David Wierichs

I don't know what the `.layout` file does... Probably it's fine then! :)

How do we stand on this now? By now someone pointed out to me that the arxiv DOIs are not used.. In my opinion it does not make any major...

I am slightly confused by the chain rule not working here :sweat_smile: Assuming f, g and x to be scalars, we have (as you wrote): `d/dx f(g(x)) = f'(g(x)) *...

Ah, I see now :) So basically a variant of `qml.gradients.param_shift` that treats the input function as a black box and just shifts the function arguments would be useful here?...

I mean that we could have the following: ```python def shift_it(fn, *params, shifts): jac = np.zeros_like(params) for idx in range(len(params)): for coeff, mult, shift in shifts[idx]: new_params = params.copy() new_params[idx]...

Yes, I think that's the right perspective...I'm not sure what the best way of proceeding is here. The idea of registering callables instead of `MeasurementProcesses` in QNodes is pretty cool,...

This PR would be good for a review :) The failing Codecov check should be due to more lines being removed than added here, and hence the reduced coverage rate.

Thanks @antalszava! I agree, this is a bug with very low visibility. Conceptually, I'd prefer if there is an error raised and a device that supports `Hamiltonian` measurements suppresses it,...

Thanks for digging this one out! :+1: The reason seems to be, that `qml.transforms.broadcast_expand` is not JIT compatible. **edit:** Actually, that's not quite true. The problem is that by default,...

I looked into this once more. 1. `default.qubit` does not longer produce the bug, because it now makes use of broadcasting. Replace it by `default.mixed` for reproducing/testing purposes. This also...