Closed anaismoller closed 1 year ago
What is the status of this Issue? It sounds like it might be a reasonable first contribution fix. If so, can you point me to the relevant area of Pippin and I can see what I can contribute?
Thanks
Hi @benjaminrose, Haven't had time to work on this, so if you wanted to give it a shot that would be great!
The relevant file is pippin/classifiers/supernnova.py
. In the __init__
method you can add new parameters to the class such as self.filters = options.get("FILTERS", ['g', 'r', 'i', 'z'])
or something like that. Your choice of key will be what key is used in the pippin input file. Next you can add the new options to the setup_dict
(approximately line 270). Finally in pippin/tasks/supernnova
add the options to the script command.
Resolved.
SuperNNova requires that information on the simulations used for training. Some values are default.
Can we please update Pippin to retrieve parameters from simulations/config file for this? I see an option of using a SuperNNova customised yml but not as part of the general Pippin config. This is due to a bug trying to classify Roman sims (see Ben Rose's work)
As in https://github.com/supernnova/SuperNNova/tree/master/configs_yml we would need to add sntypes list_filters
For Roman we would add while running --list_filters Y J H F --sntypes '{"110":"Ia"}' and others from the generated simulation from the Plasticc configs https://github.com/LSSTDESC/plasticc_alerts/blob/main/Examples/plasticc_schema/elasticc_origmap.txt
An example config file is in $WFIRST/USERS/brose3/transient_catalog/train_classifiers/BR-ROMAN-CLAS-TRAIN.yml
(these changes could help have an easy use of SuperNNova with other surveys/simulations using Pippin)
Thanks heaps!