ola bykowska

Thanks for your reply and sorry, I should have described this in the initial post. The `calculate_probability` function is a custom function that takes the observed data, contained in `experiment`...

I had a go at using `as_op`, this is what I had in the end: @t.compile.ops.as_op(itypes=[t.dscalar, t.dscalar, t.lscalar,t.dscalar,t.dscalar,t.dscalar,t.dmatrix ],otypes=[t.dscalar]) def test(parameter1,parameter2,parameter3,parameter4,parameter5,parameter6,data): log_l=calculate_probability(parameters, np.diff(experiment.spike_t), 0.001, 100, parameter4, parameter5, parameter6, data) return...