AssertionError in PyMC3 code using Deterministic distributions

The original function sampleFromSalpeterDATA itself is quite slow to evaluate (in python3 at least), so it is not very surprising that it is even slower when wrapped in theano.as_op.

The best way forward I think, it’s to rewrite sampleFromSalpeter into a vectorized numpy function (without the while loop), and then extend it into a theano function.