Numpy release causes issues

[eggerdj]

Informations

• Qiskit Experiments version: main
• Python version: 3.11
• Operating system: All

What is the current behavior?

A numpy update is causing tests to fail. See e.g.

• https://github.com/Qiskit-Extensions/qiskit-experiments/actions/runs/5311531622/jobs/9615482020
• Incompatibility with Numpy 1.25 Qiskit/qiskit#10305

Steps to reproduce the problem

Run tests with numpy >= 1.26

What is the expected behavior?

Tests should pass.

Suggested solutions

Short-term solution: pin numpy in tests.
Long-term solution: align to numpy 1.26. Mostly looks like fixing this: DeprecationWarning: productis deprecated as of NumPy 1.25.0, and will be removed in NumPy 2.0. Please useprod instead.

[jakelishman]

Fwiw: we've already merged fixes to all the new deprecation warnings into mainline and stable Terra (we haven't cut a release from them, though, so downstream CI will still likely fail), but there's some deep weirdness with some core Numpy routines that seems to be the root cause of some correctness issues in our Isometry decompositions. Well, I actually can't see why the particular weirdnesses I'm seeing in Numpy code should break our decompositions, to be fair, but there's definitely something going on.