[gmx-users] Origin of g_mindist problems located

[Volker Knecht]

Dear all,

the wrong output of g_mindist version 3.3 reported recently seems to 
have been due to an inconistency of the default value for the  flag 
"-ng". Although upon "g_mindist -h" a default value "-ng 1" is claimed, 
it seems from the code that the actual default used is 0. For "-ng 0", 
g_mindist asks for a second index group, but ignores it and replaces it 
by the first one (as it seems from the code). The program works fine if 
the flag "-ng 1" is set explicitely. Nevertheless, if I am right, I 
would suggest to set the default value "-ng" used by the program to one.

A segmentation fault is obtained with g_mindist if the names for the 
selected groups created by make_ndx are too long for complicated 
selections. If these automatically generated names are replaced by 
shorter ones manually in a text editor, the program again works fine.

Cheers,
Volker

-- 
Dr. Volker Knecht
Theory Department
Max Planck Institute of Colloids and Interfaces
Research Campus Golm
14424 Potsdam, Germany

Phone: +49-331-567-9610
Fax: +49-331-567-9612