اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدIdentifying Exoplanets with Deep Learning. IV. Removing Stellar Activity Signals from Radial Velocity Measurements Using Neural Networks
80
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Exoplanet detection with precise radial velocity (RV) observations is currently limited by spurious RV signals introduced by stellar activity. We show that machine learning techniques such as linear regression and neural networks can effectively remove the activity signals (due to starspots/faculae) from RV observations. Previous efforts focused on carefully filtering out activity signals in time using modeling techniques like Gaussian Process regression (e.g. Haywood et al. 2014). Instead, we systematically remove activity signals using only changes to the average shape of spectral lines, and no information about when the observations were collected. We trained our machine learning models on both simulated data (generated with the SOAP 2.0 software; Dumusque et al. 2014) and observations of the Sun from the HARPS-N Solar Telescope (Dumusque et al. 2015; Phillips et al. 2016; Collier Cameron et al. 2019). We find that these techniques can predict and remove stellar activity from both simulated data (improving RV scatter from 82 cm/s to 3 cm/s) and from more than 600 real observations taken nearly daily over three years with the HARPS-N Solar Telescope (improving the RV scatter from 1.47 m/s to 0.78 m/s, a factor of ~ 1.9 improvement). In the future, these or similar techniques could remove activity signals from observations of stars outside our solar system and eventually help detect habitable-zone Earth-mass exoplanets around Sun-like stars.
قيم البحث
اقرأ أيضاً
Context. Distinguishing between a signal induced by either stellar activity or a planet is currently the main challenge in radial velocity searches for low-mass exoplanets. Even when the presence of a transiting planet and hence its period are known,
stellar activity can be the main barrier to measuring the correct amplitude of the radial velocity signal. Several tools are being used to help understand which signals come from stellar activity in the data. Aims. We aim to present a new tool that can be used for the purpose of identifying periodicities caused by stellar activity, and show how it can be used to track the signal-to-noise ratio (SNR) of the detection over time. The tool is based on the principle that stellar activity signals are variable and incoherent. Methods. We calculate the Bayesian general Lomb-Scargle periodogram for subsets of data and by adding one extra data point we track what happens to the presence and significance of periodicities in the data. Publicly available datasets from HARPS and HARPS-N were used for this purpose. Additionally, we analysed a synthetic dataset that we created with SOAP2.0 to simulate pure stellar activity and a mixture of stellar activity and a planet. Results. We find that this tool can easily be used to identify unstable and incoherent signals, such as those introduced by stellar activity. The SNR of the detection grows approximately as the square root of the number of data points, in the case of a stable signal. This can then be used to make decisions on whether it is useful to keep observing a specific object. The tool is relatively fast and easy to use, and thus lends itself perfectly to a quick analysis of the data.
We present observations with the planet finder SPHERE of a selected sample of the most promising radial velocity (RV) companions for high-contrast imaging. Using a Monte Carlo simulation to explore all the possible inclinations of the orbit of wide R
V companions, we identified the systems with companions that could potentially be detected with SPHERE. We found the most favorable RV systems to observe are : HD,142, GJ,676, HD,39091, HIP,70849, and HD,30177 and carried out observations of these systems during SPHERE Guaranteed Time Observing (GTO). To reduce the intensity of the starlight and reveal faint companions, we used Principle Component Analysis (PCA) algorithms alongside angular and spectral differential imaging. We injected synthetic planets with known flux to evaluate the self-subtraction caused by our data reduction and to determine the 5$sigma$ contrast in the J band $vs$ separation for our reduced images. We estimated the upper limit on detectable companion mass around the selected stars from the contrast plot obtained from our data reduction. Although our observations enabled contrasts larger than 15 mag at a few tenths of arcsec from the host stars, we detected no planets. However, we were able to set upper mass limits around the stars using AMES-COND evolutionary models. We can exclude the presence of companions more massive than 25-28 MJup around these stars, confirming the substellar nature of these RV companions.
Precision radial velocity (RV) measurements in the near-infrared are a powerful tool to detect and characterize exoplanets around low-mass stars or young stars with higher magnetic activity. However, the presence of strong telluric absorption lines a
nd emission lines in the near infrared that significantly vary in time can prevent extraction of RV information from these spectra by classical techniques, which ignore or mask the telluric lines. We present a methodology and pipeline to derive precision RVs from near-infrared spectra using a forward-modeling technique. We applied this to spectra with a wide wavelength coverage (Y, J, and H bands, simultaneously), taken by the InfraRed Doppler (IRD) spectrograph on the Subaru 8.2-m telescope. Our pipeline extracts the instantaneous instrumental profile of the spectrograph for each spectral segment, based on a reference spectrum of the laser-frequency comb that is injected into the spectrograph simultaneously with the stellar light. These profiles are used to derive the intrinsic stellar template spectrum, which is free from instrumental broadening and telluric features, as well as model and fit individual observed spectra in the RV analysis. Implementing a series of numerical simulations using theoretical spectra that mimic IRD data, we test the pipeline and show that IRD can achieve $<2$ m s$^{-1}$ precision for slowly rotating mid-to-late M dwarfs with a signal-to-noise ratio $> 100$ per pixel at 1000 nm. Dependences of RV precision on various stellar parameters (e.g., $T_{rm eff}$, $vsin i$, [Fe/H]) and the impact of telluric-line blendings on the RV accuracy are discussed through the mock spectra analyses. We also apply the RV-analysis pipeline to the observed spectra of GJ 699 and TRAPPIST-1, demonstrating that the spectrograph and the pipeline are capable of an RV accuracy of $<3$ m s$^{-1}$ at least on a time scale of a few months.
In the context of large spectroscopic surveys of stars, data-driven methods are key in deducing physical parameters for millions of spectra in a short time. Convolutional neural networks (CNNs) enable us to connect observables (e.g. spectra, stellar
magnitudes) to physical properties (atmospheric parameters, chemical abundances, or labels in general). We trained a CNN, adopting stellar atmospheric parameters and chemical abundances from APOGEE DR16 (resolution R=22500) data as training set labels. As input, we used parts of the intermediate-resolution RAVE DR6 spectra (R~7500) overlapping with the APOGEE DR16 data as well as broad-band ALL_WISE and 2MASS photometry, together with Gaia DR2 photometry and parallaxes. We derived precise atmospheric parameters Teff, log(g), and [M/H] along with the chemical abundances of [Fe/H], [alpha/M], [Mg/Fe], [Si/Fe], [Al/Fe], and [Ni/Fe] for 420165 RAVE spectra. The precision typically amounts to 60K in Teff, 0.06 in log(g) and 0.02-0.04 dex for individual chemical abundances. Incorporating photometry and astrometry as additional constraints substantially improves the results in terms of the accuracy and precision of the derived labels. We provide a catalogue of CNN-trained atmospheric parameters and abundances along with their uncertainties for 420165 stars in the RAVE survey. CNN-based methods provide a powerful way to combine spectroscopic, photometric, and astrometric data without the need to apply any priors in the form of stellar evolutionary models. The developed procedure can extend the scientific output of RAVE spectra beyond DR6 to ongoing and planned surveys such as Gaia RVS, 4MOST, and WEAVE. We call on the community to place a particular collective emphasis and on efforts to create unbiased training samples for such future spectroscopic surveys.
To date, only 18 exoplanets with radial velocity (RV) semi-amplitudes $<2$ m/s have had their masses directly constrained. The biggest obstacle to RV detection of such exoplanets is variability intrinsic to stars themselves, e.g. nuisance signals ari
sing from surface magnetic activity such as rotating spots and plages, which can drown out or even mimic planetary RV signals. We use Kepler-37 - known to host three transiting planets, one of which, Kepler-37d, should be on the cusp of RV detectability with modern spectrographs - as a case study in disentangling planetary and stellar activity signals. We show how two different statistical techniques - one seeking to identify activity signals in stellar spectra, and another to model activity signals in extracted RVs and activity indicators - can enable detection of the hitherto elusive Kepler-37d. Moreover, we show that these two approaches can be complementary, and in combination, facilitate a definitive detection and precise characterisation of Kepler-37d. Its RV semi-amplitude of $1.22pm0.31$ m/s (mass $5.4pm1.4$ $M_oplus$) is formally consistent with TOI-178bs $1.05^{+0.25}_{-0.30}$ m/s, the latter being the smallest detected RV signal of any transiting planet to date, though dynamical simulations suggest Kepler-37ds mass may be on the lower end of our $1sigma$ credible interval. Its consequent density is consistent with either a water-world or that of a gaseous envelope ($sim0.4%$ by mass) surrounding a rocky core. Based on RV modelling and a re-analysis of Kepler-37 TTVs, we also argue that the putative (non-transiting) planet Kepler-37e should probably be stripped of its confirmed status.
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
