D I R A C   I n s t i t u t e

Stephen Portillo’s research focuses on using advances in statistics and machine learning to allow more science to be done with existing astronomical data sets. On the statistics front, he has been developing probabilistic cataloging, a Bayesian Markov chain Monte Carlo method that improves source detection and measurement in crowded images. On the machine learning front, he has been applying autoencoders, a type of deep neural network, to enable astronomers to more easily find patterns and outliers in large datasets.

Dr. Portillo is a DiRAC Postdoctoral Fellow and UW Data Science Postdoctoral Fellow. He joined the DiRAC Institute in September 2018 after finishing his PhD in Astronomy and Astrophysics at Harvard University. Before graduate studies, he completed a BSc in Astrophysics at the University of Alberta in Edmonton, Canada in 2012.

Crowded images are difficult to analyze because sources can appear blended with their close neighbors. This problem will worsen with more sensitive observatories like Rubin Observatory, James Webb Space Telescope, and Roman Space Telescope that will see unprecedented numbers of objects in the same area of sky. Unlike traditional methods that first identify sources before measuring them, probabilistic cataloging treats source identification probabilistically. Dr. Portillo has shown that this method can find stars four times fainter than state-of-the-art methods in extremely crowded images with 1 star per 10 pixels.

At DiRAC, Dr. Portillo has joined the KBMOD team, who are developing GPU-accelerated software to find Kuiper belt objects. Recently, he has been developing a method to correct for the Earth’s motion in the solar system, allowing KBMOD to better track objects over longer periods of time. He is also extending probabilistic cataloging to search for binaries among the objects found by KBMOD, because these binaries are powerful probes of the dynamical history of the outer Solar System.

Working with Prof. Connolly, Dr. Portillo has implemented a variational autoencoder to galaxy spectra from the Sloan Digital Sky Survey. He showed that the autoencoder can summarize spectra with thousands of pixels with only six numbers and easily separates known classes of galaxies. Currently, he is working with students to use this autoencoder to find massive black hole binaries, rare objects identified by unusual spectra.

Dr. Portillo is also passionate about public outreach and teaching. While at DiRAC, he has given an Astronomy at Home talk and given virtual presentations to school groups. He was also a lecturer at AstroHackWeek 2020 and is currently co-instructing an undergraduate course on astrostatistics with Prof. Juric.

Dr. Portillo is excited to be at the DiRAC Institute because it brings together researchers interested in all aspects of data-intensive science from software engineering to statistics and machine learning. He is also happy to be a part of the eScience Institute that encourages researchers across scientific fields to find commonalities in the ways they analyze data.

Over the past year, James Davenport (research assistant professor, and the Associate Director of the DiRAC Institute) and his team at the University of Washington have studied the eclipsing binary system, HS Hydrae. This star is one of rare class of eclipsing systems that is changing in real time, with its eclipses starting to disappear!

Using new observations from NASA’s exoplanet-hunting TESS mission, and a historical archive of photographic plates reaching back to 1893, Davenport and his team were able to document the complete “rise and fall” of eclipses for the system due the orbit of a third, unseen star. They believe HS Hydrae is now no longer an eclipsing binary!

This work was presented in a Press Conference at the 237th Meeting of the American Astronomical Society in early 2021.

UW News: https://www.washington.edu/news/2021/01/14/hs-hydra/

One of the significant research focuses at the DiRAC Institute has been the development of next generation asteroid and comet discovery algorithms. DiRAC researchers have published a pre-print detailing one such algorithm called “Tracklet-less Heliocentric Orbit Recovery” (THOR). Applied to observations from the Zwicky Transient Facility (ZTF), THOR recovered 97% of the known objects with at least 5 observations, a factor of 1.5-2 more than the current generation of discovery algorithms. In addition to recovering most of the known objects, THOR would have discovered nearly 500 new Solar System objects (including a parabolic/hyperbolic comet) had it been running when the observations were made in 2018. 

The Hammer and the Comet 

One of the significant research focuses at the DiRAC Institute has been the development of next generation asteroid and comet discovery algorithms. DiRAC researchers have published a pre-print detailing one such algorithm called “Tracklet-less Heliocentric Orbit Recovery” (THOR). Discovering minor planets involves having current generation astronomical surveys observe the same area of the sky at least twice in one night. The two sets of observations of the same region of the sky can then be scanned for what is known as a “tracklet”: a motion vector made of at least two observations that could represent the actual motion of a Solar System object. This observing pattern is repeated over the course of a 2-week window until enough tracklets are observed so that they can be used to discover asteroids and comets. 

Requiring tracklets for Solar System discovery has two striking consequences: first, any astronomical survey with the goal of discovering minor planets must observe the same area of sky at least twice in one night thereby limiting the amount of sky the telescope could observe in a single night. Second, any dataset that was not constructed with a tracklet building cadence is not a dataset suitable for Solar System discovery. Leveraging the latest innovations in Solar System discovery and backed by large scale computing, THOR has addressed these concerns by removing the requirement for tracklets to be made and enabling minor planet discovery without the need for a specific cadence of observations. 

As a proof-of-concept demonstration, THOR was applied to two weeks of observations from the Zwicky Transient Facility (ZTF) that were taken in early September 2018. THOR recovered 97% of all Solar System objects known at the time. ZTF’s own discovery algorithm, ZMODE, which relies on tracklet-like observations for discovery, could at best recover 68% of the same population. The Vera C. Rubin Observatory’s discovery algorithm which also relies on tracklets would at best recover 45% of the same population. In other words, by enabling Solar System discovery without requiring a specific pattern of observations, THOR can recover 1.5-2 times as many asteroids and comets as the current generation of algorithms. Of the 21,000+ orbits that were recovered by THOR, 488 were identified as high quality discovery candidates that could not be associated with any objects that were known in 2018. 

DiRAC researchers then posed the question: had THOR been running on ZTF when the data were taken in 2018, how many asteroids and comets would it have discovered? Of the 488 discovery candidates identified, THOR would have discovered at least 477 new asteroids and comets had it been running as ZTF’s discovery algorithm when the observations were made. 

The observations and best-fit trajectories of the 11 remaining objects are shown in the figure. Subsequent analysis showed that of the 11 candidates, 10 are as yet undiscovered objects. The e > 1 candidate (2nd row, 3rd column) represents “precovery” observations of parabolic/hyperbolic comet C/2018 U1 that was discovered on October 27th 2018 by the Mount Lemmon Survey. Precovery observations are observations of an object that pre-date its original discovery date. The ZTF data on which THOR was tested and developed were taken 6-8 weeks prior to the discovery date of this comet. Had THOR been running in September 2018, it would have allowed ZTF to claim the discovery of this fascinating object. 

Following their success using just two weeks of ZTF observations, the research team behind THOR is now working to process all three years of ZTF observations. They anticipate this should yield the discovery of several hundred new Solar System objects. THOR is completely open-source and available on GitHub. The eventual goal of the THOR project is to launch a discovery service where surveys can submit their observations and, powered by THOR, they can be processed and analyzed for the discovery of new asteroids and comets.

Joachim Moeyens is a graduate student in the Department of Astronomy at the University of Washington. He is interested in big data and software driven solutions to problems in astronomy. During his undergraduate studies at the University of Washington, he was presented with the opportunity to work on a research project for the Vera Rubin Observatory’s Legacy Survey of Space and Time (LSST).  For his doctoral thesis, Joachim is working on algorithms that discover minor planets in astronomical surveys, in particular, on Rubin Observatory’s Solar System Processing pipelines, and on a novel algorithm named Tracklet-less Heliocentric Orbit Recovery (THOR).

Type Ia supernovae are some of the most powerful tools for testing different theories of gravity. These supernovae are explosions of massive stars that all look remarkably similar. By measuring how bright a supernova is, we can figure out how far away it is. Type Ia supernovae were used to make the initial discovery of dark energy in 1998. We have since used supernovae to measure the properties of dark energy with better and better precision with the goal of determining what it really is. In our new work, we developed a technique that uses the spectra of Type Ia supernovae to improve how well we can measure the distances to them. Our new technique can measure these distances around twice as well as previous techniques, and our results will be very important for measurements of dark energy with upcoming surveys such as the Legacy Survey of Space and Time (LSST) at the Rubin Observatory, or for the Nancy Grace Roman Space Telescope.

Kyle Boone is a lead author on two papers published in The Astrophysical Journal that report these findings. Currently a postdoctoral fellow at the University of Washington, his research focuses on developing novel statistical methods for astronomy and cosmology. He is particularly interested in using Type Ia supernovae to probe the accelerated expansion of the universe that we believe is due to some form of “dark energy”. Dr. Boone is a former graduate student of Nobel laureate Saul Perlmutter, the Berkeley Lab senior scientist and UC Berkeley professor who led one of the teams that originally discovered dark energy. Dr. Perlmutter was also a co-author on both studies.

The Twins Embedding of Type Ia Supernovae. II. Improving Cosmological Distance Estimates

The Twins Embedding of Type Ia Supernovae. I. The Diversity of Spectra at Maximum Light

Surveys like ZTF and the LSST on the Vera C. Rubin Observatory are improving our understanding for nearly every area of modern astronomy. Sometimes, however, these large projects discover something truly unexpected…

James Davenport (UW research assistant professor, and the Associate Director of the DiRAC Institute) and Beatriz Villaroel (Stockholm University) were interviewed this April by the SETI Institute‘s Seth Shostak about the “VASCO” project to search for disappearing stars. In this hour-long discussion, Davenport and Villaroel discuss the importance of searching for intelligent life in the universe, and finding the unexpected in our data.