Hi Brian,
What version of Amber?  What compile flags for pmemd?  There is code in place through Amber 10 for sure, though it was rarely tested, so a bug is not completely out of the question.  I don't know just yet what the status in Amber 11 is, whether there may be GPU support issues, etc.
Regards - Bob Duke
________________________________________
From: brian_jo.berkeley.edu [brian_jo.berkeley.edu]
Sent: Tuesday, October 11, 2011 5:30 PM
To: amber.ambermd.org
Subject: [AMBER] Question regarding HAS_10_12 flag in PMEMD (LINMIN Failure)
Hello,
We compiled sander and pmemd with -DHAS_10_12 (10-12 Hydrogen Bonding
Potential), and set up a short, 1000-step minimization run with ubiquitin.
However, the HAS_10_12 flag in PMEMD creates massive instabilities and
inaccuracies, as I compare an example output of sander and pmemd below.
The minimization eventually terminates due to repeated LINMIN failure.
We verified that the same problem occurs on multiple machines over
multiple versions of compilers, so this problem seems to do with source
code of PMEMD and how it treats 10-12 potential.
Has anyone either encountered the same problem, or has successfully run
PMEMD with HAS_10_12? Otherwise, a look into the PMEMD 10-12 potential
source code would be greatly appreciated.
Thank you!
Brian.
sander:
   NSTEP       ENERGY          RMS            GMAX         NAME    NUMBER
      1      -1.8508E+04     3.9593E+02     4.3374E+04     Cl-      1235
 BOND    =      126.0429  ANGLE   =      276.9695  DIHED      =      704.5199
 VDWAALS =    22134.3046  EEL     =   -45257.0072  HBOND      =      -19.4807
 1-4 VDW =      389.3294  1-4 EEL =     3137.1801  RESTRAINT  =        0.0000
pmemd:
   NSTEP       ENERGY          RMS            GMAX         NAME    NUMBER
      1      -1.8365E+04     3.9887E+02     4.3374E+04     Cl-      1235
 BOND    =      126.0429  ANGLE   =      276.9695  DIHED      =      704.5199
 VDWAALS =    22134.3046  EEL     =   -45257.0072  HBOND      =      123.1766
 1-4 VDW =      389.3294  1-4 EEL =     3137.1801  RESTRAINT  =        0.0000
Also a segment where LINMIN failures begin accumulating in pmemd:
   NSTEP       ENERGY          RMS            GMAX         NAME    NUMBER
    600      -5.1128E+04     7.6120E-01     5.2002E+01     O         737
 BOND    =     5323.4082  ANGLE   =      145.5072  DIHED      =      701.6303
 VDWAALS =     8858.3346  EEL     =   -69553.0311  HBOND      =      -64.3119
 1-4 VDW =      291.4799  1-4 EEL =     3038.8655  RESTRAINT  =      130.3178
 EAMBER  =   -51258.1173
     .... RESTARTED DUE TO LINMIN FAILURE ...
   NSTEP       ENERGY          RMS            GMAX         NAME    NUMBER
    650      -5.1350E+04     4.2329E+00     4.9058E+02     H         814
 BOND    =     5446.7030  ANGLE   =      178.9044  DIHED      =      711.1491
 VDWAALS =     9143.3622  EEL     =   -70252.1102  HBOND      =      -50.7623
 1-4 VDW =      291.3955  1-4 EEL =     3037.5668  RESTRAINT  =      143.6471
 EAMBER  =   -51493.7915
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Oct 11 2011 - 18:00:04 PDT