Closed gitpeterwind closed 1 year ago
The unit tests pass with the new fixed MRCPP, but real calculations do not produce correct numbers. As far as I can tell the issue lies somewhere in the orbital::rotate
function, which does not seem to produce correct orbitals in line 237 in FockBuilder::buildHelmholtzArgument
. I have not looked further into what the issue might be.
Patch coverage: 74.57
% and project coverage change: -0.48
:warning:
Comparison is base (
57b33fb
) 70.00% compared to head (02fb075
) 69.53%.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Do you have feedback about the report comment? Let us know in this issue.
Version that uses the mrcpp version, where Bank and MPI are moved to mrcpp. Lots of names are changed. QMFunctions and Orbitals are now the same class, and is defined in mrcpp. ComplexFunction.
New treatment and defintion of NuclearPotential. The potential is now defined with a parabola to remove the cusp. The application of the potential in Fock is using the new multiplication method (nodes are stored in Bank).
It is neither faster nor use less memory... yet. But it will help to get there.
Edit 28th Feb: reversed to earlier NuclearPotential method and function. The tools developed for this are to be used anyway, but the Nuclear potential can/should perhaps be included in a more general way, by storing the potential directly in the bank.