A pinboard by
Marc Hon

Ph. D student , UNSW Sydney


Using an artificial intelligence-like approach to analyse stellar interiors and populations.

As giant hot balls of gas, stars display a lot of activity from sound waves in their interiors, which we can understand by measuring how their brightness fluctuates with time. These fluctuations form specific data patterns, which asteroseismology uses to measure properties of the star such as its size and mass through an understanding of how the sound waves travel in the star. My research uses deep learning , which is an approach to artificial intelligence, to train machines to learn these data patterns and give high efficient predictions on the properties of stars in our Galaxy.


The K2 Galactic Archaeology Program Data Release 1: Asteroseismic results from Campaign 1

Abstract: NASA's K2 mission is observing tens of thousands of stars along the ecliptic, providing data suitable for large scale asteroseismic analyses to inform galactic archaeology studies. Its first campaign covered a field near the north galactic cap, a region never covered before by large asteroseismic-ensemble investigations, and was therefore of particular interest for exploring this part of our Galaxy. Here we report the asteroseismic analysis of all stars selected by the K2 Galactic Archaeology Program during the mission's "North Galactic Cap" campaign 1. Our consolidated analysis uses six independent methods to measure the global seismic properties, in particular the large frequency separation, and the frequency of maximum power. From the full target sample of 8630 stars we find about 1200 oscillating red giants, a number comparable with estimates from galactic synthesis modeling. Thus, as a valuable by-product we find roughly 7500 stars to be dwarfs, which provide a sample well suited for galactic exoplanet occurrence studies because they originate from our simple and easily reproducible selection function. In addition, to facilitate the full potential of the data set for galactic archaeology we assess the detection completeness of our sample of oscillating red giants. We find the sample is at least near complete for stars with 40 < numax/microHz < 270, and numax_detec < 2.6*1e6 * 2e-Kp microHz. There is a detection bias against helium core burning stars with numax ~ 30 microHz, affecting the number of measurements of DeltaNu and possibly also numax. Although we can detect oscillations down to Kp = 15, our campaign 1 sample lacks enough faint giants to assess the detection completeness for stars fainter than Kp ~ 14.5.

Pub.: 29 Nov '16, Pinned: 19 Jan '18

Identifying Exoplanets with Deep Learning: A Five Planet Resonant Chain around Kepler-80 and an Eighth Planet around Kepler-90

Abstract: NASA's Kepler Space Telescope was designed to determine the frequency of Earth-sized planets orbiting Sun-like stars, but these planets are on the very edge of the mission's detection sensitivity. Accurately determining the occurrence rate of these planets will require automatically and accurately assessing the likelihood that individual candidates are indeed planets, even at low signal-to-noise ratios. We present a method for classifying potential planet signals using deep learning, a class of machine learning algorithms that have recently become state-of-the-art in a wide variety of tasks. We train a deep convolutional neural network to predict whether a given signal is a transiting exoplanet or a false positive caused by astrophysical or instrumental phenomena. Our model is highly effective at ranking individual candidates by the likelihood that they are indeed planets: 98.8% of the time it ranks plausible planet signals higher than false positive signals in our test set. We apply our model to a new set of candidate signals that we identified in a search of known Kepler multi-planet systems. We statistically validate two new planets that are identified with high confidence by our model. One of these planets is part of a five-planet resonant chain around Kepler-80, with an orbital period closely matching the prediction by three-body Laplace relations. The other planet orbits Kepler-90, a star which was previously known to host seven transiting planets. Our discovery of an eighth planet brings Kepler-90 into a tie with our Sun as the star known to host the most planets.

Pub.: 13 Dec '17, Pinned: 19 Jan '18