Noise learning bug in GP #58
Comments
bors bot
added a commit
that referenced
this issue
Jul 21, 2020
63: WIP: add SVD functionality r=odunbar a=odunbar Resolves #57 . We add the ability for the Gaussian process to learn non - diagonal covariance matrix for the observational noise by applying and SVD. Resolve #58 We add functionality to toggle whether we wish to learn the observational noise. and add (mathematically correct) default values when we are not. We also have 2 examples: - [x] Gaussian process plot. We train a GP on some training points in 2D space, and plot the mean and variance (compared with the underlying model and observational noise) - in the untransformed space - [x] Noise learning test. We train a GP with known noise, with `learn_noise = false` and with `learn_noise = true` we compare the learned `WhiteKernel` parameters to the true unlearned parameters. Coauthored with @bielim Co-authored-by: odunbar <odunbar@caltech.edu> Co-authored-by: Melanie Bieli <melanie.bieli@bluewin.ch> Co-authored-by: bielim <bielim@users.noreply.github.com> Co-authored-by: odunbar <47412152+odunbar@users.noreply.github.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
In some recent testing there appeared to be some bugs in the white noise kernel in
GPemulator.CalibrateEmulateSample.jl/src/GPEmulator.jl
Line 111 in 5fb71c2
Firstly, if users know the noise structure completely, they can provide the covariance matrix as an optional argument to
GPEmulator.Secondly, if the noise structure is unknown, then the white noise kernel should be used to learn the noise.
@ilopezgp has provided a simple working example where learning the observational noise fails even in a simple diagonal case.
The text was updated successfully, but these errors were encountered: