Constraining Uniform distribution range by another random variable

Questions v3
1

Hi!

At this point this I think this should be easy to implement but it is not working. How can I condition the lower value of a uniform distribution using another pm variable? I am trying this, but it never starts to sample,

ranges = {}
ranges['U1'] = 4, 40
ranges['U2'] = 4, 40

with pm.Model() as model:
    u1_ = pm.Uniform('ep1', ranges['U1'][0], ranges['U1'][1])
    u2_ = pm.Uniform('ep2', u1_, ranges['U2'][1])

I am using SMC since multimodal posteriors are expected.
Any feedback is welcome.

2

Samples fine for me. What version of pymc and aesara are you using and what platform are you on?

3

Thank you for the feedback. I am on Ubuntu 16.04 LTS PyMC3 3.11.4. I am trying to replicate the effect using pm.Potential with halfNormal

4

Try V4, we did a lot of improvements to SMC since the version you are using.

1 Like
5

Thank you I will try it