alex arsenovic

ok thanks for the info. i will keep an eye on #36 because i'd like to be able to reference classes within the markdown cells. otherwise its too cumbersome.

thats exactly what i was referring to. i would use it if it where included in this project or jupyter or sphinx (i dont know where it should go).

i think there are good reasons to use a 'flat' architecture rather than oo (easier to learn, easier to modify, lighter) , but i dont think closeness to math syntax...

sweet thanks a lot. fyi, i am currently working on a [ `Frame` ](https://github.com/arsenovic/clifford/blob/master/clifford/__init__.py#L1639) object. its a light object, and i was looking for simple element-wise arithmetic to work, like...

i am re-opening. because clifford is numerical package, we need to make working with arrays easier, so that broadcasting works cleaner. ie i want something like this to work. from...

i have tried a few things with `__array_ufunc__`, and `__array_wrap__` , but the difficulty of this problem doesnt make sense to me. all that is needed is for `ndarray`'s to...

so, hugo's #89 implementation works great, and i dont want to beat a dead horse. but i dont understand why there is not an easy way to prevent numpy from...

eric this works great, thanks a lot. one last question. if i want the ndarray of objects to have attributes access of the elements, like a = np.rand(3)*e1 # returns...

thanks for the definitive reply!

also, seems like big arrays of a single type of object would be common for large data sets. these would require one map from scalars->MV, and then the array of...