Orbital transformation and MO output

[Ondrej Marsalek]

Dear all,

I have hit a small bug where CP2K fails in a cryptic way rather than
giving a useful error message.

Setting CARTESIAN in

CP2K_INPUT / FORCE_EVAL / DFT / PRINT / MO

silently expands the calculation to all orbitals rather than just
occupied. It would be nice if CP2K was a bit more verbose about that,
but it is not a problem in principle. However, if you combine this
with orbital transformation, the calculation "converges" in the first
SCF iteration by encountering a very small electronic gradient, even
though it is clearly not converged, when compared to a calculation
that does not set CARTESIAN. So the run finishes, but the calculation
is really failed. Is this because OT can't work with unoccupied
orbitals? Input file attached.

Also, I would find it useful if printing the MO information would also
trigger diagonalization, like printing MO cube files does. Again, this
is not a bug, but it would save some extra work (separate calculation)
that is needed in the current situation when one wants orbital
information from an OT calculation.

Best regards,
Ondrej
-------------- next part --------------
A non-text attachment was scrubbed...
Name: H2O.inp
Type: application/octet-stream
Size: 1808 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20120316/dff24e6b/attachment.obj>