In a time of generally tight budgets, SIAM’s Committee on Science Policy (CSP) held its regular spring meeting on March 28 and 29. The committee met with key decision makers at federal agencies and in Congress to better understand the environment for research, influence congressional legislation and federal programs related to applied mathematics and computational science, discuss new research initiatives, and provide input on issues of concern to the SIAM community.
Among the guests were Steve Binkley, associate director of the Department of Energy Office of Science for Advanced Scientific Computing Research (ASCR); Michael Vogelius, director of the National Science Foundation (NSF) Division of Mathematical Sciences (DMS); Jim Kurose, assistant director of NSF for Computer and Information Science and Engineering (CISE); Melissa Flagg, Deputy Assistant Secretary of Defense for Research; Chuck Romine, director of the National Institute of Standards and Technology (NIST) Information Technology Laboratory (ITL); and Ron Boisvert, head of the Applied and Computational Mathematics Division of ITL. The first day included presentations by and discussions with the invited guests, along with a general presentation on background by SIAM’s representatives at Lewis-Burke Associates LLC. The second day featured visits with key staffers on Capitol Hill for further discussions.
As an interdisciplinary organization of members interested in applied mathematics and computational science, SIAM routinely holds discussions with several groups within the NSF at its CSP meetings. This meeting featured both mathematics and computer science.
From left to right: Sven Leyffer, David Levermore (chair), Philippe Tondeur, and Hans Kaper of the SIAM Committee on Science Policy in Washington D.C. for the spring meeting.
It is interesting to note that mathematics is currently a division (DMS) within the Directorate for Mathematical and Physical Sciences (MPS), while computer science has a full directorate encompassing all of CISE. Thirty-five years ago, mathematics and computer science were two sections within a single division (the Division of Mathematical and Computer Sciences). Around 1983, then-division head Ettore (Jim) Infante laid the plans for separating mathematics and computer science. Today, the computer science directorate has a budget of $936 million, whereas mathematics remains a division with a $234 million budget, with applied and computational mathematics receiving a fraction of that figure. Given the interdisciplinary nature of SIAM—roughly 60% of SIAM’s academic members are in mathematical sciences departments—SIAM’s CSP must cast a wide net to learn about programs of interest to members.
As background, it should be noted that the Obama administration proposed a fiscal year (FY) 2017 President’s Budget Request with an unusual feature – the request included both a proposed amount to be funded through the usual discretionary funds which would need to pass in appropriations legislation, and a proposed increment from mandatory sources (i.e. money that would require Congressional approval but would bypass the usual appropriations process). While interesting projects may have been proposed for the mandatory funding, it seems unlikely that mandatory-type funding will be provided.
DMS Division Director Michael Vogelius spoke about the budget situation for mathematics and the plans for FY 2017. Noting that recent budgets have spread increases fairly evenly among the MPS divisions, Vogelius said that not including mandatory funding, the proposed budget for FY 2017 would still leave the DMS budget less than its 2012 budget. Along with DMS, MPS has placed an emphasis on core research (applied mathematics and computational mathematics are among the DMS core programs). It was noted that mathematics does play a role in various Foundation-wide initiatives but focuses on fundamental research.
Vogelius said the DMS seeks suggestions from the community for initiatives that would be natural fits. Data science is one such area, and it is hoped that mathematics and statistics can play a larger role in big data initiatives at the NSF, for example, in algorithmic foundations of data science.
Vogelius also noted that the DMS remains committed to the institutes program and to “math+x” institutes that reach out to other disciplines; the DMS is seeking ways for biology and mathematics to mutually fund efforts, especially in areas of biology where mathematics has transformative potential.
Other mentioned initiatives include cybersecurity (mostly in CISE) and optics and photonics. SIAM is hosting an optics and photonics workshop with the NSF at the SIAM Annual Meeting this July to help familiarize a wide range of researchers with the NSF “Optics and Photonics” call for proposals, and to encourage collaborations among researchers in this field. Further topics of discussion included the math institutes and joint programs, and the role of mathematics in the Strategic Computing Initiative. Vogelius also spoke about internships for non-academic career tracks, noting as an example the Enhanced Doctoral Training (EDT) program, which includes elements to better prepare participants for non-academic careers. Additional discussions regarding internships may take place within the DMS, following the NSF-IPAM Mathematical Sciences Internship workshop held last September.
Jim Kurose of CISE initiated the discussion on programs and plans within CISE with a presentation outlining several initiatives. Current projects include big data, robotics, and the BRAIN initiative, which encompasses several CISE programs. New efforts—or ones growing substantially in FY 2017—include the National Strategic Computing Initiative, Smart and Connected Communities, Data for Scientific Discovery and Action (D4SDA), and Computer Science for All, the computer science education initiative. Kurose noted that in contrast to mathematics, where roughly 60% of research funding comes from the NSF, computer science receives 82% of federal academic funding from the NSF. The request for FY 2017 is $995 million for CISE, representing a balance among core, cross-cutting initiatives, and education.
Of course, core programs remain important in CISE. Kurose concluded with, “Submit your best ideas. It’s the heart of what we do.”
Another major area for applied mathematics and computational science is the DOE Office of Science, which is undergoing major budgetary changes. Steve Binkley discussed the Advanced Scientific Computing Research (ASCR) program, specifically focusing on applied mathematics and exascale. Binkley noted that DOE is the dominant funder of basic research in the physical sciences, and the applied mathematics program is a non-trivial source of funding in that area. The ASCR budget is proposed to increase by 7% ($42 million). Roughly half of the increase would go to basic research, and half would go to facilities.
A transformation is expected in the exascale program’s organization. Current exascale efforts within the applied mathematics program and across other ASCR programs will be pulled out into a single Exascale line called the Office of Science Exascale Computing Project (SC-ECP). As a result, $10 million will move from applied mathematics to SC-ECP. Funded programs will presumably go with this money in the short term, but this will clearly represent a shift towards programs that are more applied in the longer term. The hope is that applied mathematics and other research programs will be built back up in the future.
Binkley made an interesting observation about the role of software and algorithms in improving computer performance. He noted that software (parallelism) has contributed increasingly more than hardware (Moore’s law) to improvements from petascale to exascale. This trend will only continue, given the difficulties in moving to 10-nanometer technology and beyond.
The DOE is also considering alternative computing models, including quantum computing, with a focus on scientific computing. These models will cause this relatively small part of the budget to grow in coming years.
Chuck Romine presented information on NIST’s relevant programs, most of which are intramural research programs within ITL rather than university research-funding programs. ITL has seven divisions, one of which is Applied and Computational Math (others include two in cybersecurity and one each in core computer science, advanced networking, statistics, and information science). NIST hosts many guest researchers who work (without remuneration) with lab scientists.
NIST will play a role in the National Strategic Computing Initiative. Some of the thrusts within NIST include math of metrology, high-performance computing and visualization, advanced materials, and quantum information science. Probably the piece most well-known to the mathematics community is the Digital Library of Mathematical Functions (DLMF).
Finally, Melissa Flagg talked about basic research at the DoD, emphasizing the major role of modeling and verification. She asked how we establish trust in our projections, noting that this is a big problem in affordability of systems, for instance.
Much of what she described as research needs and thrusts concern autonomous systems and human-machine collaboration. The goal is to make faster decisions at the speed of a machine while maintaining trust, ultimately resulting in better decisions and allowing humans to do what they do well.
Flagg discussed the DoD’s future reliance on mathematics in areas such as enhancing cybersecurity defenses, modeling and simulation, computational methods for data analysis, and new analysis techniques to harness the power of big data. The DoD is looking for new ideas and big areas where mathematics can play a major role.To support FY 2017 appropriations and robust federal investment in applied mathematics and computational science, CSP members met with congressional staff from their state delegations and key congressional committees on the second day. CSP members provided information on the importance of research funding at the NSF, DOE, DoD, and the National Institutes of Health (NIH), and underscored the value of mathematics and computational science research as well as related education and workforce issues.
The CSP will meet with additional agency representatives and discuss SIAM priorities and advocacy strategy for 2017 this fall. We encourage you to lend your voice in support of federal investments in math and computational science by meeting with congressional officers for your state. If interested, please contact Miriam Quintal, SIAM’s Washington liaison, at [email protected].
||Jim Crowley is the executive director of SIAM.