Martin Roelfs

If I understand the problem correctly, it comes down to the fact that model calls return a namedtuple, which is limited to 255 in older python versions. I would therefore...

A `namedtuple` has to be initiated using `args` and `kwargs` right, so I don't think that will work. I'll start coding something up and see about your other suggestions :).

Yes, but then you wont have the nice `__getattr__` anymore, then you can only access the full list, not the entries within that list as we need. And inheriting from...

I'm not sure I would use Constants for this. I don't know actually how ROOT defines Constants so it may be a solution, but the way I see it we...

I think this might already exist since we have everything `sympy` has: `sf.Zero`. So that might indeed be a good solution. Another option is to use a normal int `0`,...

Yeah I agree, some of the root finders might offer benefits over the #254 implementation, it is definitely worth investigating which ones, and wrapping them separately. And this would definitely...

@pckroon Thanks for the review. Of course, ultimately we should implement those Objectives. However, to give a general solution for least-squares-solving a tensor equation turns out to be non-trivial ;)....

Excellent work, thanks for tackling this. In response to the second issue: `LeastSquares` could be made compatible with complex numbers relatively straightforwardly, although it is is slightly more complicated than...

I think the problem is that `u[2]` is still an array with some shape, not a scalar. The constraint functions have to return a scalar. If I understand you correctly...

@egpbos , maybe I should've been more clear, indeed in your model `u[2]` is a scalar, but numerically it is not since you call fit with ```python cone_fit = sf.Fit(cone_model,...