Re: Problems with temperature control.

From: Brian Radak (brian.radak_at_gmail.com)
Date: Mon Jul 30 2018 - 08:41:29 CDT

Note that the Langevin and reassign thermostats are completely decoupled
code paths. My guess is that the reassignIncr is too large compared to the
relaxation rate dictated by Langevin damping and the two are redundant -
personally I would only use one or the other (you can also script changes
to langevinTemp). You could also try using a smaller damping coefficient,
but this really isn't any different from not using Langevin at all.

On Mon, Jul 30, 2018 at 6:21 AM, Josep Ivan Balaguer Molins <
jobamo1_at_etsid.upv.es> wrote:

> Hi everyone!
>
> I have a system I've minimized and during minimization, it's been heat up
> to 600K using this code lines:
>
> temperature 0
>
> reassignFreq 1000
> reassignIncr 10
> reassignHold 600 # Final temperature
>
> langevin on ;# do langevin dynamics
> langevinDamping 2 ;# damping coefficient (gamma) of 1/ps
> langevinTemp 600 # Final temperature
> langevinHydrogen no ;# don't couple langevin bath to hydrogens
>
> langevinPiston on
> langevinPistonTarget 1.01325 ;# in bar -> 1 atm
> langevinPistonPeriod 200
> langevinPistonDecay 100
> langevinPistonTemp 600 # Final temperature
>
> It does work, I can see the temperature converging to 600K in VMD.
>
> After this, I perform an Isothermal simulation to see if the systems keeps
> the stability and works too:
>
> binvelocities testD1.vel # Use velocities from minimization output.
>
> langevin on ;# do langevin dynamics
> langevinDamping 2 ;# damping coefficient (gamma) of 1/ps
>
> langevinTemp 600 # I keep the 600K
>
> langevinHydrogen no ;# don't couple langevin bath to hydrogens
>
> useGroupPressure yes ;# needed for rigidBonds
> useFlexibleCell no
> useConstantArea no
>
> langevinPiston on
> langevinPistonTarget 1.01325 ;# in bar -> 1 atm
> langevinPistonPeriod 200
> langevinPistonDecay 100
> langevinPistonTemp 600 # I keep the 600K
>
> extendedSystem testD1.xsc
>
> run 400000
>
> Seeing the result in VMD, the code works well and the simulation has no
> errors, the temperature is stable at 600K.
>
> Finally to perform the final simulation, I have to cool the system down to
> 300K. So I use the output files and with Langevin I am able to cool down
> the sistem without problems to 300K. But seeing in VMD the temperature
> graph, the temperature goes down so fast. In 5000 integration steps goes
> from 600K to 300K. So I need to change the temperature graph. To do so, I
> use this code:
>
> binvelocities testD2.vel
>
> reassignFreq 100
>
> run 1000
>
> for { set TEMP 550 } { $TEMP >= 300 } { incr TEMP -10 } {
> reassignTemp $TEMP
>
> run 100
> }
>
> And I get the following error: FATAL ERROR: Bad global bond count! (14395
> vs 14400)
>
> How is that possible taking into account that before everything worked?
>
> Changing cutoff and pairlistdist parameters, I got it down to 14397
> instead of 14395. But I can get any better. And I don't understand why is
> this happening.
>
> Thank you very much.
>
>
>
>
>
>
>
>

This archive was generated by hypermail 2.1.6 : Wed Dec 11 2019 - 23:20:07 CST