Re: ABF with Alpha colvar can't be used on supercomputer?

From: Jérôme Hénin (jhenin_at_ifr88.cnrs-mrs.fr)
Date: Thu Sep 29 2011 - 05:04:42 CDT

Yes: a bug was introduced, that's what happened. Then it was fixed in
the next version.

On 29 September 2011 08:05, JC Gumbart <gumbart_at_ks.uiuc.edu> wrote:
> Something changed between NAMD versions 2.8b2 and 2.8b3 that causes this error, although I can't figure out what it is.  Hopefully Jerome can enlighten us?
>
>
> On Sep 28, 2011, at 3:38 PM, Yun Luo wrote:
>
>>
>> Dear NAMD developers,
>>
>> I want to use ABF with colvar "alpha" for a ten-alanine peptide. It seems work when I tested my script on linux cluster. But when I use NAMD2.8 on Blue Gene/P supercomputer, the output gives error message as below:
>> colvars: ----------------------------------------------------------------------
>> colvars:   Initializing a new collective variable.
>> colvars:   # name = alpha
>> colvars:   Initializing a new "alpha" component.
>> colvars:     # componentCoeff = 1 [default]
>> colvars:     # componentExp = 1 [default]
>> colvars:     # period = 0 [default]
>> colvars:     # wrapAround = 0 [default]
>> colvars:     # psfSegID = HLX
>> colvars:     # hBondCoeff = 0.5
>> colvars:     # hBondCutoff = 3.3
>> colvars:     # hBondExpNumer = 6
>> colvars:     # hBondExpDenom = 8
>> colvars:     Error: cannot add atoms to a dummy group.
>> colvars:     If this error message is unclear, try recompiling with -DCOLVARS_DEBUG.
>> FATAL ERROR: Error in the collective variables module: exiting.
>> -----------------------------------------------------------------------------------------
>>
>> Is this a bug related to supercomputer version?
>> My colvar and abf input file is shown as below:
>> colvarsTrajFrequency     100
>> colvarsRestartFrequency  100
>>
>> colvar {
>>  name alpha
>>  width 0.1
>>  lowerboundary 0.0
>>  upperboundary 1.0
>>  lowerwallconstant 10.0
>>  upperwallconstant 10.0
>>  extendedLagrangian on
>>
>>  alpha {
>>       residuerange 1-10
>>       psfSegID HLX
>>       hbondcoeff 0.5
>>       angleref    88
>>       angletol   15
>>       hbondcutoff  3.3
>>       hbondexpnumer 6
>>       hbondexpdenom 8
>> }
>> }
>>
>> abf {
>>  colvars alpha
>>  fullSamples 10000
>>  applyBias yes
>>  outputFreq 100
>>  hideJacobian  no
>> }
>>
>> Thanks !
>>
>> Lyna
>
>
>

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:20:51 CST