Letter From the Director

Welcome to the DiRAC Institute newsletter, winter edition!

One quarter into the new academic year, I’m excited to share with you new additions to DiRAC’s team, and some of the exciting work and discoveries made by our researchers.

We are very pleased to welcome two new DiRAC Postdoctoral Fellows, Azalee Bostroem and Pedro Bernardinelli. Azalee brings us the exciting world of supernovae research and applications of data science in astronomy, whereas Pedro’s work takes us to the outskirts of the Solar System in search of new (dwarf) planets and comets. You can read more about their research in this newsletter.

Read further about remarkable new machine-learning driven “supernova recognizer” by Kyle Boon. In his recently published paper Kyle introduced a new statistical model called ParSNIP, which can be distinguish different types of supernovae significantly better than other state-of-the-art methods. Kyle’s code is open source, ready to be used by researchers in the community.

Finally, we will tell you about a major, multi-year collaboration we started with Carnegie Mellon on astronomical software for LSST, co-led by our Prof. Andy Connolly and generously funded by Schmidt Futures. Through this work, a team of a dozen software engineers and scientists will work to create new software platforms to analyze extremely large astronomical datasets. These types of systems will allow us to truly harness the data from the upcoming Legacy Survey of Space and Time, and map and understand the structure of our Universe.

Mario Jurić

Director, DiRAC Institute
Professor, Department of Astronomy

Meet DiRAC’s Postdoctoral Fellow: Azalee Bostroem

Azalee (pronounced “OZ-a-lee”) is really excited to be in Seattle and joining DiRAC. She comes from a non-traditional career path, majoring in Mathematics at Vassar college where she also was certified to teach middle and high school math. A detour from those plans took her to San Diego State University for a master’s degree in astronomy – where she fell in love with the field.

She developed a strong background in programming and data analysis as a research and instrument analyst at the Space Telescope Science Institute – where she supported the Cosmic Origins Spectrograph and the Space Telescope Imaging Spectrograph, two instruments on the Hubble Space Telescope. With the goal of returning to the west coast, she left the Space Telescope Science Institute to pursue a PhD in Physics at University of California, Davis where she studied hydrogen-rich supernovae with Prof. Stefano Valenti.

Hydrogen-rich supernovae are produced when the iron cores of stars between eight and thirty times the mass of the sun collapse and produce a neutron star. Infalling material bounces off the neutron star creating a shockwave which unbinds the star in a really bright explosion we call a supernova. While entire fields of study are devoted to studying massive stars and supernovae, we are still trying to figure out how to connect our observations of supernovae to their massive star progenitors. Two aspects of this connection that Azalee is particularly interested in are using supernovae to understand how massive stars lose mass just prior to explosion and what mass stars explode as hydrogen-rich supernovae. I use supernova observations to understand progenitor mass and mass loss by modeling the light curves of supernovae, observing them at X-ray and radio wavelengths, and by modeling the spectra one to two years after explosion.

While these technique worked great for the tens of supernovae we were discovering per year a few decades ago, they have not scaled well to the thousands of supernovae we are discovering with current surveys and will be unusable on most of the hundreds of thousands supernovae discovered by the Rubin Observatory’s Legacy Survey of Space and Time (LSST). Azalee’s current focus is on building tools to model the hydrogen-rich supernova light curves produced by the LSST to measure the progenitor properties of hundreds of thousands of massive stars in the final stages of their lifetimes. 

In addition to supernova research, Azalee is actively involved with the Carpentries organization which teaches best practices for programming and data analysis to scientists in an open and inclusive way. She has been a certified instructor since 2012 and has been organizing workshops at the winter meeting of the American Astronomical Society since 2014. She is currently leading the development of a new Data Carpentry curriculum for astronomy called Foundations of Astronomical Data Science which teaches fundamental astronomy and data science skills such as working with databases and tables and communicating results through a compelling visualization. 

On the weekends you can find her exploring Seattle by bike, hiking in the mountains, or paddling around Lake Union on her paddle board. She spends her less active days baking and cuddling with her two cats.

Read more on Azalee’s Website and follow on Twitter, Github, ADS Publications.

Meet DiRAC’s Postdoctoral Fellow: Pedro Bernardinelli

Pedro Bernardinelli was born in São Paulo, Brazil and completed his undergraduate studies in Physics at the University of São Paulo. After that, he got his Ph.D from University of Pennsylvania, focusing on the development and application of new techniques for the discovery and characterization of the most distant bodies in our Solar System, trans-Neptunian objects, as a member of the Dark Energy Survey (DES).

As part of this research, Pedro has led the discovery of over 600 TNOs and the comet C/2014 UN271 (Bernardinelli-Bernstein), the largest Oort-cloud comet ever found. His research also has had deep applications to the Planet 9 hypothesis, as well as to current models of the trans-Neptunian region.

At the University of Washington, Pedro is excited to expand this research to current surveys, as well as upcoming projects such as the DECam Ecliptic Exploration Project (DEEP) and the Rubin Observatory’s Legacy Survey of Space and Time (LSST).

Pedro is also generally interested in astronomical data analysis and image reduction techniques, going from precise astrometry and photometry to detection of faint sources. Before the pandemic started, Pedro was one of the hosts and organizers of Astronomy on Tap Philly.

Outside academia, Pedro is interested in photography, fantasy/sci-fi literature, board and video games, cooking, baking, and is a great coffee enthusiast.

Read more on Pedro’s Website, and follow Twitter,  Github, ADS Publications.

 

ParSNIP: Using deep learning to identify supernovae and probe dark energy

DiRAC researchers are heavily involved in building the Vera C. Rubin Observatory, a new facility that is currently under construction in Chile. This observatory will feature the 8.4 meter Simonyi Survey Telescope and the world’s largest CCD camera which will scan the entire visible sky every three nights. It will discover and observe millions of supernovae which are powerful explosions of stars that can outshine an entire galaxy for a brief period of time.

A particular type of supernovae called “Type Ia” can be used to map out how the universe has expanded since the big bang. This led to the discovery of dark energy which was awarded the Nobel Prize in 2011. The Rubin Observatory will discover over 100 times as many Type Ia supernovae then have been observed by all surveys to date and will dramatically improve our understanding of the universe.

Extracting scientific results from this large deluge of data is a big challenge. In a paper that was recently published in the Astronomical Journal, DiRAC Fellow Kyle Boone discusses a new statistical model called ParSNIP that can be used to distinguish Type Ia supernovae from others and improve our maps of the universe. This novel work combines recent advances in computer science and deep learning with physics models of how light propagates through the universe. The resulting hybrid model is the first one that can empirically describe how the emitted light spectrum from any kind of supernova evolves over time.

This foundational work has many applications. ParSNIP will be used to identify the different kinds of supernovae that the Rubin Observatory finds, and it can do this with over twice the performance of previous models. It will also be used to hunt for new unknown kinds of supernovae in the large Rubin dataset. ParSNIP will use all of the millions of supernovae that the Rubin Observatory discovers to measure the properties of dark energy in contrast to current methods that can only use less than a tenth of the full sample. This work will transform supernova science with the Rubin Observatory and help to extract the full scientific potential.

ADS Publication: Published October, 2021, ParSNIP: Parametrization of SuperNova Intrinsic Properties

About

Kyle Boone is DiRAC Postdoctoral Fellow. Kyle’s 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”. One aspect of his research focuses on identifying Type Ia supernovae among the millions of astronomical transients that upcoming astronomical surveys such as the Large Synoptic Survey Telescope (LSST) will discover.

Read more here. GitHub here.

Carnegie Mellon, UW to Pioneer Platforms that Harness Astrophysical Data to Unravel the Universe’s Mysteries

Close your eyes and imagine the night sky filled with billions of stars, galaxies, stellar clusters and asteroids. Incredible, right? Over the next decade, those celestial images will be captured through the Legacy Survey of Space and Time (LSST) at the Vera C. Rubin Observatory in Chile. 

The University of Washington is one of the four founders of the LSST Project, which will be the most ambitious and comprehensive optical astronomy survey ever undertaken. And faculty and researchers from the University of Washington DiRAC Institute will play leading roles in developing its science capabilities and data processing pipelines.

Carnegie Mellon University and the University of Washington have announced an expansive, multi-year collaboration to create new software platforms to analyze large astronomical datasets generated by the upcoming Legacy Survey of Space and Time (LSST), which will be carried out by the Vera C. Rubin Observatory in northern Chile. The open-source platforms are part of the new LSST Interdisciplinary Network for Collaboration and Computing (LINCC) and will fundamentally change how scientists use modern computational methods to make sense of big data. 

Through the LSST, the Rubin Observatory, a joint initiative of the National Science Foundation and the Department of Energy, will collect and process more than 20 terabytes of data each night — and up to 10 petabytes each year for 10 years — and will build detailed composite images of the southern sky. Over its expected decade of observations, astrophysicists estimate the Department of Energy’s LSST Camera will detect and capture images of an estimated 30 billion stars, galaxies, stellar clusters and asteroids. Each point in the sky will be visited around 1,000 times over the survey’s 10 years, providing researchers with valuable time series data. 

Scientists plan to use this data to address fundamental questions about our universe, such as the formation of our solar system, the course of near-Earth asteroids, the birth and death of stars, the nature of dark matter and dark energy, the universe’s murky early years and its ultimate fate, among other things.

“Our goal is to maximize the scientific output and societal impact of Rubin LSST, and these analysis tools will go a huge way toward doing just that,” said Jeno Sokoloski, director for science at the LSST Corporation. “They will be freely available to all researchers, students, teachers and members of the general public.”

The Rubin Observatory will produce an unprecedented data set through the LSST. To take advantage of this opportunity, the LSST Corporation created the LSST Interdisciplinary Network for Collaboration and Computing (LINCC), whose launch was announced August 9 at the Rubin Observatory Project & Community Workshop. One of LINCC’s primary goals is to create new and improved analysis infrastructure that can accommodate the data’s scale and complexity that will result in meaningful and useful pipelines of discovery for LSST data.

“Many of the LSST’s science objectives share common traits and computational challenges. If we develop our algorithms and analysis frameworks with forethought, we can use them to enable many of the survey’s core science objectives,” said Rachel Mandelbaum, professor of physics and member of the McWilliams Center for Cosmology at Carnegie Mellon.

The LINCC analysis platforms are supported by Schmidt Futures, a philanthropic initiative founded by Eric and Wendy Schmidt that bets early on exceptional people making the world better. This project is part of Schmidt Futures’ work in astrophysics, which aims to accelerate our knowledge about the universe by supporting the development of software and hardware platforms to facilitate research across the field of astronomy.

“Many years ago, the Schmidt family provided one of the first grants to advance the original design of the Vera C. Rubin Observatory. We believe this telescope is one of the most important and eagerly awaited instruments in astrophysics in this decade. By developing platforms to analyze the astronomical datasets captured by the LSST, Carnegie Mellon University and the University of Washington are transforming what is possible in the field of astronomy,” said Stuart Feldman, chief scientist at Schmidt Futures.

“Tools that utilize the power of cloud computing will allow any researcher to search and analyze data at the scale of the LSST, not just speeding up the rate at which we make discoveries but changing the scientific questions that we can ask,” said Andrew Connolly, a professor of astronomy, director of the eScience Instituteand former director of the Data Intensive Research in Astrophysics and Cosmology (DiRAC) Institute at the University of Washington.

Connolly and Carnegie Mellon’s Mandelbaum will co-lead the project, which will consist of programmers and scientists based at Carnegie Mellon and the University of Washington, who will create platforms using professional software engineering practices and tools. Specifically, they will create a “cloud-first” system that also supports high-performance computing (HPC) systems in partnership with the Pittsburgh Supercomputing Center (PSC), a joint effort of Carnegie Mellon and the University of Pittsburgh, and the National Science Foundation’s NOIRLab. LSSTC will run programs to engage the LSST Science Collaborations and broader science community in the design, testing and use of the new tools.

“The software funded by this gift will magnify the scientific return on the public investment by the National Science Foundation and the Department of Energy to build and operate Rubin Observatory’s revolutionary telescope, camera and data systems,” said Adam Bolton, director of the Community Science and Data Center (CSDC) at NSF’s NOIRLab. CSDC will collaborate with LINCC scientists and engineers to make the LINCC framework accessible to the broader astronomical community.

Through this new project, new algorithms and processing pipelines developed at LINCC will be able to be used across fields within astrophysics and cosmology to sift through false signals, filter out noise in the data and flag potentially important objects for follow-up observations. The tools developed by LINCC will support a “census of our solar system” that will chart the courses of asteroids; help researchers to understand how the universe changes with time; and build a 3D view of the universe’s history.

“The Pittsburgh Supercomputing Center is very excited to continue to support data-intensive astrophysics research being done by scientists worldwide. The work will set the stage for the forefront of computational infrastructure by providing the community with tools and frameworks to handle the massive amount of data coming off of the next generation of telescopes,” said Shawn Brown, director of the PSC. 

Northwestern University and the University of Arizona, in addition to Carnegie Mellon and the University of Washington, are hub sites for LINCC. The University of Pittsburgh will partner with the Carnegie Mellon hub.  

Sifting through the Static

Trans-Neptunian objects provide a window into the history of the solar system, but they can be challenging to observe due to their distance from the Sun and relatively low brightness.

In the recently published paper, Sifting through the Static: Moving Object Detection in Difference Images, DiRAC researchers report the detection of 75 moving objects that could not be linked to any other known objects, the faintest of which has a VR magnitude of 25.02 ± 0.93 using the Kernel-Based Moving Object Detection (KBMOD) platform.

They recover an additional 24 sources with previously known orbits and place constraints on the barycentric distance, inclination, and longitude of ascending node of these objects. The unidentified objects have a median barycentric distance of 41.28 au, placing them in the outer solar system. The observed inclination and magnitude distribution of all detected objects is consistent with previously published KBO distributions. They describe extensions to KBMOD, including a robust percentile-based lightcurve filter, an in-line graphics-processing unit filter, new coadded stamp generation, and a convolutional neural network stamp filter, which allow KBMOD to take advantage of difference images.

These enhancements mark a significant improvement in the readiness of KBMOD for deployment on future big data surveys such as LSST.

ADS Published Paper access here.