From: Tristan Croll (tristan.croll_at_qut.edu.au)
Date: Thu Jul 18 2013 - 21:08:02 CDT
Are you using a CUDA version of NAMD by any chance? I've noticed that if any energy term goes outside of the range of single precision (which happens quite easily if your system starts out with any sort of structural problem), the energy minimization will quite keep happily keep rolling along doing nothing with energy terms at +/- 9999999999.9999. Try running your energy minimisation with a non-CUDA version of NAMD first.
Cheers,
Tristan
From: adrian palacios [mailto:adpala_at_hotmail.com]
Sent: Friday, 19 July 2013 11:33 AM
To: Tristan Croll; Aron Broom
Cc: namd-l_at_ks.uiuc.edu
Subject: RE: namd-l: Minimization/Equilibrium problem, atoms moving too fast
Thanks for that Tristan, although it was not the main problem I will try to use this instead of my improvised procedure. If it solves the problem IŽll let you know, although I donŽt think it will, but lets be optimists.
Thanks for the response!
Adrian Palacios
________________________________
From: tristan.croll_at_qut.edu.au
To: broomsday_at_gmail.com; adpala_at_hotmail.com
CC: namd-l_at_ks.uiuc.edu
Date: Fri, 19 Jul 2013 11:23:35 +1000
Subject: RE: namd-l: Minimization/Equilibrium problem, atoms moving too fast
The cell origin should be the center of mass of your system:
measure center $all.
Unless you have used the "rotate to minimize volume" option in Solvate, your center could easily be tens or hundreds of Angstroms from {0 0 0} and your simulation will go crazy trying to move the system there.
If you're using the built-in solvate and ionize GUI tools in VMD, you should be able to just type:
pbc get
to get the box dimensions.
From: owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] On Behalf Of Aron Broom
Sent: Friday, 19 July 2013 11:06 AM
To: adrian palacios
Cc: namd-l_at_ks.uiuc.edu
Subject: Re: namd-l: Minimization/Equilibrium problem, atoms moving too fast
I'm not super familiar with exactly how NAMD does the periodic stuff or again what that VWD Boundary term means, but I wonder what would happen in your original larger simulation if you replaced:
"cellOrigin 0 0 0"
with
"cellOrigin 80.5 73.0 72.0"
Maybe nothing, but worth testing since it's a quick change.
On Thu, Jul 18, 2013 at 8:58 PM, adrian palacios <adpala_at_hotmail.com<mailto:adpala_at_hotmail.com>> wrote:
Well, my procedure to define the cell dimensions is running the "measure minmax $sel" where sel is the selection of the water box and using the diference between those vectors as each of the cell dimensions, for example, if it gives me {-80.5 . .} and {80.4 . .} I give a cell vector {161 0 0} as I did in my configuration file, trying to give numbers rounded to the greater integer. And sorry for my incorrect expression, I was thinking of the absolute value of energy, but you are correct in that point.
Before you answered I looked for the vdw "NAMD Plot" from the minimization log file and it does represent the greater contribution to this unstability, but how could I fix this? I tried running a similar system in a smaller box and it worked, but still is the second time I have this problem related to CaCl Ionization neutralizing the system.
Thanks for your reply, it was very accurate.
Adrian Palacios
________________________________
Date: Thu, 18 Jul 2013 17:12:32 -0400
Subject: Re: namd-l: Minimization/Equilibrium problem, atoms moving too fast
From: broomsday_at_gmail.com<mailto:broomsday_at_gmail.com>
To: adpala_at_hotmail.com<mailto:adpala_at_hotmail.com>
CC: namd-l_at_ks.uiuc.edu<mailto:namd-l_at_ks.uiuc.edu>
>From the log, your energies actually go crazy low after minimization, not crazy high. And this seems to result from the ' VDW BOUNDARY' term. I'm not sure what this term applies to, but how confident are you in the cell dimensions you gave the system?
On Thu, Jul 18, 2013 at 4:28 PM, Adrian <adpala_at_hotmail.com<mailto:adpala_at_hotmail.com>> wrote:
Hello,
I have been receiving an error during equilibration that says:
OPENING EXTENDED SYSTEM TRAJECTORY FILE
ERROR: Atom 224 velocity is -3.36524e+09 -5.3944e+10 -6.93782e+10 (limit is 14000, atom 76 of 587 on patch 328 pe 7)
ERROR: Atom 226 velocity is 5548.87 -31899.4 21857.1 (limit is 14000, atom 78 of 587 on patch 328 pe 7)
ERROR: Atom 231 velocity is 4.01192e+10 6.43101e+11 8.27101e+11 (limit is 14000, atom 83 of 587 on patch 328 pe 7)
ERROR: Atoms moving too fast; simulation has become unstable (3 atoms on patch 328 pe 7).
ERROR: Exiting prematurely; see error messages above.
But I have seen that this velocity problems arise from the minimization, which only runs for 100 steps and suddenly energies goes crazy high (see below).
I have searched in namd-l and even in gromacs forums (they seem to have this problem too sometimes) and the only solutions I found are given for equilibration, such as gradual heat, which I tried but doesn't work (and shouldn't) as my systems becomes unstable from minimization...
what could I do?
Below theres a part of the minimization log file where the energies suddendly jump an go crazy:
******************************************************************************************************
LINE MINIMIZER BRACKET: DX 0 7.46617e-05 DU 0 804922 DUDX -1.17635e+11 -1.17635e+11 9.89855e+09
ENERGY: 98 82987.3409 32267.5604 378.1408 398.7799 -1379409.6821 220635.2885 0.0000 0.0000 0.0000 -1042742.5716 0.0000 -1042742.5716 -1042742.5716 0.0000
LINE MINIMIZER BRACKET: DX 0 2.85182e-05 DU 0 648998 DUDX -1.17635e+11 -1.17635e+11 -1.41585e+10
ENERGY: 99 53742.1489 26794.8140 378.9026 391.4425 -1379049.7368 88822.2874 0.0000 0.0000 0.0000 -1208920.1414 0.0000 -1208920.1414 -1208920.1414 0.0000
LDB: ============= START OF LOAD BALANCING ============== 12.3247
LDB: Largest compute 8450 load 0.006281 is 0.4% of average load 1.470207
LDB: Average compute 0.001140 is 0.1% of average load 1.470207
LDB: Partitioning computes with target load 0.147021
LDB: Increased migratable compute count from 3456 to 3456
LDB: Largest unpartitionable compute is 0.006281
LDB: ============== END OF LOAD BALANCING =============== 12.3267
Info: useSync: 1 useProxySync: 0
LDB: =============== DONE WITH MIGRATION ================ 12.3279
LINE MINIMIZER BRACKET: DX 0 1.0893e-05 DU 0 482821 DUDX -1.17635e+11 -1.17635e+11 -6.00576e+09
Info: Initial time: 8 CPUs 0.0917141 s/step 1.06151 days/ns 545.605 MB memory
ETITLE: TS BOND ANGLE DIHED IMPRP ELECT VDW BOUNDARY MISC KINETIC TOTAL TEMP POTENTIAL TOTAL3 TEMPAVG
ENERGY: 100 50712.3949 26491.2341 381.0279 384.0135 -1378906.7507 153570.1053 0.0000 0.0000 0.0000 -1147367.9752 0.0000 -1147367.9752 -1147367.9752 0.0000
LINE MINIMIZER BRACKET: DX 0 4.16075e-06 DU 0 544373 DUDX -1.17635e+11 -1.17635e+11 4.31289e+10
OPENING COORDINATE DCD FILE
WRITING COORDINATES TO DCD FILE AT STEP 100
The last position output (seq=100) takes 0.016 seconds, 545.797 MB of memory in use
ENERGY: 101 50478.8814 26312.4298 378.2435 378.8770 -1378841.8835 -9999999999.9999 0.0000 0.0000 0.0000 -9999999999.9999 0.0000 -9999999999.9999 -9999999999.9999 0.0000
LINE MINIMIZER BRACKET: DX 1.58927e-06 2.57149e-06 DU -1.38707e+11 1.38708e+11 DUDX -1.17635e+11 1.17012e+11 4.31289e+10
ENERGY: 102 50518.8841 26374.8347 378.6484 378.0852 -1378867.4574 -436743.2275 0.0000 0.0000 0.0000 -1737960.2326 0.0000 -1737960.2326 -1737960.2326 0.0000
LINE MINIMIZER BRACKET: DX 1.58927e-06 9.82221e-07 DU -1.38707e+11 1.38707e+11 DUDX -1.17635e+11 1.17012e+11 1.25833e+11
ENERGY: 103 50481.9200 26279.3515 378.2880 379.4739 -1378826.8819 -9999999999.9999 0.0000 0.0000 0.0000 -9999999999.9999 0.0000 -9999999999.9999 -9999999999.9999 0.0000
*******************************************************************************************************
Here is my minimization configuration file:
#############################################################
## JOB DESCRIPTION ##
#############################################################
# Minimization of 1AML_Ca2mM
#############################################################
## ADJUSTABLE PARAMETERS ##
#############################################################
structure 1AML_Ca2mM.psf
coordinates 1AML_Ca2mM.pdb
outputname minCa2mM
set temperature 298
firsttimestep 0
#############################################################
## SIMULATION PARAMETERS ##
#############################################################
# Input
paraTypeCharmm on
parameters /usr/local/lib/vmd/plugins/noarch/tcl/readcharmmpar1.2/par_all27_prot_lipid_na.inp
temperature $temperature
# Periodic Boundary conditions
if {0} {
cellBasisVector1 161 0 0
cellBasisVector2 0 146 0
cellBasisVector3 0 0 144
cellOrigin 0 0 0
}
wrapAll on
# Force-Field Parameters
exclude scaled1-4
1-4scaling 1.0 ;# already a default value
cutoff 14.0
switching on ;# already a default value
switchdist 12.0
pairlistdist 16.0
# Integrator Parameters
timestep 1.0
rigidBonds water
nonbondedFreq 2
fullElectFrequency 4
stepspercycle 20
restartfreq 2000
dcdfreq 100
xstFreq 100
outputEnergies 200
#############################################################################
## EXTRA PARAMETERS ##
#############################################################################
binaryoutput no
binaryrestart no
#############################################################
## EXECUTION SCRIPT ##
#############################################################
minimize 1000
Thanks for your attention
Adrian Palacios
-- Aron Broom M.Sc PhD Student Department of Chemistry University of Waterloo -- Aron Broom M.Sc PhD Student Department of Chemistry University of Waterloo
This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:23:30 CST