Re: Restarting problem with Langevin dynamics

From: Brian Bennion (brian_at_youkai.llnl.gov)
Date: Wed Apr 07 2004 - 13:50:00 CDT

Yes your second run started at 1.6M even though your timestep was set at
1.5M. the firsttimestep variable is just a place holder to make things
more or less convenient.

I to have had problems with restarting energies lately and haven't figured
it out. It seems to only happend with certain protein systems and not
others even though I am using the latest binary release.

In my case setting my equilibration to 298 works fine until I restart from
a binary restart file and then it zips right up to 307K in 500 steps and
then stays there.

ANY HINTS JIM?

Regards,
Brian

On Wed, 7 Apr 2004, Xin Chen wrote:

> Hi,
>
> Thank you so much. Do you mean that the second run actually restarted at
> 1.6M based on the last coordinates file even I set the first timestep at
> 1.5M? But if so, the total energy should be kept as a constant, as in the
> first run, the total energy is a constant after about 0.1M timestep. But in
> the second run, the total energy starts from another value and then converge
> to the constant value, as if it starts from time step 1. Thanks.
>
> Sincerely yours,
>
> Xin Chen
> ----- Original Message -----
> From: "Brian Bennion" <brian_at_youkai.llnl.gov>
> To: "Xin Chen" <xinchen_at_ic.sunysb.edu>
> Cc: <namd-l_at_ks.uiuc.edu>
> Sent: Wednesday, April 07, 2004 10:59 AM
> Subject: Re: namd-l: Restarting problem with Langevin dynamics
>
>
> > Hi
> >
> > Aren't you starting with a totally different restart file?
> > >From your config below, it seems that you are using the last coordinates
> > of the 1.6M run to start again at 1.5M, so the energies are going to be
> > different for the period 1.5M-1.6M in the second run. Even though you have
> > renamed the firsttimestep to 1.5M that doesn't do anything for the
> > restartfile that you set to get over written every 5000 steps.
> >
> > Regards,
> > Brian
> >
> >
> > On Wed, 7 Apr 2004, Xin Chen wrote:
> >
> > > The starting seed is same as set as 12345. The total energy is almost a
> > > constant for the first run at 1.5M to 1.6M timestep, while in the second
> run
> > > restarting at 1.5M timestep, the total energy starts from another value
> and
> > > then converge to the constant value, as if it starts from time step 1,
> > > although I set the firsttimestep 1500000. Thanks.
> > >
> > > ----- Original Message -----
> > > From: "Mike McCallum" <mmccallum_at_pacific.edu>
> > > To: "Xin Chen" <xinchen_at_ic.sunysb.edu>
> > > Sent: Tuesday, April 06, 2004 11:32 AM
> > > Subject: Re: namd-l: Restarting problem with Langevin dynamics
> > >
> > >
> > > > How different? Perhaps the starting seed is reset?
> > > >
> > > > Cheers,
> > > >
> > > > Mike
> > > > On Apr 5, 2004, at 22:05, Xin Chen wrote:
> > > >
> > > > > Hi, all
> > > > >
> > > > > I used Langevin dynamics to control the temperature in the
> simulation.
> > > > > But
> > > > > it seems that I cannot restart my simulation with it. First time, I
> did
> > > > > simulation to 1.6M timestep, and then I restart it from 1.5M
> timestep.
> > > > > The
> > > > > energy results from 1.5M to 1.6M timestep for these two simulation
> are
> > > > > different. The force field I used is AMBER. I wonder if anyone know
> > > > > why and
> > > > > how to solve this problem. Thanks. The configuration files is as
> > > > > follows
> > > > >
> > > > >
> > > > > First one
> > > > > ===================================================
> > > > > numsteps 1600000
> > > > >
> > > > > # initial config
> > > > > temperature 373.16
> > > > > seed 12345
> > > > > #rigidBonds all
> > > > > # output params
> > > > > #cwd ./out0
> > > > > outputname botulinum_water
> > > > > binaryoutput no
> > > > > outputEnergies 500
> > > > > outputMomenta 0
> > > > > outputTiming 500
> > > > > outputPressure 0
> > > > > # integrator params
> > > > > timestep 1.0
> > > > >
> > > > > # force field parameters
> > > > > exclude scaled1-4
> > > > > 1-4scaling 1.0
> > > > > switching on
> > > > > switchdist 10.0
> > > > > cutoff 12.0
> > > > > pairlistdist 14.0
> > > > > pairlistsPerCycle 10
> > > > > margin 4.0
> > > > > stepspercycle 20
> > > > >
> > > > > langevin on # langevin dynamics
> > > > > langevinFile 3bta_wat.pdb
> > > > > langevinDamping 5.0 # damping coefficient of 5/ps
> > > > > langevinTemp 373.16 # random noise at this level
> > > > >
> > > > > maximumMove 0.5
> > > > >
> > > > > wrapWater on # wrap water to central cell
> > > > > wrapAll on # wrap other molecules too
> > > > > wrapNearest off # use for non-rectangular cells
> > > > >
> > > > > # we are using amber force field
> > > > >
> > > > > amber yes
> > > > > parmfile 3bta_wat.top
> > > > > ambercoor 3bta_wat.crd
> > > > > readexclusions no
> > > > > scnb 1.0
> > > > >
> > > > > binaryrestart yes
> > > > > DCDfile bta_wat.dcd
> > > > > DCDfreq 10000
> > > > > DCDUnitCell yes
> > > > > velDCDfile bta_wat.vcd
> > > > > velDCDfreq 20000
> > > > >
> > > > > restartname bta_wat
> > > > > restartfreq 5000
> > > > > restartsave yes
> > > > >
> > > > > ===================================================
> > > > >
> > > > > Second one:
> > > > >
> > > > > Same, add one line:
> > > > >
> > > > > firsttimestep 1500000
> > > > >
> > > > >
> > > > > ===================================================
> > > > > Sincerely yours,
> > > > >
> > > > > Xin Chen
> > > > >
> > > > >
> > > > --
> > > > C. Michael McCallum
> > > > http://chem.cop.uop.edu/cmmccallum.html
> > > > Associate Professor
> > > > Department of Chemistry, UOP
> > > > mmccallum .at. pacific .dot. edu (209) 946-2636 v /
> > > > (209) 946-2607 fax
> > > >
> > > >
> > > > --
> > > > C. Michael McCallum
> > > > http://chem.cop.uop.edu/cmmccallum.html
> > > > Associate Professor
> > > > Department of Chemistry, UOP
> > > > mmccallum .at. pacific .dot. edu (209) 946-2636 v /
> > > > (209) 946-2607 fax
> > > >
> > >
> >
> > *****************************************************************
> > **Brian Bennion, Ph.D. **
> > **Computational and Systems Biology Division **
> > **Biology and Biotechnology Research Program **
> > **Lawrence Livermore National Laboratory **
> > **P.O. Box 808, L-448 bennion1_at_llnl.gov **
> > **7000 East Avenue phone: (925) 422-5722 **
> > **Livermore, CA 94550 fax: (925) 424-6605 **
> > *****************************************************************
> >
>

*****************************************************************
**Brian Bennion, Ph.D. **
**Computational and Systems Biology Division **
**Biology and Biotechnology Research Program **
**Lawrence Livermore National Laboratory **
**P.O. Box 808, L-448 bennion1_at_llnl.gov **
**7000 East Avenue phone: (925) 422-5722 **
**Livermore, CA 94550 fax: (925) 424-6605 **
*****************************************************************

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 05:18:10 CST