Fwd: Failure of namd2.10 in keeping trace of cell in t-remd

From: Francesco Pietra (chiendarret_at_gmail.com)
Date: Tue Feb 18 2014 - 12:45:35 CST

While walking I was re-thinking about this issue. Perhaps, I should have
commented out the cellBasisVectors, as we do on restarting MD. However, how
could I have foreseen that the box has enlarged, and so much, in order to
resetting the PMEGridSize? Is it implied that one has to check each replica
to verify the cell? On restarting MD I never had to reset the PMEGridSize,
although I recognize that here there is a shift between different
temperature conditions (314-600K was the range I used).

PME yes
cellBasisVector1 67.05 0. 0.
cellBasisVector2 0. 56.58 0.
cellBasisVector3 0. 0. 64.64

cellOrigin -14.232172012329102 29.93465805053711 -67.4001693725586
PMEGridSpacing 1.0
PMEGridSizeX 70
PMEGridSizeY 60
PMEGridSizeZ 68

francesco pietra

---------- Forwarded message ----------
From: Francesco Pietra <chiendarret_at_gmail.com>
Date: Tue, Feb 18, 2014 at 12:44 PM
Subject: namd-l: Failure of namd2.10 in keeping trace of cell in t-remd
To: NAMD <namd-l_at_ks.uiuc.edu>

In a t-remd with a 41 aa peptide, 32 replicas, restarting t-remd does not
keep trace of the status of colvars (rmsd applied to a small, initial
stretch of the peptide), as I posted before for GB t-remd.

Now, for the same t-remd, in a periodic box with TIP3 water, on restarting
the initial, correct t-remd, the status of the cell is incorrectly
reported in the .xsc file for replicas 20, 22, 23, and 25, with crash

>From the .err file (or the signle log files in the various replicas)

Reason: REPLICA 23 FATAL ERROR: PMEGridSizeX 70 is too small for cell
length 73.054500 and PMEGridSpacing 1.000000

namd2: machine.c:708: LrtsAbort: Assertion `0' failed.
------------- Processor 0 Exiting: Called CmiAbort ------------
Reason: REPLICA 22 FATAL ERROR: PMEGridSizeX 70 is too small for cell
length 72.344600 and PMEGridSpacing 1.000000

namd2: machine.c:708: LrtsAbort: Assertion `0' failed.
------------- Processor 0 Exiting: Called CmiAbort ------------
Reason: REPLICA 20 FATAL ERROR: PMEGridSizeX 70 is too small for cell
length 70.826300 and PMEGridSpacing 1.000000

------------- Processor 0 Exiting: Called CmiAbort ------------
Reason: REPLICA 25 FATAL ERROR: PMEGridSizeX 70 is too small for cell
length 74.891500 and PMEGridSpacing 1.000000

Actually, the same can be expected for all replicas, as the other replicas
had not yet reached the stage of looking for .xsc (same problem observed in
GB for the status of colvars)

No such problems with the same files for MD in explicit water: both the
cell status and the status of colvars are correctly transferred to the
restarting MD.

francesco pietra

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2015 - 23:20:31 CST