Matrix Multiplication With Multiple Dimensions in PYMC Model

[ricardoV94]

@kiernan9811

Out of interest could you help me understand why sampling mean[:, product_data] fails? The reason for my confusion is this is similar to the syntax used to slice tensors in tensorflow. For PYMC are tensors held a little differently? my intuition would be we have a tensor generated here for each input, so we need to take the respective tensors then select the product value internally from that tensor, is that anywhere near reality?

Indexing in Aesara works just like Numpy (sans bugs):

import numpy as np

x = np.arange(10).reshape((5, 2))
print(x[:, [0, 1, 0, 1, 0]].shape)  # (5, 5)
print(x[[0, 1, 2, 3, 4], [0, 1, 0, 1, 0]].shape)  # (5,)