10 Questions for a Scientist: Dr. Anubhav Jain

Dr. Anubhav Jain of Lawrence Berkeley National Laboratory

Anubhav Jain vacationing in Hanoi, Vietnam. Dr. Jain recently received the 2014 early career award for innovative use of HPC from DOE’s National Energy Research Scientific Computing Center.

Anubhav Jain vacationing in Hanoi, Vietnam. Dr. Jain recently received the 2014 early career award for innovative use of HPC from DOE’s National Energy Research Scientific Computing Center.

Anubhav Jain, a research scientist/chemist in Berkeley Lab’s Environmental Energy Technologies Division, received the 2014 early career award for innovative use of HPC from DOE’s National Energy Research Scientific Computing Center (NERSC). Jain’s work focuses on discovering new materials using high-throughput computations, a process he helped speed up by developing the FireWorks software. FireWorks is free, open-source software for defining, managing and executing scientific workflows. As one of DOE’s Computational Science Graduate Fellows, Jain first came to Berkeley Lab in the summer of 2010. After graduating from MIT in 2011 with his Ph.D. in materials science and engineering, Jain returned to Berkeley Lab as a Luis W. Alvarez Postdoctoral Fellow.

1. How did you get started in computational materials science?

Anubhav Jain: I studied applied physics as an undergraduate, but I noticed that the materials scientists were having a lot more fun working on some very important problems while I was doing problem sets. When I entered graduate school, I was planning to research biological materials, like artificial tissue or drug delivery devices. That didn’t work out, so I became interested in high-throughput computing to explore materials for batteries. It’s not like I had a grand plan, but it worked out well. And I actually had done something somewhat similar at a prior internship at Brookhaven National Lab, where I wrote software to automate protein crystallography experiments for high-throughput.

2. What are the major challenges or roadblocks that scientists and researchers are working on in high-throughput computing?

Anubhav Jain: I work on a lot of different projects, and many of them need a large amount of small computing, which is an increasingly common situation. For example, the Materials Project has over a hundred thousand calculations on solid materials, and we have an Electrolyte Genome project running tens of thousands of calculations on small molecules, and much of the same thing is happening in biology and cosmology projects at the lab and elsewhere. While there is a lot of workflow software already out there, they weren’t really capable of some of the more dynamic and error-correcting things we were doing and weren’t built on tools like Python and MongoDB that a lot of scientists are switching to.

In our materials work, we have calculated various properties for about 50,000 different materials, although each material requires several distinct calculations. Each of these calculations runs on from 100 to 1,000 “CPU-hours,” which would take a day to a week on a personal computer but is still small by supercomputer standards. But in all, we have used 25 million CPU-hours at NERSC. To do this efficiently and to be able to track all these jobs, we spent a lot of time on the workflow; I led the development of a software package called FireWorks that automatically launches these small jobs and keeps the computer’s queues full of jobs waiting to run. This in itself is simple, but because we included some special capabilities, this software works well for other scientists with similar situations. Now FireWorks is starting to take off as others adopt it. Several academic groups already use it, we’ve gotten interest from a European consortium on multiscale modeling, and at a recent meeting I was surprised to hear staff from DOE’s Joint Genome Center are thinking of using FireWorks to manage their computations.

3. What are you working on now that you are most excited about?

Anubhav Jain: I’m really excited about working with the Joint Center for Energy Storage Research (JCESR) because we are using our software to design new materials for better batteries. For instance, we have computationally screened thousands of potential “multivalent” cathode materials. Such multivalent batteries can potentially store more energy than lithium ion batteries because multivalent ions like magnesium and calcium carry two electrons, whereas each Li ion carries only a single electron. Some of these materials are now being tested in the lab. We’re having similar experiences with a different project investigating thermoelectric materials, which can convert waste heat to electrical energy. We are working with a group at Caltech on potential new materials and they are making some of the ones we designed on the computer.

It’s always exciting when things jump out of the computer and into real life – something that’s happening more and more in materials science.

4. What do you like most about working in computational materials science?

Anubhav Jain: I like programming and I’m interested in the mechanics of open source software, where you can get the community involved. A great thing about the computational materials science field is that each year, we get more computing power and better theories, so our capability to actively guide and participate in the materials design problem is always to becoming more powerful. And it’s exciting when people come out of the woodwork and tell you that they are using your software.

5. What research are you interested in other than your own?

Anubhav Jain: Although I work on materials science software, I’m also interested in the computer science side of software. Like, how does open source Linux compete with billion-dollar commercial software? It’s really interesting to see how software – if built and managed correctly – can scale without huge budgets.

6. How do you like to unwind away from your job?

Anubhav Jain: I often switch between hobbies. A month ago, I was really into graphic novels. Now I’m trying to relearn how to play the piano. I used to compose music in high school and college. I like a lot of different kinds of music – I listen to techno trance when I’m programming, but I also like classical and jazz. My favorite band is The Smashing Pumpkins. I also enjoy traveling with my wife, although I don’t really enjoy traveling to conferences.

7. Who most inspired you to become a scientist?

Anubhav Jain: My dad is a physicist at Brookhaven National Laboratory, so obviously he inspired me to a great degree. I think another good question is why have I remained a scientist, especially since in the computer science world there are industry jobs requiring similar skill sets that are also very attractive. The answer is that while scientists have many day-to-day struggles, the big picture is often really unique and mind-boggling. I mean for JCESR, we’re using some of the world’s largest computers to design new materials starting at the atomic/molecular level with the goal of making electric vehicles and large grid storage mainstream, thereby having massive implications on world climate and energy use. How many other people are building that app?

8. What is your favorite sci-fi movie?

Anubhav Jain: I think the visuals in “2001: A Space Odyssey” are the best of any sci-fi movie. Do you know Danny Boyle’s movie “Sunshine?” The first half is amazing, and the second half is terrible.

9. Which famous inventor or scientist do you admire most and why?

Anubhav Jain: I think many of us are biased toward the scientists we hear about the most, like Richard Feynman and Albert Einstein. I recently read a biography of Marie Curie, called “Radioactive,” and found her life inspiring. The book had drawings and text, like a children’s book, but it was for adults. The colors in the drawings were amazing and reminded me of Marc Chagall’s work. I also learned that a duel was once fought over Marie Curie. Fortunately no one died.

I also like Elon Musk, who is very impressive. He became a billionaire as co-founder of PayPal and other companies and has used his money to create SpaceX (a space company), Solar City (an energy company), and Tesla (which builds electric cars). He is really everywhere.

Also on my list is Andre Geim, who shared the 2010 Nobel Prize in Physics for his research on graphene, a new form of carbon that is the thinnest and strongest. I actually don’t follow his research much, but it was interesting to read that in his laboratory, he gives the staff free time during “Friday night experiments” to explore crazy and offbeat ideas, and their work in graphene came out of that research freedom.

10. What’s the coolest thing about your work?

Anubhav Jain: I get to interact with a lot of different projects and a lot of different people, which allows me to learn about a lot of different topics. But the really cool thing I have is access to the supercomputers at NERSC and I get to run millions of processor-hours on those supercomputers. Not many people get to do that.