search

mackelab / delfi

Density estimation likelihood-free inference. No longer actively developed see https://github.com/mackelab/sbi instead
http://www.mackelab.org/delfi
Other
72 stars 28 forks source link

Two moons - discrepancy with APT/SNPE-C paper #72

Closed MaverickMeerkat closed 2 years ago

MaverickMeerkat commented 2 years ago

https://github.com/mackelab/delfi/blob/62a99a879145bdc675917fc33eed69293b964048/delfi/simulator/TwoMoons.py#L115

Inverting equation (26) in the paper "Automatic Posterior Transformation for Likelihood-free Inference", we get: p_0 \frac{|\theta_0 + \theta_1|}{\sqrt2}=x_0 p_1 + \frac{-\theta_0 + \theta_1}{\sqrt2}=x_1

Moving sides and defining the rhs as q's: \frac{|\theta_0 + \theta_1|}{\sqrt2}=p_0-x_0:=q_0 \frac{-\theta_0 + \theta_1}{\sqrt2}=x_1-p_1:=q_1 \frac{\theta_0 + \theta_1}{\sqrt2}= \pm q_0

subtracting/adding to cancel the variables we get: \sqrt2 \theta_1=\pm q_0 + q_1 \sqrt2\theta_0=\pm q_0 - q_1

Now c is simply 1/\sqrt2, and s is -1/\sqrt2. So it comes out that the code is putting theta1 in theta0, and putting -theta0 in theta1.

MaverickMeerkat commented 2 years ago

Looking back on this, I think I missed the fact that the angle has a minus in front of it - so s=c=1/sqrt(2) and there's no problem in the code. (though it's a bit cumbersome with all the s, c and angles...)

I remove this issue.