Wednesday, December 11, 2013

Intelligence in Astronomy: The Growth of My Intelligence

When I was in graduate school at UC Berkeley, I had a very rough first year. I started astronomy graduate school with a B.S. in physics from a small mid-western school and zero preparation in astronomy. I didn't use a telescope until I was 21 years old, I hadn't taken an astronomy course as an undergrad, and upon my arrival at Berkeley I couldn't tell you why the moon went through phases. Seriously. I learned moon phases as a TA of Astro 10.
Campbell Hall at UC Berkeley. My office was next to the
dome on the right side. The building was torn down a few
years ago. 
I remember very clearly heading down to the sixth floor of Campbell Hall for my Stellar Structure class, taught by Prof. Frank Shu. As I walked down the stairs with the other students, two of the second-years, Jason Wright and Erik Rosolowsky, were engaged in an intense discussion the likes of which I had never heard before from students. They were discussing whether the forward-scattering of light is the same as inverse Compton scattering, and under which conditions one description is better than the other (or at least this is how I recall the conversation).

I remember chuckling and thinking, "Yeah, right, they're seriously discussing physics outside of class like they're professors. Hilarious!" But then it sank in: they were dead serious. The joke was on me, and it was clear that intellectually I was a long way away from these high-power second-years. They were Smart. I was not.

Interactions like this continued at more or less a steady state in my first year, and I felt less and less capable compared to my peers. Two of my classmates came from Caltech and Harvard, respectively. The third came from Maryland, and he had completed most of his graduate course work as a PhD student there before transferring to Berkeley. I came from the University of Missouri. At Rolla. People generally don't even know how to pronounce Rolla. (BTW, UMR is now Missouri S&T)

The second-years were taking classes with me in the morning, but working on mysterious-looking astronomy data late into the evening and talking in a foreign tongue while doing so. I knew classroom physics, but my fellow students had taken the next step and could describe how all of that problem-set physics applied to things such as interstellar dust grains and the structure of the Sun. I was lost, and I was feeling increasingly stupid.

What saved me was a class taught by a postdoc named Doug Finkbeiner. As a former BADGrad and newly-minted Berkeley postdoc, Doug decided to teach a late-night, unofficial course on astronomical computer programming using this new and exciting scripting language called IDL. About a dozen students gathered in the seventh-floor undergraduate laboratory to use the new Sun workstations that a donor had purchased (oddly, first-year students used Sun SPARC workstations essentially as dummy terminals to login to faster computers. The undergrads had the real computing power with their SunUltra 10s.)

Doug taught a decidedly untraditional class. Each week he would teach a couple new concepts (e.g. array-based math operations), and he'd hand us some data and have us analyze the data using the programming techniques he just described. Because of my extensive past computer science experience I took to the analysis problems like a duck to water. Plus, the vector-based nature of IDL really meshed with the way my brain thinks about the world. Suddenly, all of those classroom lessons on Linear Algebra and even calculus were coming to life on my screen as I manipulated astronomical images. Fourier transforms took on a much deeper meaning after Doug gave us a radio telescope time series of the Crab Pulsar. Not to mention the fact that we were looking at the Crab frikkin' Pulsar!


I took my new-found IDL expertise and applied it to my Stellar Structure and Radiative Processes course work. IDL's plotting tools were just what my visual-manipulation brain needed to see past the confusing, abstract mathematics. Multi-dimensional integration, which to this day is often incomprehensible to me on the written page, became my best friend first using IDL's TOTAL() function and later my own numerical integrators (oooohhhhh, that's how the trapezoidal rule works!).

Doug's fly-by-night class gave me the tools I needed to not just get by in my courses, but start to excel in grad school. HW sets were no longer a slog, but were rather exercises that strengthened my physical intuition, built on my growing problem-solving toolset, and increased my vocabulary. With these traits growing in strength, I was able to start engaging in the science discussions that my older classmates were having, and as I did so my vocabulary, problem-solving sense, and intuition also grew. This positive feedback process led to an exponential growth in all of the traits that eventually led me to become a tenured professor at Harvard.

My intelligence grew exponentially.
Figure by Nathan Sandars on Astrobites.org. For more see Nathan's interview of yours truly.
More importantly, I didn't quit in the face of difficulty. Why? It's tough to say. I was truly on my way out the door during my first year. One important ingredient was taking a class that was taught in a way that really resonated with the way my brain works. Another ingredient was having Doug Finkbeiner as a teacher. He was, and is, an astrophysical god in my mind, a wizard of the highest order. Yet his approach to science was so humble and down to Earth. He had these sayings that I use to this day. When tackling a tough, complicated problem, he'd say "Just do the stupidest possible thing first." If that doesn't work, start increasing the sophistication slowly until you find a solution. Astrophysical problems can at first seem daunting. But they can all be reduced to sophomore-level physics to get a first-order solution that can be implemented in code. And given the nature of most astrophysical data, first-order is often good enough!

Another ingredient was vividly seeing how the effort I invested paid off in direct proportion in my ability to do astrophysics. More effort in, more growth out. I may not be smart right at this moment, but give me a week or two and I'll be just as smart, if not smarter than you. More importantly, I saw my classmates working their asses off. I figured out that Jason Wright didn't arrive at Berkeley with his encyclopedic knowledge. He arrived with perhaps a quarter of an encyclopedia. At any given moment, I could walk into his office and find him reading an article or book, talking science with his office mates, doing math at his whiteboard, or programming. He worked. Hard. If I wanted to be like Jason, and I most certainly did, I'd have to put in the work.

My intelligence grew exponentially. And it's still growing because when challenges come my way, or when I encounter something that "I should know because I'm a Harvard professor," I don't back down. I don't try to find an easy path around it. I find someone who knows, like Smadar Naoz or Ruth Murray Clay or Avi Loeb, and I ask them "dumb" questions. I ask them to send me to the chalk board so I can struggle through basic problems. And you know what? They don't seem to judge me as being stupid. They help me learn.

At this pace, it's absolutely scary to think of how intelligent I'll be in 10 years.

Epilogue

This year Doug Finkbeiner and I both became tenured professors at Harvard. Doug was tenured jointly in Astronomy and Physics, and I was tenured in Astro. I'm still learning from him and it's an honor to work in the same department as him.

No comments: