Codecov Report

Attention: Patch coverage is80.00000% with4 lines in your changes missing coverage. Please review.

Project coverage is 84.47%. Comparing base(646255a) to head(5faa17f).
Report is 78 commits behind head on master.

@@            Coverage Diff             @@##           master    #1974      +/-   ##==========================================- Coverage   84.47%   84.47%   -0.01%==========================================  Files         314      314                Lines       19057    19075      +18       Branches     2387     2391       +4     ==========================================+ Hits        16099    16113      +14- Misses       2199     2201       +2- Partials      759      761       +2

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report?Share it here.

shouldnt the conversion from vector to integer just fail if the constraint is violated?

I mean if get_conn returns a state outside the Hilbert space one should just error or ignore it ?

Indeed, that is what happens currently.
But the error is not very clear, as testified by#1973.

It would be nice if we could automatically make the operators work in the chosen subspace.
I was plying with doing that.
But it'd be more work.

I am not sure we should do this... at first sight it makes the code harder to read, adding another layer of logic we don't really need... I think we should just error and it should be the job of the operator to make sure it acts correctly on the given (constrained) hilbert space !

Yeah, I'm also unsure.

Though automatic stuff is very handy sometimes...
This kind of stuff is what is hard to work with.
Sometimes writing the projector for elaborate constraints is painful...