Open FlorianDeconinck opened 1 month ago
Thanks for reporting @FlorianDeconinck . Are you planning to also work in a fix or is it not high priority for you at the moment?
Unclear. We need a fix for sure but we have a mounting amount of physics pattern difficult to write so other feature requests might take precedence. I'd say we are a few weeks away from deciding where we put our efforts
The
numpy
backend has a limitation on literal accessor not shared by other backends. Consider the following:We expect
n_iter
to be some kind of local scalar variable. Butnumpy
is trying to use as much inner parallelism and masking as possible leading to this code giving aMeanwhile, running this code under a
gt:X
ordace:X
backend gives the expected results (withn_iter
a local scalar`).PS: this was unearth looking at the Aer Activation physics code for NASA's GEOS model