TURBOMOLE Users Forum

TURBOMOLE Modules => Define => Topic started by: willows9803 on February 08, 2008, 03:23:26 PM

Title: What should I do if the B-matrix is singular?
Post by: willows9803 on February 08, 2008, 03:23:26 PM

I tried to run dscf on a big molecule with 52 atoms. However, the relax found the B-matrix was singular (the internal coordinates were linearly dependent) and stopped abnormally. Is there anything I can do to eliminate this singularity in the Define script?

Thanks a lot!

Title: Re: What should I do if the B-matrix is singular?
Post by: christof.haettig on February 12, 2008, 09:55:35 PM

Hello,

it's quite unusual that define is not able to construct useful redundant internal coordinated with the 'ired' option. But it can happen for molecules with a very complicated structure (several rings and cages, possibly mixed with intermolecular connections are a nightmare). Than that easiest way out is to do the geometry optimization in cartesian coordinates. This is not very efficient, but sometimes the only solution. After some geometry cycles in cartesian coordinates one can try again to construct redundant internal coordinates.

christof

Title: Re: What should I do if the B-matrix is singular?
Post by: willows9803 on February 19, 2008, 02:39:29 PM

Yes, first cartesian, then internal, it works!
Thanks a lot, christof! :D

Quote from: christof.haettig on February 12, 2008, 09:55:35 PM

Hello,

it's quite unusual that define is not able to construct useful redundant internal coordinated with the 'ired' option. But it can happen for molecules with a very complicated structure (several rings and cages, possibly mixed with intermolecular connections are a nightmare). Than that easiest way out is to do the geometry optimization in cartesian coordinates. This is not very efficient, but sometimes the only solution. After some geometry cycles in cartesian coordinates one can try again to construct redundant internal coordinates.

christof