Adding own supernova and galaxy templates

[steveschulze]

Hi Daniel,

Thanks for developing such a powerful software package. In issue #16 you mention a future feature to add your own SN templates to the library. Is this feature added? If yes, Is it possible to define my own class of transients?

A related question, if I would have the spectrum/SED of the host galaxy of a particular supernova, can this be used instead of the generic galaxy templates?

Cheers,

Steve

[daniel-muthukrishna]

Hi Steve,

Thanks for the feedback and for submitting an issue. While it is possible to retrain on your own data, I haven't yet made this easy to do within the python library yet. I have recently received some requests to add SLSNe and Kilonovae as additional classes and am planning to add that while also completing issue #16 within the next few weeks to a month.

You can certainly add your own host spectra to the training set too. I'll be sure to make this an option.

If there is some urgency before the next month, you can add/remove spectrum files to the astrodash/templates/training_set/ directory and the templist.txt, as well as the astrodash/templates/superfit_templates/gal/ for the host galaxy spectra, and then carefully run the create_and_save_all_data_files.py module to retrain the classifier on your new training set. I suspect there may be some trivial issues in doing this, however. Feel free to let me know issues you have. I'll keep you updated when a new version with a customizable training set is available in any case.

Cheers,

Daniel