[gmx-users] g_covar Segmentation fault problem

gozde ergin gozdeeergin at gmail.com
Thu Jul 16 16:23:01 CEST 2015

Dear gromacs user,

In order to estimate system absolute entropy value I want to use g_covar +
g_anaeig commands.

However when I *'g_covar -f nvt.trr -s nvt.tpr'*  I get a segmentation
fault after *'Diagonalizing ....'* step from gromacs 4.6.6 and 5.0.3.

*Constructing covariance matrix (12258x12258) ...Last frame       4000 time
4000.000   Read 4001 framesTrace of the covariance matrix: 12454.6
(nm^2)Diagonalizing ...Sum of the eigenvalues: 12454.6 (nm^2)Segmentation
fault (core dumped)*

But the same command, executed with g_covar of gromacs 4.6.3 works without
any error or fault.

Was it a bug and corrected for newer versions? I saw  the release note of *
'Fixed a complicated bug in g_anaeig and g_covar, if the number of frames
for a covariance analysis is fewer than the number of degrees of freedom'  *for
gromacs 4.6.6.

Do I need to lower the number of atoms to analyze or higher the frame
numbers or is it ok if I use 4.6.3?


More information about the gromacs.org_gmx-users mailing list