Python numpy 'same' convolution different from Matlab's?

Question

I\'m trying to replicate Matlab\'s convolution aka conv function.

Matlab\'s conv works like this:

w = conv(u,v,shape) returns a subsection of the convoluti