David Wierichs

For the first bug, the problem seems to stem from the behaviour of `tape.trainable_params` together with the Hamiltonian: ```python dev = qml.device('default.qubit', wires=2) qubit_operator = of.QubitOperator('X0 Y1', 1.0) hamiltonian =...

The second bug in the [minimal example comment](https://github.com/PennyLaneAI/pennylane/issues/2195#issuecomment-1128486227) above is now fixed due to #2584 being merged.

Good call, thanks! I found an incompatibility with qulacs and one with `braket.local.qubit` and fixed those. The current version works with (using `shots=None`, with `shots=1000` more tests seem to fail...

>That makes me wonder why other tests may be failing when using this PR. Thanks for the pointer, I wasn't aware of that repo! Given that the combinations of latest...

This is indeed just not implemented yet. But I have been working on the templates a bit. Let me get back to you tomorrow with a fix :) And yes,...

The above mentioned #2810 introduces broadcasting to templates. Note that the speedup only can be achieved once #2627 is merged! If it is required immediately, one could checkout the branch...

I think this should be resolved now and can be closed? @Qottmann

I can no longer reproduce this bug on master. @eddddddy, @josh146 , can you confirm?

Made the mistake of closing this, because it is fixed for `"default.qubit"`, although I had mentioned already that this will happen :facepalm: : >With the introduction of parameter broadcasting to...

It seems to me that all proposed changes here have been implemented by now. @cvjjm @josh146 what do you think?