James Simons’s Foundation Starts New Institute for Computing, Big Data

By KENNETH CHANG

Part of the work at the Flatiron Institute involves complicated numerical simulations of exploding stars, as shown here.CreditChris Hayward/Center for Computational Astrophysics/Flatiron Institute; and Phil Hopkins/California Institute of Technology/FIRE Collaboration

A new private research institute financed by the billionaire James H. Simons in New York will develop software tools and apply cutting edge computing techniques to science often not possible in academia and industry.

Researchers this week will start moving into the Flatiron Institute, across the street from the downtown offices of the Simons Foundation, a nonprofit that finances basic science research.

Mr. Simons, a mathematician turned wealthy hedge fund manager and now philanthropist, and his wife, Marilyn, the foundation president, hope the new institute will fill an overlooked niche.

Computers have been a fixture for decades in astrophysics and many other fields of science. But typically, the computer programs are written by graduate students, often abandoned after they finish their programs. “Those people aren’t great coders, for the most part,” Mr. Simons said.

At the Flatiron Institute, a good fraction of the staff will be professional computer programmers, producing software not only for the in-house scientists but also available for anyone else who needs it.

Continue reading the main story

“The computational motif that runs through the organization, there really isn’t anything quite like that,” Mr. Simons said. “A university department cannot hire programmers.”

Photo

James and Marilyn Simons. CreditFred R. Conrad/The New York Times

Ms. Simons said the impetus for the institute evolved from a brainstorming workshop about “what we might do to help move the needle in science.” A Belgian physicist and mathematician, Ingrid Daubechies, suggested an effort to develop better computational tools.

“She said there is a real need for analyzing big data,” Ms. Simons said. “She felt these kinds of technologies and approaches would be really helpful in addressing some of the scientific problems we had now.”

The first area of focus was computational biology.

The foundation reached out to Leslie F. Greengard, a mathematician at New York University who also has a medical degree. He gave a talk at Simons outlining some possibilities for what a computational biology endeavor might look like.

“Jim said, ‘Let’s have lunch tomorrow,’ ” Dr. Greengard said. “Then, ‘Why don’t you come and start this?’ ”

That was three years ago, and Dr. Greengard accepted. Most of the work of the Simons Foundation goes to grants to scientists around the country; this was its first foray into hiring its own scientists and embarking on its own research.

One project is developing software for collecting and analyzing the electrical pulses recorded from electrodes implanted in animal brains. Originally, it was just a few electrodes. Now it is hundreds and soon it will be thousands, Dr. Greengard said.

Photo

A supercomputer simulation of colliding galaxies. CreditChris Hayward/Center for Computational Astrophysics/Flatiron Institute; and Phil Hopkins/California Institute of Technology

From the voltages, scientists need to identify which neurons are firing when, and that is complex computational challenge. Currently, different research groups use their own custom-written computer programs, making it difficult for scientists to compare their results with others, or to reproduce them.

The Flatiron software will be available for all scientists.

“We can build teams where methodology development is not sort of secondary to asking the science question,” Dr. Greengard said. “It’s critical to asking the science question.”

Dr. Greengard’s efforts, until now housed in the foundation offices, proved successful, so his group and three other centers will be working at the institute.

That led to the invitation to David N. Spergel, a Princeton astrophysics professor, to start a similar effort for astrophysics, and a third group focusing on materials science is in the works. Eventually, the institute will add a fourth center — topic not yet decided — and grow to about 200 full-time employees with an $80-million annual budget.

The foundation also hopes to remove institute scientists from the typical academic pressures of churning out journal articles in the chase of tenure and scrambling for the next grant to finance their work.

Because of the competition for grants, scientists often propose projects that they know will work rather than on ideas that are more adventurous with potentially greater payoffs.

“These are really interesting questions, and we can think longer than the three-year grant cycle,” Marilyn Simons said. “They can tackle tough questions and put the time in that’s necessary.”