[gmx-developers] Nr of graph edges per atom and distance restraints

[Berk Hess]

Juergen Haas wrote:

>On Thursday 08 March 2007 10:03, Berk Hess wrote:
>  
>
>>Hi,
>>
>>I assume this mail was meant for Bert de Groot, not the developers list?
>>
>>Anyhow, the problem you mention has not been fixed (yet).
>>We should make the graph memory allocation fully dynamic.
>>To run your simulation, you need to increase the number 4
>>on line 229 of src/gmxlib/mshift.c. Currently the code
>>only supports 4 restraints per atom.
>>    
>>
>I have slowly increased this value until I do not get this error message 
>anymore.
>mdrun, however, segfaults around line 166.c in orires.c...
>  
>
Line 166 is for the ensemble averaging code.
Are you really using ensemble averaging?
I don't see what could go wrong around that line.
The best thing would be to compile Gromacs with CFLAGS=-g
and run in the debugger.

>to explain the restraints I apply:
>I am using restraints originating from four atoms. In each case there is one 
>restraint for almost every bb-N-H, but never more than one..
>which means on four atoms we have around 140 constraints arising from the 
>restraints from this atom, as well as to this atom(+3, from the other 3 
>atoms).
>
>  
>
I don't undertand this last line.
How do you mean restraints originating from four atoms?
Orientations restraints only work on atom pairs.
And what do constraints have to do with orientation restraints?

>Since I have just started using restraints I might consider this a user 
>problem:-)
>
>regards
>
>Juergen
>
>  
>