Open mkhattab940 opened 6 years ago
The expected is wrong here, at least for failure 0, the 'got' is indeed what I expected.
Failure 1 is weird - I don't understand how a plate of normals becomes two gammas when inlining.
OK, fixed the 0-test. Failure looks like this now:
### Failure in: 6:RoundTrip:0:6:t_laplace_to_stdChiSq:1
haskell/Tests/TestTools.hs:130
expected:
X1 <~ normal(+0/1, 1/1)
X2 <~ normal(+0/1, 1/1)
return real2prob(X1 ^ 2 + X2 ^ 2)
but got:
X5 <~ gamma(1/1, 1/1)
Y3 <~ gamma(1/1, 1/1)
return abs(prob2real(X5) + prob2real(Y3) * (-1/1) + (-1/1)) * (2/1)
Cases: 172 Tried: 97 Errors: 0 Failures: 11
Cases: 172 Tried: 98 Errors: 0 Failures: 11
This one slipped through the cracks. Here's the failure: