# Finding the Needle in a Haystack: Detrending Photometric Timeseries Data of Strictly Periodic Astrophysical Objects

## TL;DR

This paper introduces a novel detrending algorithm for astrophysical light curves with strictly periodic signals, effectively disentangling signals from baseline noise by modeling the periodic component as a probability distribution.

## Contribution

The paper presents a new method leveraging periodicity to improve detrending of astrophysical time series, with an open-source implementation available.

## Key findings

- Successfully applied to synthetic and real data cases.
- Effectively disentangles periodic signals from baseline noise.
- Provides insights into instrumental response dependence.

## Abstract

Light curves of astrophysical objects frequently contain strictly periodic signals. In those cases we can use that property to aid the detrending algorithm to fully disentangle an unknown periodic signal and an unknown baseline signal with no power at that period. The periodic signal is modeled as a discrete probability distribution function (pdf), while the baseline signal is modeled as a residual timeseries. Those two components are disentangled by minimizing the length of the residual timeseries w.r.t. the per-bin pdf fluxes. We demonstrate the use of the algorithm on a synthetic case, on the eclipsing binary KIC 3953981 and on the eccentric ellipsoidal variable KIC 3547874. We further discuss the parameters and the limitations of the algorithm and speculate on the two most common use cases: detrending the periodic signal of interest and measuring the dependence of instrumental response on controlled instrumental variables. A more sophisticated version of the algorithm is released as open source on github and available via pip.