Using the Hubble Telescope to Discover Elements of Interest - Week 2

Week 2: A Day in the Life...

by Rabeea Ahmed

Every morning, when I enter the Space Science Institute (STScI) to begin my work for the day, breath-taking pictures of space that line the walls of the lobby greet me. Although I have walked through the lobby every day for over three weeks, the one-tenth-scale replica of the Hubble Space Telescope that hangs on the ceiling of the lobby always catches my attention. It infuses within me the thrill of being a part of an expedition that ventures into the haunting depths of space to learn more about our universe and our place in it.

Most mornings at the STScI begin with a talk scheduled for student interns to learn more about the great work that takes place at the STScI. Dr. Alberto Conti, who is currently an Innovation Scientist for the upcoming James Webb Space Telescope (JWST), delivered today’s talk, entitled “A New Kind of Astronomy”. As a computer science and astronomy student, the title of the talk piqued my interest. I was curious to know what new things were at the frontier of astrophysics today and what the future was like for the field. Indeed, the focus of the talk was

about astronomy as I imagined it would be in the future - more data dependent. The
talk helped me to appreciate the scale of data received from telescopes both in space and on the ground and to realize that within a few years, these massive amounts of data will need ground breaking ideas and algorithms in order to be optimized. Indeed, Dr. Conti emphasized the need for the astrophysicists and computer scientists of tomorrow to realize that the data received in the coming years from JWST and other similar telescopes will be the ideal testing bed for complex and beautiful algorithms that will refine the data, detect patterns in it, and visualize it to increase the pace of research and discovery.

After the presentation and accompanying talk were over, it was time to head-back to my workstation and to think more deeply about my tasks for the day. This week, I am going to test the algorithms that are designed to extract points of light (aka sources) from images taken from the Hubble Space Telescope. For the sake of completeness, I decided to focus my tests on different types of images of space to see if the algorithm worked better with pictures containing certain features over others. Ergo, I broke down my image categories into galaxies, stellar fields, clusters, and nebulae after scanning many images online and finding major differences between these categories. After analyzing hundreds of thousands of data points on these images, I finally began to see trends in missing, false, or incomplete data etc. and having done so, I began to note down the problems that I had found in field-guide documents. My next task was to figure out exactly what might be the cause of these problems. To do this, however, I needed the guidance of my mentor at STScI - Dr. Brad Whitmore.

Reverse engineering an algorithm without actually seeing the code typed out might seem
counterintuitive at first. When I began my internship, I was quite worried about how this would be done, since I had been rigorously trained to find logical errors in code in my Computer Science courses at Harvard. This new technique of reverse engineering the algorithm without seeing the code (but with full knowledge about how the algorithm is designed to work from reading research
papers) seemed as exciting as it seemed challenging. Together, Dr. Whitmore and I began to go over the list of problems that I had found earlier in the data and then, after finding certain points of starlight that we found in the images, we looked at the array of numbers that held the scientifically valuable information about those points. Soon enough, the pictorial trends became translated into beautiful numbers - numbers that seemed to have very specific boundaries at
which point the algorithms seemed to get confused. A simple brute force methodology to find these magical number might have caused problems, but digging deeply into the test results drastically reduced the time needed to figure out these cut-offs and boundaries. This new method of debugging - hitherto unknown to me - was as elegant as it was exciting. I felt refreshed and, in a way, inspired by this unconventional way of thinking. As Dr. Conti had mentioned, astrophysics really was the ideal test bed for new ideas.

I made my way through the maze of corridors and out through the building with a great sense of fulfillment. I had had a chance to look into the future of astronomy and had contributed my infinitesimal bit to it as I analyzed pixels of light that had left their homes in the distant corners of the galaxy so many thousands of years ago. I had jumped on a train full of space explorers as it
took me to an utterly unknown destination - but the view into the past and the future that it afforded me made it well-worth it.

Source: http://heritage.stsci.edu/2006/01/index.html

---

About the Blogger

Rabeea Ahmed is an Astrophysics and Computer Science major at Harvard (class of 2014). She was born and raised in Pakistan, where she spent her formative years before moving to the United States for college. She's been fascinated by science since elementary school. She also enjoys roller-blading, having conversations about government and foreign cultures with my friends and absolutely loves cooking! 

"Historically, women have been underrepresented within many scientific fields and this gender gap exists in this day and age as well. There is a need for women to have a platform within which to share their experiences in science and encourage more young girls to pursue science. I think that the Scientista Foundation is a great step towards that goal." - Rabeea Ahmed