The Q-space equations in OATDCC are based on the anti-symmetrized u-matrix elements, however Kvaal's 2012-article states that these should be derived from the non-anti-symmetrized matrix elements, as seen in eq 33 or more explicitly on page 5 of the appendix.
It seems like this requires us to also store and update a version of the non-anti-symmetrized matrix elements, which is a shame in computational terms. I have managed to implement a fix in the my hacked DVR-basis-oatdcc, which already stores u-matrix elements in their original form and anti-symmetrizes when needed (due to the sparse 2d representation). However, I find it an extremely open question how to best implement this in the "official" OATDCC.
The Q-space equations in OATDCC are based on the anti-symmetrized
u
-matrix elements, however Kvaal's 2012-article states that these should be derived from the non-anti-symmetrized matrix elements, as seen in eq 33 or more explicitly on page 5 of the appendix.It seems like this requires us to also store and update a version of the non-anti-symmetrized matrix elements, which is a shame in computational terms. I have managed to implement a fix in the my hacked DVR-basis-oatdcc, which already stores u-matrix elements in their original form and anti-symmetrizes when needed (due to the sparse 2d representation). However, I find it an extremely open question how to best implement this in the "official" OATDCC.