Added seeding to draws in tests #195
Conversation
Added seeding to tests that draw from random variables and check the values drawn.
| with pm.Model() as m_old: | ||
| x = pm.Normal("x", 0, 1e-3) | ||
| y = pm.Normal("y", x, 1e-3) | ||
| z = pm.Normal("z", y + x, 1e-3) | ||
|
|
||
| assert -5 < pm.draw(z) < 5 | ||
| assert -5 < pm.draw(z, random_seed=rng) < 5 |
There was a problem hiding this comment.
These ones didn't need seeding but no harm I guess. They were selected to be many standard deviations away from the constraint
There was a problem hiding this comment.
I noticed that, but in my recent and limited experience with the project I observed that random test failure gets flagged as a flaky test issue, so I thought it would be better to seed. Do you have a sense of what probability of test failure the project would accept?
There was a problem hiding this comment.
Those are the sum of 3 Normals each with 1e-3 standard deviation, so to get out of the (-5, 5) range, it should take a couple of universe lifetimes :)
There was a problem hiding this comment.
Haha, yes, in this case the chance is very small. For my own learning, I just wondered where (approximately) you might draw the line.
Fixes #192
Added seeding to tests that draw from random variables and check the values drawn.