dPCA analysis ends in error "Error in LAPACK ssyev() call: 663"

[chet...@gmail.com]

Dear Dr, Glykos,

I have been using grcarma for quite some time and recently struggling with this error where the dPCA analysis ends in an error after the second pass to calculate dihedral angle variances. I don't see apparent errors in the PSF and dcd files.

I have tried to remove the last few frames where it usually ends in error.

I also tried to run it in different folders and also using the latest carma release.

Error:

N    CA   C    atoms are declared in the PSF file.
[32m [1mIt appears that this DCD file contains unit cell information. [0m
Number of coordinate sets is 147334
Starting timestep         is 1
Timesteps between sets    is 1
Number of atoms in sets   is 2351
[33mStarting dPCA with 332 torsion angles

Maximum of variance-covariance matrix is -999999995904.000000
Minimum of variance-covariance matrix is 999999995904.000000
Writing data (formatted ASCII) to the file wwAMD.dcd.dPCA.varcov.dat.
Writing postscript file wwAMD.dcd.dPCA.varcov.ps.
Calculation of eigenvectors and eigenvalues ...
Asking for optimal workspace size : 22576
Starting the calculation ...
[31m [1mError in LAPACK ssyev() call: 663 [0m

[Nicholas Glykos]

Maximum of variance-covariance matrix is -999999995904.000000
Minimum of variance-covariance matrix is 999999995904.000000

These numbers are completely out of scale and context. Something is

wrong, but I can not guess what. I would start from the PSF, especially if 

it has ligands and/or multiple chains. Are the 332 torsion angles correct ?