Josh Izaac

Don't you need to use `x.at[i].set(x[i] + y[i])` here?

ah, so it happens at the 'get value' stage, got it!

Thanks @Qottmann! Just double checking, does #6075 aim to resolve this? It wasn't clear to me from the PR (as the description is blank)

Thanks! I guess one question; does the open PR _improve_ the behaviour even if it doesn't resolve it fully?

Got it, thanks @dwierichs!

Thanks for catching this @mehrdad2m! How involved would you say the fix is -- is it straightforward, or would it require some exploration?

We could take the middle ground, and only check it on the tape. So allow ```python qml.QubitStateVector(..., wires=0) qml.QubitStateVector(..., wires=1) qml.RX(0.3, wires=0) ``` but disallow ```python qml.QubitStateVector(..., wires=0) qml.RX(0.3, wires=0)...

@mews6 yep that's correct! We never updated the parameter-shift rule to support broadcasting, however it seems that the torch layer is providing the parameter-shift rule with a broadcasted tape, which...

Hey @austingmhuang, just wondering here how come you are adding `argnum` to the various qchem functionalities? I would expect that we simply need to change the logic in the qchem...

> For functions like `overlap_matrix`, we need to know what's differentiable in order to know what to do. So functionally, doing `qml.qchem.overlap_matrix()` will need that information somehow. Yep, and we...