SNAD is a project devoted to the anomaly detection problem in the Open Supernova Catalog. The results of this work can be found in Pruzhinskaya et al., 2019. Here, we present all the code and the data from the SNAD analysis.
data/
Lists of supernovae that pass our selection criteria, i.e. have 3 observations in each passband. These objects are used for interpolation/extrapolation. 3-day bin width is applied:
-
min3obs_B,R,I.csv
-
min3obs_g,r,i.csv
-
min3obs_g_pr,r_pr,i_pr.csv
Files that contain photometry in the range of [-20:100] days relative to the maximum in r/r' band of the SNe extrapolated light curves in gri, g'r'i', and BRI (transformed to gri) passbands. The results of the extrapolation were not checked by eye:
-
extrapol_-20.0_100.0_B,R,I_uncut.csv
-
extrapol_-20.0_100.0_g,r,i_uncut.csv
-
extrapol_-20.0_100.0_g_pr,r_pr,i_pr.csv
Files that contain photometry in the range of [-20:100] days relative to the maximum in r/r' band of the SNe extrapolated light curves in gri, g'r'i', and BRI (transformed to gri) passbands. The results of the extrapolation were checked by eye. These files are mainly used for the ML analysis:
-
extrapol_-20.0_100.0_B,R,I.csv
-
extrapol_-20.0_100.0_g,r,i.csv
-
extrapol_-20.0_100.0_g_pr,r_pr,i_pr.csv
tsne/ contains files with dimensionality-reduced data sets corresponding to 2 to 9 t-SNE features.
isolation_forests/ contains results of isolation forest algorithm run on 10 datasets:
-
weirdSN_isoforest_GPfit.dat
data set of 364 photometric characteristics (121×3 normalized fluxes, the LC flux maximum)
-
weirdSN_isoforest_GPparam.dat
data set of 10 parameters of the Gaussian process (9 fitted parameters of the kernel, the log-likelihood of the fit)
-
weirdSN_isoforest_tSNE_*.dat
8 data sets obtained by reducing 374 features to 2-9 t-SNE dimensions
fig/ contains plots of the supernova light curves in gri, g'r'i', and BRI passbands together with their Gaussian processes approximation.