Closed mtfishman closed 6 years ago
I agree we should definitely get the apply MPO methods to be consistent as much as possible. And yes, none of them should normalize the input wavefunction (onto which the MPO is being applied). So if any of them do, let’s treat it as a bug and fix asap!
Oh sorry I meant that "Fit" has an option to normalize the output wavefunction (not the input wavefunction). I don't see any harm in keeping that as an option, but I will make it so it is not the default, since I think people would expect by default that applyMPO(K,psi) -> K*psi, not K*psi/|K*psi|. I am working on an update now.
While writing the documentation I realized it is not ideal to have "Fit" normalize the wavefunction by default but "DensityMatrix" does not (and doesn't even have an option to). It seems like they should both default to not normalize the wavefunction, and maybe add an option to normalize the output of "DensityMatrix" if desired.