nanoseconds, but
> then the simulation stopped with the error:
>
> colvars:
> ----------------------------------------------------------------------
> colvars: Collective variables biases initialized, 1 in total.
> colvars:
> ----------------------------------------------------------------------
> colvars: Restarting from file
> "output/0/1ign_silcs.job4.0.colvars.state".
> colvars: Restarting collective variable "COMDistance" from value:
> 0.334768
> colvars: Restarting harmonic bias "harmonic1".
> colvars: Reading current stage from the restart.
> colvars: # stage = 0 [default]
> colvars: Error: current stage is missing from the restart.
> colvars: If this error message is unclear, try recompiling with
> -DCOLVARS_DEBUG.
> FATAL ERROR: Error in the collective variables module: exiting.
>
>
>
> My colvars.conf is like this:
>
> colvarsTrajFrequency 100
>
> colvar {
> name COMDistance
> distance {
> group1 {
> atomNumbers 4 11 30 50 57 67 86 96 103 123 142 157 171 178 202
> 217 224 241 260 277
> }
> group2 {
> dummyAtom (-0.5685726404190063, -0.11478058248758316,
> -0.2689979374408722)
> }
> }
> }
>
>
> harmonic {
> colvars COMDistance
> centers 0
> forceConstant 1.0
> }
>
>
>
> I looked in the restart files of the simulation and for some replicas
> there is the stage variable.
> I also restarted the simulation with the same results.
> What could be the cause of this?
>
> I hope I gave all the necessary configs.
>
> Andrei
>
>


This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:23:44 CST