python-mle
python-mle copied to clipboard
Implement convolution of PDFs
Theano implements fast convolution via FFT.
We could implement a Conv distribution that takes two distributions and represents their convolution.
You would use it like Conv(x, dist1, dist2). If either dist1 or dist2 don't contain the variable x, we should throw an error.
There may be distributions that we can convolute symbolically. If that's the case, Conv should automatically use the fastest approach.