Whew, what a week! And where to start? This weekend was mostly dedicated to the onboarding process. I was able to get access to University of Washington IT services (and a UW email!) that will eventually give me access to the powerful network of computers that will absolutely trounce my workstation. Speaking of my workstation, it was finished this week–just in time for my DREU! My spouse built this water-cooled system that can handle a lot of data crunching and computing quickly and efficiently, although it can’t possibly compare to the computing power that the DIRAC institute has. I’m not there yet, so my workstation is more than enough, but I can’t wait to get to running analytics on a couple hundred thousand light curves.
I guess I should take this time to explain my science and what I’ll be researching this summer. It might be different from a lot of the DREU researchers, but it’s no less exciting!
Like most sciences, astronomy is dealing with an overwhelming amount of data. The technology for gathering data gets better year after year, and that only adds to the amount of computational power that’s required to categorize and find patterns in that data. The existing modeling techniques already take a hefty amount of computational power and time, and the increase in data resolution and data is simply too much. This is why advances in machine learning as they apply to astronomy is important, and is the area of research I want to get into during and after grad school. When the Vera C. Rubin Observatory in Chile is completed, it will be taking in 20 terabytes of raw data per night–and it’s not the only major telescope project that’s being built.
While the Vera C. Rubin Observatory won’t be operational until after I graduate in 2022, I get to work on data from a telescope in space! NASA’s Transiting Exoplanet Survey Satellite (TESS) takes pictures of stars every two minutes and measures the fluctuations in their brightness. It will stare at the same piece of sky for 27.4 days, producing nearly 20,000 data points per star…and there are tons of stars in each picture! If you graph the brightness vs. time, you get what is called a light curve. If there is a transit–meaning one object is going in front of another–there will be a decrease in brightness. TESS’s goal is to find exoplanets, which are planets orbiting a star that is not our sun, but my research goal is to identify eclipsing binaries: stars that orbit each other in pairs.
Each light curve has nearly 20,000 data points and TESS will be observing 200,000 stars. That is a lot of data to go through, which is why I will be researching how we can effectively use machine learning on TESS data! Dr. Davenport and I agreed that my goal for this summer should be to develop a machine learning model that can identify obvious binary star systems in the TESS data set by finding the eclipses. If time permits, we’d also like to identify their orbit periods. Now we can only identify binary stars where the stars cross nearly directly between Earth and the other star, but there are still a lot of stars we can identify!
But we still have to start with the basics: training data. My goal (which I just finished!) for this week was to successfully open and read the TESS data, which is kept in a fits file format. They’re tricky but are able to hold a lot of information! Luckily Astropy had a wonderful guide to help me out with this.
That’s all I have for now, but stay tuned for next week’s update!