[gmx-users] Re: Hessian Matrix

[Justin A. Lemkul]

#ZHAO LINA# wrote:
> Hi,
> 
> I just simply did a change in .mdp file. Now the integrator	= nm	
> after grompp, 
> 
> 1) below is the error of  mdrun -s em.tpr -mtx em.mtx
> 
> Program mdrun, VERSION 4.0.7
> Source code file: ../../../../src/gmxlib/smalloc.c, line: 147
> 
> Fatal error:
> Not enough memory. Failed to calloc 4118944041 elements of size 4 for full_matrix
> (called from file ../../../../src/mdlib/minimize.c, line 2219)
> 

You might want to take a step back and consider what you're doing.  I didn't 
suggest you use the nm integrator ad hoc to get some arbitrary output.  You have 
to understand what normal modes calculations are before you just dive in and try 
them.  They actually require quite a bit of finesse:

http://www.gromacs.org/Documentation/How-tos/Normal_Mode_Analysis

You're asking mdrun to do a huge amount of calculation, which your system may 
not be able to handle.  Perhaps if you can define why (you think) you need a 
Hessian you'll get some more specific advice.

> 2) Seems the one run in cluster choked there like this in md.log.
> Linking all bonded interactions to atoms
> There are 2106 inter charge-group exclusions,
> will use an extra communication step for exclusion forces for PME
> 
> The initial number of communication pulses is: X 1 Y 1
> The initial domain decomposition cell size is: X 2.76 nm Y 1.07 nm
> 
> The maximum allowed distance for charge groups involved in interactions is:
>                  non-bonded interactions           1.000 nm
>             two-body bonded interactions  (-rdd)   1.000 nm
>           multi-body bonded interactions  (-rdd)   1.000 nm
> 
> 

I don't understand how this is related.  These are normal messages related to 
domain decomposition.

-Justin

> 
> Thanks for further suggestion,
> 
> lina
> 
> 
> #ZHAO LINA# wrote:
>> Hi everyone,
>>
>> I tried to use mdrun to get the .mtx file, but it does not work, I mean...no mtx file output.
>> Below is the command I used in the scripts of my last two trial.
>>
>> 1)
>> ## To run on 16 cpus
>> #PBS -l nodes=2:ppn=8
>>
>> ## program to run
>> mpirun -np $NCPUS mdrun_mpi -mtx em.mtx -deffnm em 
>>
>> 2)
>> ## program to run
>> mpirun -np $NCPUS mdrun_mpi -s em.tpr -mtx em.mtx 
>>
> 
> Probably because the Hessian is only relevant when doing normal mode 
> calculations (or perhaps L-BFGS minimization).  I suspect that if you're not 
> using the nm integrator, then you're not going to get the Hessian.
> 
> -Justin
> 
>> Thanks and best,
>>
>> lina
>>
> 

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================