Use internal coordinates for optimization

Dear all

I would like to avoid the generation of internal coordinates in the optimization (optking) program, as I already have the defined connectivity information of all my molecules. So first question is of course why psi4 disregards this information even if the molecules is input as a zmat?

Next off I have tried to generate the intco.dat file (success), and have that file in the same directory where the input file is. But no matter what I call that intco file (intco.dat, I always see a:
Internal coordinates to be generated automatically.
Detected frag 1 with atoms: 1 2 3 4 5 6 7 8
before the first optimization step…

How should I provide my own intco file for my optimization job, or even better force optking to use the provided zmat for internal coordinates??


When Psi4 creates a molecule object to store that geometry, I’m fairly sure that it doesn’t keep the internal coordinates around but stores the Cartesian coordinates.

As for supplying your own files, I’ve stumbled onto a very similar problem myself just last night.

Psi4 looks for an intco.dat file… except it expects something like intco.XXXX.dat, where XXXX is the process id, a number that isn’t known until you actually run the job. (This is so that if you’re running jobs in parallel, the intermediate files on different processors don’t overwrite each other.) Short of modifying Psi4 yourself, I’m not aware of a way to get around this for an optimization run, though I would be quite interested to learn of one.

Yes, that also causes me trouble e.g. with reading orbitals from a file.

on topic: There is very little reason forcing psi4 to use a particular set of internal coordinates. The automatically generated redundant internals are usually superior regarding convergence to a z-matrix, and also check for “bad” angles close to 180deg, I believe.