In this series of posts, we sit down with a few of the keynote speakers of the 241st AAS meeting to learn more about them and their research. You can see a full schedule of their talks here, and read our other interviews here!
Dr. Dan Foreman-Mackey has become a special figure in the methodology of analyzing astronomical data, and he also helps astronomy research with his insights on software building. His most famous contribution is the python package emcee, which now has over one thousand stars on GitHub. He is also the author of the packages like exoplanet, corner.py, and daft, and has contributed to many more.
Dr. Foreman-Mackey completed his PhD at New York University and is now an Research Scientist at the Flatiron Institute, which is equivalent to the position of Associate Professor in a traditional academic institution.
Time-dependent research interests
Dr. Foreman-Mackey is quite a unique astronomer. If you ask him what he is working on, his answer would be “data analysis” with almost no specification (not “exoplanets” or not “black holes”), because he is interested in it all. So, his research is definitely a function of time, and you might want to ask what he is focused on at this very moment.
“Right this second, I am extremely deep in the weeds at solving Kepler’s equation extremely fast and in a parallelizable way, which is something that I come back to every few months because it’s my favorite test problem that has some impact in actual astronomy research. For me, the big motivating thing is learning new things. So, every time I come back to it, I have a completely new vision of how I want to approach the problem and learn a lot about […] computational methods. It’s such a simple problem that has hundreds of years of literature and methodology behind it, but there are still new things to do and new things to learn.”
However, the problem of solving Kepler’s equation efficiently is not a bottleneck in his actual research – it is more of a personal computational challenge and an opportunity to learn new things.
“In my career, I’ve always been more motivated by the statistical challenges rather than the specific research domains. I ended up working on exoplanets [first] because I started my PhD right as the original Kepler data releases were coming out. And so that presented a whole set of new problems that were really interesting, and it was a nice opportunity to confront these things. […] I went to NYU specifically because I knew papers that David Hogg and his group had been working on, and I was really excited about the way that they were approaching data analysis in all sorts of different domains.”
Dan’s eyes fire up when he talks about various statistical and computational challenges. Yet I had always been wondering: one can find those challenges in any other field, why did he choose to do astronomy? The answer is collaboration.
“One thing about astronomy is the collaborative community. […] That’s a really nice place to work as someone like me who likes to contribute to other people’s projects. And I like to be in a situation where I don’t have to mastermind the research project, but I can provide this core support of the data analysis methodology or the software infrastructure. […] I think astronomy in some ways has started to recognize the importance of those kinds of contributions. […] I like the data access policies. I like the way that we approach collaborations […] And I think it’s a wonderful place to work because of that.”
What is the value of astronomical software?
Astronomers code. A lot. It helps tremendously when you don’t have to reinvent the wheel and just use existing software. However, if you’re the one trying to help and write the software, then there’s a tradeoff–you might spend too much time on building the software instead of doing the actual research. I decided to touch on this topic with Dan since he spends a lot of time on software development projects.
“What is the value of astronomers who code versus coders who astronomize? And there are various projects where if you think about hiring professional software developers with astronomy backgrounds to work on developing code, or there might be tools that are developed in other fields that out of the box can be applied to astronomy. I think that there are legs to those ideas. But there’s also a huge amount of benefit to having astronomers understand how to write good software […] because in any scientific domain there are specific problems that you need to solve that aren’t relevant in other fields and the way that you’ll approach them. Actually bringing in domain expertise can be really beneficial. I think it’s easier to identify what problems need to be solved if you’re embedded in the community. My long-term hope is that astronomical software gets the support and credibility that it obviously deserves. […] I think it’s really important that our field has people with a very diverse set of research interests.”
What should graduate students be focusing on?
Dr. Foreman-Mackey told me that everything in statistics and coding he knows is self-taught. That momentarily added one thousand more things to my to-do list, because there is indeed so much to learn and the list will only be adding up. Therefore, I decided to ask him what the best strategies would be for graduate students who want to learn software development or graduate students who are overwhelmed by the number of things to learn. It turns out the answer is simple: work on things that make you happy.
“I think it’s most important to be, like, working on what you find motivating, because, hopefully, it makes you happy. […] One thing I would say is that people shouldn’t be shy or self-conscious about their efforts. I hear a lot that people don’t want to put their software work out there because they’re worried that it’s not very good or people are going to think poorly of them because they’re going to go and look at the code. […] That’s very much not true. If someone’s looking at the code, then you’ve done something right, and there’s a chance that someone is going to learn something from it or you’re going to start a new collaboration. […] We’re all learning all the time and I’m devastatingly embarrassed about some of the early software that I’ve developed and also the stuff I wrote yesterday.
In the context of being stressed about career things. Have some faith, things will probably be good. […] It’s important to remember that you can’t necessarily anticipate exactly where you want to land. If you’re highly motivated by what you’re working on, then you’ll do a better job at it.”
To hear more about data analysis and how to build software, be sure to attend Dr. Foreman-Mackey’s Plenary Lecture at 3:40 PM PDT on Thursday, January 12 at #AAS241!
Astrobite edited by Katya Gozman
Featured image credit: AAS