Davide Sandonà

Results 65 comments of Davide Sandonà

Note: setting `L = symbols("L", positive=True)` allows the computation to be quickly solved. I guess it is better to include a note in the documentation of `Beam` that a symbolic...

> I just remembered that in the past we have always recommended implementing functions that return complex numbers as a vector and add the logic in the loss function. >...

Hello @alex47 , any news on the new implementation?

I'm going to attempt a fix in a couple of days...

I'm battling with the same problem. With trials and errors I understood that: * it only happens with Chrome when using headless mode. I never encountered the problem by running...

> This needs to be more explicit. What exactly is breaking? Can you say explicitly what sort of downstream code would break? Most users won't note any difference because they...

> I don't oppose this change but I just want to note that from a high level this seems like a step backwards to me in design terms. My experience...

> Is it possible to avoid __new__ by defining a function? I suppose it can be done. Just pushed a new commit, let's see what happens.

If you place a `__new__` into the base class `Plot`, you have to place a `__new__` also on subclasses, otherwise you end up with infinite recursion when creating a new...

Than you would remove `MatplotlibBackend` and `TextBackend` from the method of `Plot.__new__`, making it superfluous.