search

ICSM / pgmuvi

Python gaussian processes for inference on multi-wavelength light curves
https://pgmuvi.readthedocs.io
GNU General Public License v3.0
12 stars 6 forks source link

Human-readable results output #8

Open pscicluna opened 1 year ago

pscicluna commented 1 year ago

pgmuvi should create an output table so that people can understand what the results are and what they mean.

This might involve two separate implementations for MAP and MCMC solutions, as credible intervals need to be reported in the latter case.

pscicluna commented 1 year ago

Partial solution for 1D lightcurves in https://github.com/ICSM/pgmuvi/pull/22

pscicluna commented 1 year ago

Testing on a 2D lightcurve results in the following error:

Traceback (most recent call last):
  File "/home/psciclun/Downloads/light curves-20221213T145300Z-001/light curves/pgmuvi2Dtest.py", line 42, in <module>
    lightcurve_2d.print_results()
  File "/home/psciclun/code/pythondev/pgmuvi/pgmuvi/lightcurve.py", line 395, in print_results
    print(f"{key}: {results_tmp[:,0].flatten()}, {results_tmp[:,1].flatten()}")
IndexError: index 1 is out of bounds for axis 1 with size 1

We will have to figure out what shape the results are in for each variable in 2D. It did successfully output the loss, delta loss, likelihood.noise_covar.raw_noise, mean_module.constant, and covar_module.raw_mixture_weights, which as expected are all 1D. It couldn't output the mixture means or scales.