search

LLNL / MuyGPyS

A fast, pure python implementation of the MuyGPs Gaussian process realization and training algorithm.
Other
23 stars 11 forks source link

`fast_posterior_mean` needs to be rephrased as a composition #106

Closed bwpriest closed 1 year ago

bwpriest commented 1 year ago

We need the fast posterior mean to be rewritten as a composition, similar to how the posterior mean function works. This should be in the form of a functor class similar in form to PosteriorMean less the optimization members (since the fast mean is not involved in optimization).

bwpriest commented 1 year ago

Same thing goes for build_fast_regress_coeffs.

bwpriest commented 1 year ago

Same thing goes for build_fast_regress_coeffs.

Also, we need to change build_fast_posterior_mean_coeffs to take pairwise_diffs_fast and train_nn_targets_fast to be consistent with the other MuyGPS functions. Event though that function is not supported in distributed memory, we want to keep the control flow as similar as possible for all workflows. So, we should always expect something like:

(raw data, indices) -> tensors -> MuyGPS functions

Done

bwpriest commented 1 year ago

build_fast_posterior_mean_coeffs is a lot of characters and involves an abbreviation (which I have been trying to avoid). It would be nice to rename it to something more concise, along the lines of fast_coefficients or something more descriptive of your choice.

Done

bwpriest commented 1 year ago

PR #107 makes this a little easier to coalescing the noise model application to a single function call

alecmdunton commented 1 year ago

build_fast_posterior_mean_coeffs is a lot of characters and involves an abbreviation (which I have been trying to avoid). It would be nice to rename it to something more concise, along the lines of fast_coefficients or something more descriptive of your choice.

Done

alecmdunton commented 1 year ago

Done.