Atmospheric retrievals on exoplanets usually involve computationally intensive Bayesian sampling methods. Large parameter spaces and increasingly complex atmospheric models create a computational ...bottleneck forcing a trade-off between statistical sampling accuracy and model complexity. It is especially true for upcoming JWST and ARIEL observations. We introduce ExoGAN, the Exoplanet Generative Adversarial Network, a new deep-learning algorithm able to recognize molecular features, atmospheric trace-gas abundances, and planetary parameters using unsupervised learning. Once trained, ExoGAN is widely applicable to a large number of instruments and planetary types. The ExoGAN retrievals constitute a significant speed improvement over traditional retrievals and can be used either as a final atmospheric analysis or provide prior constraints to subsequent retrieval.
Molecular line lists (a catalogue of transition frequencies and line strengths) are important for modelling absorption and emission processes in atmospheres of different astronomical objects, such as ...cool stars and exoplanets. In order to be applicable for high temperatures, line lists for molecules like methane must contain billions of transitions, which makes their direct (line-by-line usage) application in radiative transfer calculations impracticable. Here we suggest a new, hybrid line list format to mitigate this problem, based on the idea of temperature-dependent absorption continuum.
The line list is partitioned into a large set of relatively weak lines and a small set of important, stronger lines. The weaker lines are then used either to construct a temperature-dependent (but pressure-independent) set of intensity cross sections or are blended into a greatly reduced set of 'super-lines'. The strong lines are kept in the form of temperature-independent Einstein
coefficients.
A line list for methane (CH
) is constructed as a combination of 17 million strong absorption lines relative to the reference absorption spectra and a background methane continuum in two temperature-dependent forms of cross sections and super-lines. This approach significantly eases the use of large high temperature line lists as the computationally expensive calculation of pressure- dependent profiles (e.g. Voigt) only need to be performed for a relatively small number of lines. Both the line list and cross sections were generated using a new 34 billion methane line list (known as 34to10), which extends the 10to10 line list to higher temperatures (up to 2000 K). The new hybrid scheme can be applied to any large line lists containing billions of transitions. We recommend using super-lines generated on a high resolution grid based on a resolving power of
= 1,000,000 to model the molecular continuum as a more flexible alternative to the temperature-dependent cross sections.
The James Webb Space Telescope (JWST) is expected to revolutionize the field of exoplanets. The broad wavelength coverage and the high sensitivity of its instruments will allow characterization of ...exoplanetary atmospheres with unprecedented precision. Following the Call for the Cycle 1 Early Release Science Program, the Transiting Exoplanet Community was awarded time to observe several targets, including WASP-43b. The atmosphere of this hot Jupiter has been intensively observed but still harbors some mysteries, especially concerning the day-night temperature gradient, the efficiency of the atmospheric circulation, and the presence of nightside clouds. We will constrain these properties by observing a full orbit of the planet and extracting its spectroscopic phase curve in the 5-12 m range with JWST/MIRI. To prepare for these observations, we performed extensive modeling work with various codes: radiative transfer, chemical kinetics, cloud microphysics, global circulation models, JWST simulators, and spectral retrieval. Our JWST simulations show that we should achieve a precision of 210 ppm per 0.1 m spectral bin on average, which will allow us to measure the variations of the spectrum in longitude and measure the nightside emission spectrum for the first time. If the atmosphere of WASP-43b is clear, our observations will permit us to determine if its atmosphere has an equilibrium or disequilibrium chemical composition, eventually providing the first conclusive evidence of chemical quenching in a hot Jupiter atmosphere. If the atmosphere is cloudy, a careful retrieval analysis will allow us to identify the cloud composition.
ABSTRACT
Over the last several years, spectroscopic observations of transiting exoplanets have begun to uncover information about their atmospheres, including atmospheric composition and indications ...of the presence of clouds and hazes. Spectral retrieval is the leading technique for interpretation of transmission spectra and is employed by several teams using a variety of forward models and parameter estimation algorithms. However, different model suites have mostly been used in isolation and so it is unknown whether the results from each are comparable. As we approach the launch of the James Webb Space Telescope, we anticipate advances in wavelength coverage, precision, and resolution of transit spectroscopic data, so it is important that the tools that will be used to interpret these information-rich spectra are validated. To this end, we present an intermodel comparison of three retrieval suites: TauREx, nemesis, and chimera. We demonstrate that the forward model spectra are in good agreement (residual deviations on the order of 20–40 ppm), and discuss the results of cross-retrievals among the three tools. Generally, the constraints from the cross-retrievals are consistent with each other and with input values to within 1σ. However, for high precision scenarios with error envelopes of order 30 ppm, subtle differences in the simulated spectra result in discrepancies between the different retrieval suites, and inaccuracies in retrieved values of several σ. This can be considered analogous to substantial systematic/astrophysical noise in a real observation, or errors/omissions in a forward model such as molecular line list incompleteness or missing absorbers.
Abstract
Current endeavours in exoplanet characterization rely on atmospheric retrieval to quantify crucial physical properties of remote exoplanets from observations. However, the scalability and ...efficiency of said technique are under strain with increasing spectroscopic resolution and forward model complexity. The situation has become more acute with the recent launch of the James Webb Space Telescope and other upcoming missions. Recent advances in machine learning provide optimization-based variational inference as an alternative approach to perform approximate Bayesian posterior inference. In this investigation we developed a normalizing-flow-based neural network, combined with our newly developed differentiable forward model,
Diff
-
τ
, to perform Bayesian inference in the context of atmospheric retrievals. Using examples from real and simulated spectroscopic data, we demonstrate the advantages of our proposed framework: (1) training our neural network does not require a large precomputed training set and can be trained with only a single observation; (2) it produces high-fidelity posterior distributions in excellent agreement with sampling-based retrievals; (3) it requires up to 75% fewer forward model calls to converge to the same result; and (4) this approach allows formal Bayesian model selection. We discuss the computational efficiencies of
Diff
-
τ
in relation to
TauREx3
's nominal forward model and provide a “lessons learned” account of developing radiative transfer models in differentiable languages. Our proposed framework contributes toward the latest development of neural network–powered atmospheric retrieval. Its flexibility and significant reduction in forward model calls required for convergence holds the potential to be an important addition to the retrieval tool box for large and complex data sets along with sampling-based approaches.
Aims. Molecular line lists (catalogues of transition frequencies and line strengths) are important for modelling absorption and emission processes in atmospheres of different astronomical objects, ...such as cool stars and exoplanets. In order to be applicable for high temperatures, line lists for molecules like methane must contain billions of transitions, which makes their direct (line-by-line usage) application in radiative transfer calculations impracticable. Here we suggest a new, hybrid line list format to mitigate this problem, based on the idea of temperature-dependent absorption continuum. Methods. The line list is partitioned into a large set of relatively weak lines and a small set of important, stronger lines. The weaker lines are then used either to construct a temperature-dependent (but pressure-independent) set of intensity cross sections or are blended into a greatly reduced set of “super-lines”. The strong lines are kept in the form of temperature-independent Einstein A coefficients. Results. A line list for methane (CH4) is constructed as a combination of 17 million strong absorption lines relative to the reference absorption spectra and a background methane continuum in two temperature-dependent forms of cross sections and super-lines. This approach significantly eases the use of large high temperature line lists as the computationally expensive calculation of pressure-dependent profiles (e.g. Voigt) only need to be performed for a relatively small number of lines. Both the line list and cross sections were generated using a new 34 billion methane line list (known as 34to10), which extends the 10to10 line list to higher temperatures (up to 2000 K). The new hybrid scheme can be applied to any large line lists containing billions of transitions. We recommend using super-lines generated on a high resolution grid based on a resolving power of R = 1 000 000 to model the molecular continuum as a more flexible alternative to the temperature-dependent cross sections.
Abstract
The study of exoplanetary atmospheres relies on detecting minute changes in the transit depth at different wavelengths. To date, a number of ground- and space-based instruments have been ...used to obtain transmission spectra of exoplanets in different spectral bands. One common practice is to combine observations from different instruments in order to achieve a broader wavelength coverage. We present here two inconsistent observations of WASP-96 b, one by the Hubble Space Telescope (HST) and the other by the Very Large Telescope (VLT). We present two key findings in our investigation: (1) a strong water signature is detected via the HST WFC3 observations and (2) a notable offset in transit depth (>1100 ppm) can be seen when the ground-based and space-based observations are combined. The discrepancy raises the question of whether observations from different instruments could indeed be combined. We attempt to align the observations by including an additional parameter in our retrieval studies but are unable to definitively ascertain that the aligned observations are indeed compatible. The case of WASP-96 b signals that compatibility of instruments should not be assumed. While wavelength overlaps between instruments can help, it should be noted that combining data sets remains risky business. The difficulty of combining observations also strengthens the need for next-generation instruments that possess broader spectral coverage.
Abstract
We present a new open source python package, based on PyLightcurve and PyTorch Paszke et al., tailored for efficient computation and automatic differentiation of exoplanetary transits. The ...classes and functions implemented are fully vectorised, natively GPU-compatible and differentiable with respect to the stellar and planetary parameters. This makes PyLightcurve-torch suitable for traditional forward computation of transits, but also extends the range of possible applications with inference and optimization algorithms requiring access to the gradients of the physical model. This endeavour is aimed at fostering the use of deep learning in exoplanets research, motivated by an ever increasing amount of stellar light curves data and various incentives for the improvement of detection and characterization techniques.
Abstract
Deep-learning algorithms are growing in popularity in the field of exoplanetary science due to their ability to model highly nonlinear relations and solve interesting problems in a ...data-driven manner. Several works have attempted to perform fast retrievals of atmospheric parameters with the use of machine-learning algorithms like deep neural networks (DNNs). Yet, despite their high predictive power, DNNs are also infamous for being “black boxes.” It is their apparent lack of explainability that makes the astrophysics community reluctant to adopt them. What are their predictions based on? How confident should we be in them? When are they wrong, and how wrong can they be? In this work, we present a number of general evaluation methodologies that can be applied to any trained model and answer questions like these. In particular, we train three different popular DNN architectures to retrieve atmospheric parameters from exoplanet spectra and show that all three achieve good predictive performance. We then present an extensive analysis of the predictions of DNNs, which can inform us–among other things–of the credibility limits for atmospheric parameters for a given instrument and model. Finally, we perform a perturbation-based sensitivity analysis to identify to which features of the spectrum the outcome of the retrieval is most sensitive. We conclude that, for different molecules, the wavelength ranges to which the DNNs predictions are most sensitive do indeed coincide with their characteristic absorption regions. The methodologies presented in this work help to improve the evaluation of DNNs and to grant interpretability to their predictions.
ABSTRACT
Retrieval methods are a powerful analysis technique for modelling exoplanetary atmospheres by estimating the bulk physical and chemical properties that combine in a forward model to best fit ...an observed spectrum, and they are increasingly being applied to observations of directly imaged exoplanets. We have adapted taurex3, the Bayesian retrieval suite, for the analysis of near-infrared spectrophotometry from directly imaged gas giant exoplanets and brown dwarfs. We demonstrate taurex3’s applicability to sub-stellar atmospheres by presenting results for brown dwarf benchmark GJ 570D which are consistent with previous retrieval studies, whilst also exhibiting systematic biases associated with the presence of alkali lines. We also present results for the cool exoplanet 51 Eri b, the first application of a free chemistry retrieval analysis to this object, using spectroscopic observations from GPI and SPHERE. While our retrieval analysis is able to explain spectroscopic and photometric observations without employing cloud extinction, we conclude this may be a result of employing a flexible temperature-pressure profile which is able to mimic the presence of clouds. We present Bayesian evidence for an ammonia detection with a 2.7σ confidence, the first indication of ammonia in a directly imaged exoplanetary atmosphere. This is consistent with this molecule being present in brown dwarfs of a similar spectral type. We demonstrate the chemical similarities between 51 Eri b and GJ 570D in relation to their retrieved molecular abundances. Finally, we show that overall retrieval conclusions for 51 Eri b can vary when employing different spectral data and modelling components, such as temperature–pressure and cloud structures.