Open
Description
What is the current behavior?
lmfit
is now warning for invalid covariances after we suppressed CurveAnalysis warnings in #1098 for the visualization tutorial:
WARNING: Cell printed to stderr:
2182
/home/runner/work/qiskit-experiments/qiskit-experiments/.tox/docs/lib/python3.8/site-packages/lmfit/minimizer.py:855: RuntimeWarning: invalid value encountered in sqrt
2183
par.stderr = np.sqrt(self.result.covar[ivar, ivar])
2184
/home/runner/work/qiskit-experiments/qiskit-experiments/.tox/docs/lib/python3.8/site-packages/lmfit/minimizer.py:862: RuntimeWarning: invalid value encountered in sqrt
2185
(par.stderr * np.sqrt(self.result.covar[jvar, jvar])))
(From this run)
I've also seen this warning from numpy
:
WARNING: Cell printed to stderr:
.tox/docs/lib/python3.8/site-packages/numpy/linalg/linalg.py:2139: RuntimeWarning: divide by zero encountered in det
r = _umath_linalg.det(a, signature=signature)
.tox/docs/lib/python3.8/site-packages/numpy/linalg/linalg.py:2139: RuntimeWarning: invalid value encountered in det
r = _umath_linalg.det(a, signature=signature)
Suggested solutions
The tutorials should have fixed seeds so this doesn't cause frequent failures. In the meantime, we should investigate which seeds can reproduce these warnings reliably.