Hi all,
The DETCI module was written to perform full CI and near-full-CI computations. It can technically do CISD computations, but only on very small systems. It will run out of memory and crash if it has very many orbitals (like ~200 in this case).
The reason for this is that, to make the full CI as fast as possible, lots and lots of metadata is stored in memory for each alpha and beta string in the CI. This is acceptable for full CI, because the number of alpha and beta strings is roughly the square root of the number of determinants. But for CISD, the number of strings is roughly proportional to the number of determinants. And the metadata size itself is also proportional to the number of orbitals (times the number of electrons). So, while the algorithm works just fine for full CI, it really isn’t appropriate for CISD.
I think Eugene DePrince implemented CISD for closed shell as part of his fnocc module, but unfortunately that only works for closed-shell molecules. DETCI is the fall-back for open-shell CISD, but just can’t handle problems as large as this.
Best,
David