How to fix likelihood hyperparameter (observation noise variance)

[chaozg]

Hi there,

Is it possible to fix the observation noise variance (i.e., the likelihood noise hyperparameter) to a constant value, rather than allowing it to be optimized during training?

In my use case, I have prior knowledge about the noise level in the observations, and I’d prefer to keep this parameter fixed to ensure consistency across experiments.

This post is potentially related to #481, though I didn't see any replies there.

Regards,
Charlie

[thomaspinder]

Hey! Yes, you can do it by the following

import gpjax as gpx
from gpjax.parameters import Static

sigma_true = 0.2
likelihood = gpx.likelihoods.Gaussian(num_datapoints=n, obs_stddev=Static(sigma_true))

[chaozg]

Thanks very much @thomaspinder!