Re: [ESPResSo] Pressure calculation

[Torsten Stuehn]

The bug has been fixed in the newest CVS version.

For those of you who do not want to (or can not) upgrade to the
newest cvs version, here is what you have to change in your code:

find the following lines in pressure.h, in routine calc_bonded_force() [around line 300, change 
"i" to "(*i)" 4 times]:

---------------------------------------------------- snip 
-------------------------------------------------
     /* since it is not clear at the moment how to handle a many body 
interaction here, I skip it */
   case BONDED_IA_ANGLE:
     i++; force[0] = force[1] = force[2] = 0; break;
   case BONDED_IA_DIHEDRAL:
     i+=2; force[0] = force[1] = force[2] = 0; break;

#ifdef TABULATED
   case BONDED_IA_TABULATED:
     // printf("BONDED TAB, Particle: %d, P2: %d TYPE_TAB: 
%d\n",p1->p.identity,p2->p.identity,iparams->p.tab.type);
     switch(iaparams->p.tab.type) {
       case TAB_BOND_LENGTH:
          calc_tab_bond_force(p1, p2, iaparams, dx, force); break;
       case TAB_BOND_ANGLE:
         i++; force[0] = force[1] = force[2] = 0; break;
       case TAB_BOND_DIHEDRAL:
         i+=2; force[0] = force[1] = force[2] = 0; break;
---------------------------------------------------- snip 
-------------------------------------------------

and replace them with:

---------------------------------------------------- snip 
-------------------------------------------------
     /* since it is not clear at the moment how to handle a many body 
interaction here, I skip it */
   case BONDED_IA_ANGLE:
     (*i)++; force[0] = force[1] = force[2] = 0; break;
   case BONDED_IA_DIHEDRAL:
     (*i)+=2; force[0] = force[1] = force[2] = 0; break;

#ifdef TABULATED
   case BONDED_IA_TABULATED:
     // printf("BONDED TAB, Particle: %d, P2: %d TYPE_TAB: 
%d\n",p1->p.identity,p2->p.identity,iparams->p.tab.type);
     switch(iaparams->p.tab.type) {
       case TAB_BOND_LENGTH:
          calc_tab_bond_force(p1, p2, iaparams, dx, force); break;
       case TAB_BOND_ANGLE:
         (*i)++; force[0] = force[1] = force[2] = 0; break;
       case TAB_BOND_DIHEDRAL:
         (*i)+=2; force[0] = force[1] = force[2] = 0; break;
---------------------------------------------------- snip 
-------------------------------------------------

This will solve your problem.

Torsten.


Yelash, Dr. Leonid wrote:
> Hi Christine and other Espresso-Users,
>
> we also have similar problem in both the mpi and the fake versions:
>
> FAKE VERSION:
>
> add_bonded_virials: WARNING: Bond type 136622228 , atom 9 unhandled, Atom 2: 25
> /remote/home/sokrates/yelash/Espresso-2.0.5e-nompi/obj-Pentium_4_64-pc-linux/Espresso:
>  line 45: 11147 Segmentation fault      $ESPRESSO
>
> MPI VERSION RUN ON 1 CPU:
>
> add_bonded_virials: WARNING: Bond type 137329180 , atom 23 unhandled, Atom 2: 25
> add_bonded_virials: WARNING: Bond type 89 , atom 24 unhandled, Atom 2: 25
> add_bonded_virials: WARNING: Bond type 137328856 , atom 26 unhandled, Atom 2: 25
> p0_11474:  p4_error: interrupt SIGSEGV: 11
>
> Regards,
> Leonid
>
> _______________________________________
> From: address@hidden address@hidden On Behalf Of Ben Reynwar address@hidden
> Sent: Monday, January 28, 2008 5:29 PM
> To: Christine Peter
> Cc: address@hidden
> Subject: Re: [ESPResSo] Pressure calculation
>
> Hi Christine and all,
> I'm the one that moved stuff into calc_bonded_force.  It was to avoid 
> duplicating code for the pressure tensor and pressure calculations, but it's 
> entirely possible I screwed something up.  Unfortunately if I did, I don't know 
> where.
> Ben
>
> On 1/28/08, Christine Peter <address@hidden<mailto:address@hidden>> wrote:
> Hi all,
>
> I have a problem with the pressure calculation in Espresso.
> After Torstens email from last week about the bug in the pressure
> calculation with tabulated bonded potentials I checked out the latest
> release. This release gives idential results to my old runs if I do not
> execute the analyze pressure command, so everything I observe seems to
> be related to the pressure calculation:
>
> As soon as I execute analyze pressure I get a segmentation fault with
> the preceding error message:
>
> add_bonded_virials: WARNING: Bond type 9277 , atom 8925 unhandled, Atom
> 2: 8926
>
>
> It seems that at some point while looping through the atoms  and the
> bonds in add_bonded_virials a pointer goes astray. But I could not
> identify why at this atom and at this bond. My system has 1323 molecules
> with 9 atoms each. Each molecule has 8 bonds, 7 angles and 6 dihedral
> angles. Some of the bonds and angles are tabulated. The error occurs
> sort of unmotivated it seems.
>
> The debugger has just helped to identify that the error occurs in pressure.h
> at
> 356      p2 = local_particles[p1->bl.e[i++]];
>
> Could it be that it has sth to do with the change in pressure.h where
> parts from add_bonded_virials have been taken out and are now executed
> in calc_bonded_force (I do not know who mad that change) ?
>
> BTW: The error is not caused by Torsten's change to calculate the
> pressure for tabulated bonds, if I take that out I still get the
> segmentation fault.
>
> Has anybody observed anything similar?
>
> Cheers,
> Christine.
>
>
> --
> Christine Peter
> MPI for Polymer Research
> Ackermannweg 10
> 55128 Mainz / Germany
>
> Tel. +49-6131-379267
> email: address@hidden<mailto:address@hidden>
>
> _______________________________________________
> ESPResSo mailing list
> address@hidden<mailto:address@hidden>
> https://fias.uni-frankfurt.de/mailman/listinfo/espresso
>
> This email was Anti Virus checked by Astaro Security Gateway. 
> http://www.astaro.com
>
>
>
> _______________________________________________
> ESPResSo mailing list
> address@hidden
> https://fias.uni-frankfurt.de/mailman/listinfo/espresso
>
> This email was Anti Virus checked by Astaro Security Gateway. 
> http://www.astaro.com


--
Dr. Torsten Stühn
MPI für Polymerforschung
Ackermannweg 10
55128 Mainz / Germany

Tel.  +49-(0)6131-379268
Fax   +49-(0)6131-379100 
EMail address@hidden

smime.p7s
Description: S/MIME Cryptographic Signature