search

tapios / risk-networks

Code for risk networks: a blend of compartmental models, graphs, data assimilation and semi-supervised learning
Other
2 stars 2 forks source link

Restrict DA of model parameter to only nodes of interest #154

Closed odunbar closed 4 years ago

odunbar commented 4 years ago

This PR primarily restricts the learning of nodally defined model parameters to only over nodes that are observed. Prior to this if we have a nodally defined model parameter array, we would feed the entire node-length array into the assimilator ( even if only a few nodes were observed).

I have also pulled out the getting and setting of model parameters into new functions get_model_parameters_to_update(...) set_updated_model_parameters(...) so that DataAssimilator.update(...) is a cleaner function.