Re: Large Box Sizes and Total Energies Observed in NAMD 2.6b1 Binaries of AIX-POWER-MPI and Origin2000-MPI

From: Joshua D. Moore (jdmoore_at_unity.ncsu.edu)
Date: Fri Aug 19 2005 - 14:42:02 CDT

I sent another email previously with some attached files (~200k). It
didn't seem to post. Maybe it was too large?

Anyway, I think that I have figured out that my problem is with the
minimization. Minimizing for 2000 steps was plenty before, but now either
the conjugate gradient searching is not finding the minimum in 2.6b1 or
2000 steps is not long enough in 2.6b1 to find the minimum? Why is there
such a dramatic difference between 2.6b1 and 2.5 in this regard?

Eventually, I was smart enough to cheat. I ran for 1 ps on the AIX/IBM-SP
machines, and then used the .coor, .vel, and .xsc output as input in the
Origin machine. Everything seems to look ok. The energies match now.

Hopefully I can figure out how to use the minimization in 2.6b1, because
this will be a hassle if I have to relax my systems every time on a
different machine.

Thanks for everyone's help.

Joshua Moore

On Fri, August 19, 2005 7:38 am, Joshua D. Moore wrote:
> Thank you Mauricio. I tried this and I tried it again with smaller
> temperature increments, but it doesn't seem to help. I think the problem
> is that at the end of the minimization, the energy is way too high! Any
> addition of energy (from a Kinetic Contribution) causes the total energy
> to go way too high, and the system blows up!
>
> I have run the same system with the same configuration file on two
> different machines and 3 different binaries of NAMD (one compiled by me).
> Attached are the results and comparisons between the last step of the
> minimization (TS 2000). I am also attaching my output files for these
> three as well. Maybe the beginning information may be useful for
> determining what I am doing wrong. I believe the results ran on NAMD 2.5
> are correct, because eventually, it produces the correct g(r). Plus, the
> system does not blow up when I add kinetic energy in subsequent steps.
>
> I used 'minimize 2000' for the minimization steps for all. Everything is
> the same except in the 2.6b1 runs where I have included the
> 'vdwGeometricSigma' option because I am using OPLS. In the 2.5 version, I
> compiled it with an altered LJTable.C file to change the sigma from the
> arithmetic to geometric combining rule. The total energy for the 2.6b1
> runs is much higher, as well as differences in the Bonded, Angle, and
> Dihedral contributions. However, there are also differences between the
> 2.6b1 runs between the two different machines, but they seem to be around
> the same.
>
> My parameter file is in Charmm format, which I indicate by using
> 'paraTypeCharmm on'. Has the format/units for the parameters changed
> between versions of NAMD?? What can be my problem?
>
> Josh
>
> On Thu, August 18, 2005 9:04 pm, Mauricio Carrillo Tripp wrote:
>> try to increase the temperature slowly, instead of putting the kinetic
>> energy
>> all at once and see what happens...
>>
>> On 8/18/05, Joshua D. Moore <jdmoore_at_unity.ncsu.edu> wrote:
>>>
>>> Hi,
>>>
>>> After minimization, the reinitializing of the temperature causes large
>>> forces and energies and NAMD immediately crashes. In two other cases,
>>> it
>>> didn't actually crash, but the box size was huge. It should have been
>>> around 50 in that case, and it was 460!!
>>>
>>> I have run this same exact script (minus the vdwGeometricSigma option)
>>> in
>>> NAMD 2.5 (which I compiled myself) without any problems. Are there some
>>> syntax changes that I haven't figured out?
>>>
>>> Is there a problem with my configuration file (see below), or could it
>>> just be that the binaries are not behaving properly on these two
>>> systems?
>>>
>>> Thanks.
>>>
>>> Joshua Moore
>>>
>>> #############################################################
>>> ## JOB DESCRIPTION ##
>>> #############################################################
>>>
>>> # Minimization and Equilibration of
>>> # ethanol in a Water Box
>>>
>>>
>>> #############################################################
>>> ## ADJUSTABLE PARAMETERS ##
>>> #############################################################
>>>
>>> structure 7ethanol64_w1991.psf
>>> coordinates 7ethanol64_w1991.pdb
>>>
>>> set temperature 295
>>>
>>> #temperature: should only be set when starting from scratch
>>> temperature $temperature
>>> set outputname 7ethanol_out
>>>
>>> firsttimestep 0
>>>
>>> # continuing a run
>>> #set inputname 7ethanol_in
>>> #binCoordinates $inputname.coor
>>> #binVelocities $inputname.vel
>>> #extendedSystem $inputname.xsc
>>> #firsttimestep 502000
>>>
>>> #run stops when this step is reached
>>> #numsteps 1250000
>>>
>>> #set outputname 7ethanol_out
>>>
>>> #############################################################
>>> ## SIMULATION PARAMETERS ##
>>> #############################################################
>>>
>>> # Input
>>> paraTypeCharmm on
>>> parameters oplsaa_SPCE.prm
>>> vdwGeometricSigma yes
>>>
>>>
>>> # Force-Field Parameters
>>> exclude scaled1-4
>>> 1-4scaling 0.5
>>> cutoff 12.
>>> switching on
>>> switchdist 10.
>>> pairlistdist 13.5
>>>
>>>
>>> # Integrator Parameters
>>> timestep 1.0
>>> rigidBonds water
>>> nonbondedFreq 1
>>> fullElectFrequency 2
>>> stepspercycle 10
>>>
>>>
>>> # Constant Temperature Control
>>> langevin on ;# do langevin dynamics
>>> langevinDamping 5 ;# damping coefficient (gamma) of 5/ps
>>> langevinTemp $temperature
>>> langevinHydrogen off ;# don't couple langevin bath to hydrogens
>>>
>>>
>>> # Periodic Boundary Conditions: should only be used in starting a run
>>> cellBasisVector1 42.5 0. 0.
>>> cellBasisVector2 0. 42.5 0.
>>> cellBasisVector3 0. 0. 42.5
>>> cellOrigin 0. 0. 0.
>>>
>>>
>>> # PME (for full-system periodic electrostatics)
>>> PME yes
>>> PMEGridSizeX 48
>>> PMEGridSizeY 48
>>> PMEGridSizeZ 48
>>>
>>>
>>> # Constant Pressure Control (variable volume)
>>> useGroupPressure yes ;# needed for rigidBonds
>>> useFlexibleCell no
>>> useConstantArea no
>>>
>>> langevinPiston on
>>> langevinPistonTarget 1.01325 ;# in bar -> 1 atm
>>> langevinPistonPeriod 100.
>>> langevinPistonDecay 50.
>>> langevinPistonTemp $temperature
>>>
>>>
>>> # Output
>>> outputName $outputname
>>> #binaryoutput no
>>>
>>> restartfreq 500 ;# 500steps = every 1ps
>>> #binaryrestart no
>>> dcdfreq 250
>>> xstFreq 250
>>> outputEnergies 100
>>> DCDUnitCell yes
>>> outputPressure 100
>>>
>>>
>>> #############################################################
>>> ## EXTRA PARAMETERS ##
>>> #############################################################
>>>
>>>
>>> #############################################################
>>> ## EXECUTION SCRIPT ##
>>> #############################################################
>>>
>>> # Minimization
>>> minimize 2000
>>> reinitvels $temperature
>>>
>>> run 10000
>>>
>>>
>>>
>>
>>
>> --
>>
>> Mauricio Carrillo Tripp, PhD
>> Department of Chemistry
>> Wabash College
>> trippm_at_wabash.edu
>> http://trippm.bajacast.com/
>>
>
>
> --------------------------------------------------
> Joshua D. Moore
> Graduate Student
> North Carolina State University
> Department of Chemical and Biomolecular Engineering
> Box 7905 Centennial Campus
> Engineering Building I
> 911 Partners Way
> Raleigh, NC 27695-7905
> Phone: (919) 513-2051
> Fax: (919) 513-2470
> Email: jdmoore_at_unity.ncsu.edu
> --------------------------------------------------
>
>

--------------------------------------------------
Joshua D. Moore
Graduate Student
North Carolina State University
Department of Chemical and Biomolecular Engineering
Box 7905 Centennial Campus
Engineering Building I
911 Partners Way
Raleigh, NC 27695-7905
Phone: (919) 513-2051
Fax: (919) 513-2470
Email: jdmoore_at_unity.ncsu.edu
--------------------------------------------------

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:39:51 CST