Index error when number of samples is not a multiple of 10

[kdolum]

I say sampler.sample(x0, 128, SCAMweight=30, AMweight=15, DEweight=50) and I get IndexError: index 12 is out of bounds for axis 0 with size 12. 12 is the quotient of 128 over 10. It seems to be completely repeatable, giving an error for any number of samples that is not a multiple of 10.

[paulthebaker]

It looks like the issue comes from when the chain arrays get initialized at PTMCMCSampler.py#L182. The default thin is 10. The issue occurs whenever Niter/thin is not an integer ratio. For instance

sampler.sample(p0, 128, thin=8)

works correctly. It looks like it would be solved by justing adding 1 to N.

[paulthebaker]

Adding 1 to N causes an extra unused index at the end of the arrays with Niter/burn is an integer ratio. It seems the best thing to do is round:

N = int(np.ceil(maxIter / thin))

[kdolum]

Fixed by #41.