problem with adding to numpy.float64, ufunc issue?

[arsenovic]

i am working on changing the operators (on this branch) to match those in galgebra: ie &->* and from *->|

however, i am getting some strange behavior which i dont undertand, which seems to be related to some operator overloads with numpy dtypes. it appears as though the float64.__add__ is taking precedence over cliffords MultiVector.__radd__. not sure why the operator changes created this problem.

@moble, @rkern perhaps you all have some ideas on how to solve this?

here is demo of error,
https://gist.github.com/13b0ee10dde76817c79aab567b682efa

[rkern]

The left operand gets the first shot at implementing the operation. Only if it fails does the right operand attempt with its __radd__ method. Because MultiVector implements the Sequence interface, np.asarray() knows how to convert it to a numpy array via np.asarray(). The numpy scalar types will attempt to convert the other operand to a numpy array via np.asarray(). Since this succeeds, np.float64.__add__() succeeds, and MultiVector.__radd__() does not get a chance to run.

I don't have a good idea offhand how to get around this.

[arsenovic]

thanks very much for the info.

is it clear to you that the operator change would cause this? the exp() function operating on a MultiVector now fails also, which i suspect is related.

a similar question is raised in this SO post. after looking at their suggestions the __array_priority__ does not work, but __array_wrap__ does. however this hook seems to be after the numpy __add__ has happened...

they also suggest to subclassing ndarray. however, i am not sure that i have the time/ability to make this change to the whole class, or to decide whether it is a good idea.

[arsenovic]

so if the left hand operator fails, the right operand attempt with its __radd__ method, then would it be possible to make the ufunc 'fail' by doing something in __array_prepare__

[arsenovic]

arsenovic@bd3bb72 . . . embarrasing

[hugohadfield]

@arsenovic is this now fixed or still a problem?

[arsenovic]

i think we can close