Multilayer Perceptron and Geometric Albedo Spectra for Quick Parameter Estimations of Exoplanets

Abstract

Future space telescopes now in the concept and design stage aim to observe reflected light spectra of extrasolar planets. Assessing whether given notional mission and instrument design parameters will provide data suitable for constraining quantities of interest typically requires time consuming retrieval studies in which tens to hundreds of thousands of models are compared to data with a given assumed signal to noise ratio, thereby limiting the rapidity of design iterations. Here we present a novel machine learning approach employing five Multilayer Perceptron's (MLP) trained on model albedo spectra of extrasolar giant planets to estimate a planet's atmospheric metallicity, gravity, effective temperature, and cloud properties given simulated observed spectra. The stand-alone C++ code we have developed can train new MLP's on new training sets, within minutes to hours, depending upon the dimensions of input spectra and desired output. After the MLP is trained, it can classify new input spectra within a second, potentially helping speed observation and mission design planning. Four of the MLP's were tuned to work with various levels of spectral noise. The fifth MLP was developed for cases where the user is uncertain about the noise level. The MLP's were trained using a grid of model spectra that varied in metallicity, gravity, temperature, and cloud properties. We tested the MLP on noisy models and observed spectra of both Jupiter and Saturn. The root mean squared error when applied to noisy models were on the order of the model grid intervals. The results show that a trained MLP is a robust means for reliable in-situ estimations, providing an elegant artificial intelligence that is simple to customize and quick to use.