Astraea: Predicting Long Rotation Periods with 27 Day Light Curves

Lu, Yuxi; Angus, Ruth; Agüeros, Marcel A.; Blancato, Kirsten; Ness, Melissa; Rowland, Danielle; Curtis, Jason L.; Grunblatt, Samuel K.

doi:10.3847/1538-3881/abada4

Cited by 20 publications

(39 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the tests with 128 and 150 days, the neural network had some success predicting periods longer than the maximum visible period in the periodogram, even for the noisy data. This suggests that neural networks can predict periods even when the period at maximum power is beyond the range of the plot, consistent with the results of Lu et al (2020). This is encouraging for period predictions for stars outside the TESS continuous viewing zones, where observations are substantially less than a year in duration.…”

Section: Data Processing/wavelet Transformsupporting

confidence: 81%

“…They obtained unambiguous rotation periods for 131 stars, but all were shorter than the 13.7-days TESS orbital period. Lu et al (2020) trained a random forest (RF) regressor to predict rotation periods from 27-day sections of Kepler light curves coupled with Gaia stellar parameters. They then evaluated the trained model on single sectors of TESS data for the same stars.…”

Section: Comparisons To Other Period Recovery Attemptsmentioning

confidence: 99%

“…Furthermore, while there is some overlap between TESS and Kepler, the TESS observations of the Kepler field are short, spanning only 27 days. This is enough to recover and validate short rotation periods (Blancato et al 2020), and possibly some subset of longer periods (Lu et al 2020), but not enough to be useful for the broader population of stars in our Galaxy. Moreover, even large rotation samples from Kepler are likely contaminated with mismeasured periods (Aigrain et al 2015).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Recovery of TESS Stellar Rotation Periods Using Deep Learning

Claytor,

van Saders,

Llama

et al. 2021

Preprint

View full text Add to dashboard Cite

We used a convolutional neural network to infer stellar rotation periods from a set of synthetic light curves simulated with realistic spot evolution patterns. We convolved these simulated light curves with real TESS light curves containing minimal intrinsic astrophysical variability to allow the network to learn TESS systematics and estimate rotation periods despite them. In addition to periods, we predict uncertainties via heteroskedastic regression to estimate the credibility of the period predictions. In the most credible half of the test data, we recover 10%-accurate periods for 46% of the targets, and 20%accurate periods for 69% of the targets. Using our trained network, we successfully recover periods of real stars with literature rotation measurements, even past the 13.7-day limit generally encountered by TESS rotation searches using conventional period-finding techniques. Our method also demonstrates resistance to half-period aliases. We present the neural network and simulated training data, and introduce the software butterpy used to synthesize the light curves using realistic star spot evolution.

show abstract

Section: Data Processing/wavelet Transformsupporting

confidence: 81%

Section: Comparisons To Other Period Recovery Attemptsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Recovery of TESS Stellar Rotation Periods Using Deep Learning

Claytor,

van Saders,

Llama

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…This is consistent with TOI-1634 being 26 days. Furthermore, we cannot hope to obtain a precise P rot measurement with just one TESS sector if indeed P rot is as long as we expect (Lu et al 2020).…”

Section: Archival Photometric Monitoringmentioning

confidence: 99%

TOI-1634 b: An Ultra-short-period Keystone Planet Sitting inside the M-dwarf Radius Valley

Cloutier

Charbonneau

Stassun

et al. 2021

View full text Add to dashboard Cite

show abstract

“…In asteroseismology, RF algorithms have already been used to estimate stellar surface gravities (log g, Bugnet et al 2018) and to automatically recognise solar-like pulsators (Bugnet et al 2019). They have also recently been used to perform analyses linked to surface stellar rotation, but only in order to infer long rotation periods from TESS 27-day-long light curves (Lu et al 2020). In this work, we present the Random fOrest Over STEllar Rotation (ROOSTER), which is designed to select a rotation period for stars observed by Kepler through a combination of RF classifiers applied to a variety of methods used to extract P rot and different ways to correct the light curves.…”

Section: Introductionmentioning

confidence: 99%

ROOSTER: a machine-learning analysis tool for Kepler stellar rotation periods

Breton

Santos

Bugnet

et al. 2021

A&A

View full text Add to dashboard Cite

In order to understand stellar evolution, it is crucial to efficiently determine stellar surface rotation periods. Indeed, while they are of great importance in stellar models, angular momentum transport processes inside stars are still poorly understood today. Surface rotation, which is linked to the age of the star, is one of the constraints needed to improve the way those processes are modelled. Statistics of the surface rotation periods for a large sample of stars of different spectral types are thus necessary. An efficient tool to automatically determine reliable rotation periods is needed when dealing with large samples of stellar photometric datasets. The objective of this work is to develop such a tool. For this purpose, machine learning classifiers constitute relevant bases to build our new methodology. Random forest learning abilities are exploited to automate the extraction of rotation periods in Kepler light curves. Rotation periods and complementary parameters are obtained via three different methods: a wavelet analysis, the autocorrelation function of the light curve, and the composite spectrum. We trained three different classifiers: one to detect if rotational modulations are present in the light curve, one to flag close binary or classical pulsators candidates that can bias our rotation period determination, and finally one classifier to provide the final rotation period. We tested our machine learning pipeline on 23 431 stars of the Kepler K and M dwarf reference rotation catalogue for which 60% of the stars have been visually inspected. For the sample of 21 707 stars where all the input parameters are provided to the algorithm, 94.2% of them are correctly classified (as rotating or not). Among the stars that have a rotation period in the reference catalogue, the machine learning provides a period that agrees within 10% of the reference value for 95.3% of the stars. Moreover, the yield of correct rotation periods is raised to 99.5% after visually inspecting 25.2% of the stars. Over the two main analysis steps, rotation classification and period selection, the pipeline yields a global agreement with the reference values of 92.1% and 96.9% before and after visual inspection. Random forest classifiers are efficient tools to determine reliable rotation periods in large samples of stars. The methodology presented here could be easily adapted to extract surface rotation periods for stars with different spectral types or observed by other instruments such as K2, TESS or by PLATO in the near future.

show abstract

Astraea: Predicting Long Rotation Periods with 27 Day Light Curves

Cited by 20 publications

References 57 publications

Recovery of TESS Stellar Rotation Periods Using Deep Learning

Recovery of TESS Stellar Rotation Periods Using Deep Learning

TOI-1634 b: An Ultra-short-period Keystone Planet Sitting inside the M-dwarf Radius Valley

ROOSTER: a machine-learning analysis tool for Kepler stellar rotation periods

Contact Info

Product

Resources

About