IDEA 5 Repost

IDEA 5: Detection of Periodic and Aperiodic phenomena in a large catalog
Hi All,

Here is a challenge I'd like to propose for teams to tackle in the hackathon :

Background:

Various authors (Graham 2012, 2013, Feigelson etal. 2018) have shown periodic and aperiodic phenomena are sensitive to different algorithmic detection and there are potentially many different models which are each best suited to find small classes/types of signals. For example traditional Fourier analysis works well with periodic, sinusoidal waveforms but performs poorly for other types of signals. With the advent of large survey telescopes like TESS and LSST it will become a pressing problem as to how we will be able to classify and uncover as much of the observed variability in the sky as possible.

Problem:

Pioneer a means to find as efficiently as possible (in compute time) as many periodic and aperiodic phenomena from a large catalog of point source detections as possible.

Dataset(s) to use:

Any large dataset which contains lightcurves. Some possible sources of data include:

ASAS Variable Star Catalog : https://asas-sn.osu.edu/variables (I will bring a thumb drive with these data)
Kepler Catalog : http://archive.stsci.edu/kepler/data_search/search.php
Catalina Survey : http://nesssi.cacr.caltech.edu/DataRelease/

References

A comparison of period finding algorithms
https://ui.adsabs.harvard.edu/#abs/2013MNRAS.434.3423G/abstract

Data challenges of time domain astronomy
https://ui.adsabs.harvard.edu/#abs/2012arXiv1208.2480G/abstract
Autoregressive Times Series Methods for Time Domain Astronomy
https://doi.org/10.3389/fphy.2018.00080