Re: PME fails no matter what

From: David Hardy (dhardy_at_ks.uiuc.edu)
Date: Tue Jul 29 2008 - 12:39:04 CDT

Hi Roman,

This is an issue with the spatial decomposition of your periodic cell
into NAMD "patches" NOT pme. Your use of PMEGridSize[XYZ] = 32 should
be adequate given your basis vector lengths. Maybe try adjusting your
margin. I notice that you didn't post your choices of cutoff, margin,
and hGroupCutoff, nor did you show the output of NAMD's patch
decomposition. These are the things at issue as your periodic cell
shrinks during NPT simulation, not your choice of pme parameters.

Best regards,
Dave

On Tue, Jul 29, 2008 at 10:00:11AM -0700, Joshua Adelman wrote:
> To follow up on this, Peter helped me with a similar problem a number
> of months ago. The complete thread is found here:
> http://www.ks.uiuc.edu/Research/namd/mailing_list/namd-l/7256.html
>
> Basically the solution was check the xst file and then take steps to
> make sure there wasn't a large change in the cell size. Some things
> you might want to try:
> - Slowly heating the system during equilibration
> - Equilibrating the solvent while restraining the protein
> - Start with no rigid bonds and a small time step (for a particularly
> nasty system I started with 0.5fs, brought it to 1 fs and then once I
> was confident that things were ok, then turned rigidBonds on and used
> 2.0 fs)
>
> Hope that helps.
>
> Josh
>
>
> On Jul 29, 2008, at 9:38 AM, Peter Freddolino wrote:
>
> >Hi Roman,
> >please note, since it's important to understand what the actual
> >problem here is, that this is not a problem with PME but rather one
> >with your periodic cell (and, most notably, the fact that it is
> >shrinking significantly early on in the simulation). You may want to
> >run writing an xst file every timestep and then check whether your
> >cell shrinks rapidly after minimization -- it probably does. If this
> >is the case, you will need to either follow the advice of restarting
> >from a recent checkpoint, or reduce your initial periodic cell size
> >to be closer to what the actual, equlibrated cell size will be. How
> >did you solvate your input system?
> >Best,
> >Peter
> >
> >Roman Petrenko wrote:
> >>Dear all,
> >>my system size is (min and max)
> >>{-16.3659992218 -18.3840007782 -13.1579999924} {17.0270004272
> >>17.5310001373 7.91699981689}
> >>
> >>the center is
> >>0.253832936287 -0.271096855402 -2.62966871262
> >>
> >>additionally i add 2.5A in each direction for the basis vectors
> >>TO_NAMD:cellBasisVector1 35.892999649 0. 0.
> >>TO_NAMD:cellBasisVector2 0. 38.4150009155 0.
> >>TO_NAMD:cellBasisVector3 0. 0 23.5749998093
> >>TO_NAMD:cellOrigin 0.253832936287 -0.271096855402
> >>-2.62966871262
> >>TO_NAMD:wrapAll on
> >>
> >>now minimization is ok for 100 steps, but the main run (it is NPT
> >>simulation) fails no matter what parameter i put for PMEGridSize.
> >>i've
> >>tried starting from 32 to 90.
> >>
> >>TO_NAMD:PME yes
> >>TO_NAMD:PMEGridSizeX 32
> >>TO_NAMD:PMEGridSizeY 32
> >>TO_NAMD:PMEGridSizeZ 32
> >>
> >>
> >>every time i get this error:
> >>------------- Processor 0 Exiting: Called CmiAbort ------------
> >>Reason: FATAL ERROR: Periodic cell has become too small for
> >>original patch grid!
> >>Possible solutions are to restart from a recent checkpoint,
> >>increase margin, or disable useFlexibleCell for liquid simulation.
> >>
> >>Charm++ fatal error:
> >>FATAL ERROR: Periodic cell has become too small for original patch
> >>grid!
> >>Possible solutions are to restart from a recent checkpoint,
> >>increase margin, or disable useFlexibleCell for liquid simulation.
> >>------------------------------------------------------------------------
> >>
> >>I can't follow these instructions, because my useFlexibleCell is
> >>aready off
> >>
> >>useGroupPressure yes ;# needed for rigidBonds
> >>useFlexibleCell no
> >>useConstantArea no
> >>
> >>Please, help.
> >>
> >>
> >>
> >>
> >
>
> ------------------------------------------------------------------------------------------------------
> Joshua L. Adelman
> Biophysics Graduate Group Lab: 510.643.2159
> 218 Wellman Hall Fax: 510.642.7428
> University of California, Berkeley http://nature.berkeley.edu/~jadelman
> Berkeley, CA 94720 USA jadelman_at_berkeley.edu
> ------------------------------------------------------------------------------------------------------
>
>

-- 
David J. Hardy           office: 3047 BI
Beckman Institute        fax:    (217) 244-6078
University of Illinois   phone:  (217) 244-5493
405 N. Mathews Ave       email:  dhardy_at_ks.uiuc.edu
Urbana, IL 61801, USA    web:    http://www.ks.uiuc.edu/~dhardy/

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:48:06 CST