A Toolkit in 0S and 1S

JOHN H ARVARD'S JOURNAL A Toolkit in 0s and 1s Efthimios Kaxiras (left) and I! "#$ %&"$ 1930s, Harvard graduate stu- Hanspeter P!ster in front of a portion of Howard Aiken’s dent Howard Aiken was dreaming of a Mark I computing machine computing machine. His doctoral thesis in physics required numerical solutions to nonlinear di'erential equations, involving tedious calculations beyond the capacity of calculators. It was clear to Aiken that these mathematical operations could be standard- ized and mechanized, so he began designing a machine to carry them out. The potential, he thought, was immense: he could already envision applications in mathematics, science, and even sociology. Certain areas of science, he argued, were at an impasse, lim- ited by the mathematical power of computing equipment. Aiken’s computing machine, now bet- ter known as the Harvard Mark I, was built by IBM and presented to Harvard in August 1944. In the next 15 years, the Mark I and its successors ran calculations for scientific research as well as for the in scattered academic fields, but was em- modes of theory and experiment. U.S. military, including a series of implo- ployed mostly on an ad hoc basis, lacking an But harnessing the potential of compu- sion calculations for Los Alamos National academic foundation of its own. tation requires more sophisticated train- Laboratory (later revealed to be part of Now, nearly 70 years after the Mark I ing. To that end, the School of Engineering the atomic bomb project). Aiken himself, was built, computational science has come and Applied Sciences has launched a new as a professor of applied mathematics, into its own, says Efthimios Kaxiras, Van degree program, the master of science in directed Harvard’s Computation Labora- Vleck professor of pure and applied phys- computational science and engineering tory—built to house the Mark I, and later ics and founding director of Harvard’s In- (CSE), to provide rigorous training in the renamed in his honor—until 1961. In 1947 stitute for Applied Computational Science use of computational methods to solve and 1949 respectively, he developed a mas- (IACS). No longer do computers merely research problems. The 28 students in its ter’s, and then a doctoral, program in com- speed up processes that could be done inaugural class, drawn from a pool of 147 puter science, the first of their kind in the by hand, as in Aiken’s time; computation, applicants, began their studies this fall. nation. The age of computing had begun. once just a tool, has become so powerful The year-long program combines a core Yet in the decades following Aiken’s that it is changing the nature of inquiry. of project-based computer science and retirement, large-scale computing largely “The hardware, algorithms, and ideas be- applied math courses with electives that left Harvard behind. Computer science— hind computing have all evolved so that allow students to explore connections especially the theory of computation and computers can now solve real problems,” to other fields, from engineering to eco- the development of software systems— Kaxiras explains. So-called “big data” is nomics and global health. “The program took o' as an academic discipline in the opening new lines of research in fields is meant to be a toolkit,” says Rosalind 1970s, but computational science lagged be- from basic science to advertising; already, Reid, former IACS executive director (she hind, burdened by its roots in calculation. it has given rise—via complex computa- recently became executive director of the Early computing machines like the Mark I tional models, coupled with statistics—to Council for the Advancement of Science seemed like enormous calculators; though developments like election forecasts more Writing). “And it’s a toolkit taught via ap- powerful, they remained merely tools, and accurate than ever before. People are call- plication.” few people thought to study their use. ing computation the “third leg” of sci- The program itself is the product of Computational methodology persisted ence, he says, adding it to the traditional careful design by IACS, which was found- ed in 2010 to oversee interdisciplinary research and educational initiatives related IN THIS ISSU E to computing. “The new degree is ground- ed in real needs,” Kaxiras says. IACS con- 62 Harvard Portrait 67 Brevia vened a panel of experts from industry and 63 The Endowment Rises 69 Sports national research laboratories to assess the 64 Teaching Tech 72 Alumni current research environment and brain- 65 Yesterday’s News 76 The College Pump storm what an academic program could 66 The Undergraduate accomplish. (The advisory board has met

Photograph by Stu Rosner H&()&(* M&+&,-!$ 61 Reprinted from Harvard Magazine. For more information, contact Harvard Magazine, Inc. at 617-495-5746 JOHN H ARVARD'S JOURNAL

periodically since to continue the dia- HARVA RD PORTRAIT logue.) The unanimous suggestion, he says, was to create a master’s degree—a recom- mendation supported by academics on the grounds that computational science and engineering still lack the clear career path and disciplinary focus needed to justify a doctoral degree (though Kaxiras predicts that this will soon change). The Institute itself has begun laying the intellectual groundwork for this new field by developing courses in areas like scientific computing, data science, and numerical methods. “Computational science is always changing, and its needs and topics are changing,” says Kaxiras, “but the key ideas remain the same.” There’s a need, he explains, for sophisticated techniques from computer science and applied math that are properly matched to the complex problems at hand. For instance, situations from card games to market investments are often represented by systems of equations impossible for even the most high-powered computers to solve mathematically be- cause of the near-infinite range of possibilities. Kaxiras has therefore developed and taught a course on stochastic optimization: a set of probabilistic methods that tries a large number of possible strategies to come up with a best guess—all that’s possible, he says, for many real-world problems. In computational science, he continues, there’s a need for people who can “marry the hardware to the right software and the Jennifer Lewis right algorithm.” For instance, big data now requires immense computing power, Jennifer Lewis’s engineered materials look to nature as a guide. The new Wyss pro- well beyond what any single processor fessor of biologically inspired engineering uses 3-D printing to build minuscule devic- can accomplish alone. “How do you break es, from microbatteries to synthetic spider webs of threads a micron thick. Now she down a problem to take advantage of par- works to “print” biology, motivated by “a bit of naiveté mixed with a strong desire to allel computing?” he asks. “It requires a dif- bene!t society.” Her lab develops “inks” with functional properties: cell-laden ones to ferent way of thinking from the sequential print 3-D tissues, or conductive inks that "ow through rollerball pens at room temper- mode of thinking we’re accustomed to. ature to draw functional circuits on paper. Lewis works with high-school teachers to in- These developments have elevated the dis- corporate these inexpensive pen-on-paper electronics in their classes, so students can cipline and approach to something with se- explore engineering through circuit design. Her educational interest draws on personal rious intellectual potential.” Wang profes- experience: despite coming from a family of engineers—her father worked for Gen- sor of computer science Hanspeter Pfister, eral Electric, and her sister is a chemical engineer—Lewis !rst encountered materials the current IACS director, has developed science in college at the University of Illinois. She later joined the faculty and taught and taught courses in visualization, com- there for 20 years (after earning her S.D. at MIT), returning to Cambridge for her new puter graphics, parallel programming, and appointment in January. The move back East has given Lewis, an avid basketball player, computational science; this fall, he and Jo- a chance to pick up her squash racket again, as well. She’s also been exploring Boston seph Blitzstein, professor of the practice in through another longtime hobby: one of this !ction fan’s recent favorites is The Dante statistics, introduced a new course in data Club, a whodunit set in Civil War Cambridge, which helped immerse Lewis in her new science that is jointly o!ered by the com- community. She and her partner, Lori Sanders, who also studies biomaterials, live near puter science and statistics departments. the undergraduate Houses, “right in the heart of things,” and Lewis enjoys the intel- Meanwhile, the new degree is prom- lectual stimulation of her new home: “It’s time to stretch and grow in new directions.” ising as much for its applications as its academic focus. “Students come up with

62 N"#$%&$' - D$($%&$' 2013 Photograph by Jim Harrison Reprinted from Harvard Magazine. For more information, contact Harvard Magazine, Inc. at 617-495-5746 T he E ndowment R ises

Harvard’s endowment was valued at $32.7 billion last June Dartmouth investments yielded a 12.1 percent return, too. The 30, the end of !scal year 2013, up $2.0 billion (6.5 percent) from University of Pennsylvania disclosed a 14.4 percent return, rais- the prior year, but still 11.4 percent less than the $36.9 billion ing its endowment to $7.7 billion. The University of Virginia In- peak reported in !scal 2008, just before the !nancial crisis. The vestment Management Company realized a 13.4 percent return. growth, as disclosed on September 24 in Harvard Management And MIT recorded an 11.1 percent return, raising its endowment Company’s (HMC) annual report, was driven by an 11.3 percent to $10.9 billion. investment return on assets—a turnaround from !scal 2012, HMC president and CEO Jane L. Mendillo, noting that her when a modestly negative return (-0.05 percent) and annual and organization is now fully staffed, expressed cautious optimism one-time distributions from the endowment reduced its value about the outlook, given current economic and political condi- by some $1.3 billion. (Endowment distributions, the largest tions: source of Harvard’s operating revenues, accounted for 35 per- Questions abound about !scal and monetary policy here cent of income in !scal 2012.) and abroad…about the impact of new market regulations The !scal 2013 appreciation re"ects the positive investment and participants, and about the prospects for economic return, minus distributions to support University operations and growth across global markets in sometimes shaky political for other purposes, plus gifts received (to be reported in the environments.…However, looking beyond some of the annual !nancial report later this fall). In a year of strong stock- shorter-term issues…we are con!dent that there is plen- market returns, HMC’s public equities (about one-third of assets) tiful opportunity for long-term investors like Harvard. gained 16.3 percent, slightly above their benchmark. Absolute- For a detailed report on the endowment’s performance, see http:// return assets (hedge funds and high-yield bonds, about one-sixth harvardmag.com/endowment. of assets) gained 13.2 percent, nearly double their benchmark. The !xed-income portfolios (now less than 10 percent of assets) Harvard Management Company produced positive returns and yielded the widest performance 2013 Investment Performance margin relative to market results. Real assets (real estate, timber and farmland, and commodities, 25 percent of the investments) Asset Class HMC Return Benchmark Difference were the only category to trail market returns—an anomaly for Return HMC. The overall return was in line with the results of public Public equities 16.3% 14.5% 1.8% pension funds and investors following a more conservative 60/40 Private equity 11.0 10.6 0.4 allocation strategy, relying on public stock and bond holdings (the Absolute return* 13.2 6.8 6.4 former mostly up strongly in !scal 2013, the latter mostly down). Real assets** 7.0 7.2 (0.2) Among peers, Yale reported a 12.5 percent return and 7.8 Fixed income 3.3 (3.4) 6.7 percent endowment appreciation, to $20.8 billion. Stanford’s To t a l e n d o w m e n t 11.3 9.1 2.2 investment return was 12.1 percent; its endowment grew 9.7 *Includes high-yield bonds percent to $18.7 billion as of August 31, the end of its !scal year. **Includes real estate, commodities, and natural resources amazing and innovative ideas about how of his residency and found himself drift- problems to tackle. to apply what they’ve learned,” Kaxiras re- ing back to his undergraduate studies in “Many data-science students come from ports: in the final project for his class, they chemical engineering at Tulane University. fields in science that they naturally gravi- developed methods to optimize everything “I hope to add engineering to medicine,” he tate to, that form how they think and what from class time and room assignments says. As a medical student, he had observed problems they choose to approach,” says (to minimize between-class travel) to re- that much of healthcare is empirical, and Reid. At the same time, she says that sci- source distribution and routes to first-aid he believes it could benefit from a stronger ence and computer science speak di!erent facilities (in case of a major disaster). foundation in computer models and simula- languages, even though computational sci- The inaugural master’s-degree students tion. He has already arranged to work with ence must make use of both. “We’re trying come from diverse academic backgrounds; associate professor of radiation oncology to create multilingual people here,” she about a third earned undergraduate degrees Lance Munn on a research project to cre- explains. Daniel Weinstock, the program’s outside the STEM (science, technology, en- ate computational models of metastasis. assistant director of graduate studies, who gineering, and mathematics) fields. Most He’s not sure yet whether his path will take holds a Ph.D. in theoretical and computa- have taken time o! since college, and a third him back into medicine, or into academia— tional chemistry, points out that computa- already hold advanced degrees in fields as “I think this year will help me refine what tion “is frequently taught for the purpose diverse as finance, sociology, and science I want to do”—but like many of his class- of problem-solving, but only informally. writing. Physician Ryan King, for example, mates, he is interested in learning the ap- The [program’s] courses give students felt unfulfilled in the purely clinical aspect proaches that will allow him to choose what rigorous training in a computational area

H"#$"#% M"&"'()* 63 Reprinted from Harvard Magazine. For more information, contact Harvard Magazine, Inc. at 617-495-5746 JOHN H ARVARD'S JOURNAL that they can then take back to their dis- istic questions,” he says, and he’s excited dents from the graduate and professional ciplines.” Pfister adds that IACS intends by the potential of big data to transform schools, as well as the Extension School. to foster collaborations across schools at epidemiology and public health. Meh- Next fall, a two-year master of engineer- Harvard, and to connect students with rotra, who studied math and economics, ing degree program will join the current alumni and practitioners in industry says she’s “always enjoyed using econom- one-year master of science program in through internships and research projects. ics to understand how people interact, and CSE, and a secondary-field citation is Several students have already worked their impact on the world around them,” also available for other graduate students. in industry, giving them valuable per- and hopes that by strengthening her back- Kaxiras says he receives frequent inquiries spectives on the importance of computa- ground in programming and Bayesian sta- from other institutions seeking to develop tion in the marketplace. After graduating tistics she will be able to explore projects similar programs. “The need is urgent,” he from the University of California, Berke- with social impact in fields like develop- says. “There are so many things happen- ley, in 2011, Anita Mehrotra joined the mental economics, education, and public ing in science, engineering, and the wider technology-consulting firm Accenture’s health. She calls the Harvard program the world that rely on computation, and the research and development labs, where she “perfect mix” of strong applied-math and more people you have who are highly witnessed firsthand the growing need for computer-science classes in an interdisci- trained, but not narrowly focused, the data scientists and statisticians. Peter Bull, plinary setting. “The degree can really take more exciting the possibilities will be.” who graduated from Yale in 2008, worked you anywhere,” says Bull, echoing many of He is happy to see the surge in inter- for five years as a software engineer at his classmates. est. “When I first came to Harvard over 20 Microsoft, where he became interested in “I think these [computational-training] years ago, there was quite a bit of skepti- learning how to manage and gain insights courses are meeting a real need, and proof cism that this type of scientific approach from large amounts of data. He sees some of that is the great number of students could yield interesting results and really connections between data science and who are attending,” says Kaxiras. This advance the scientific endeavor,” he says. his undergraduate focus on philosophy: fall, over 400 students enrolled in the new “I think there’s real potential in the future “Both represent rigorous, technical ap- data science course; more than half were for scientific breakthroughs using compu- proaches to finding answers to human- undergraduates, and the rest a mix of stu- tational approaches. It’s mind-boggling.”

Teaching Tech e*ect on students and hence allow to “accelerate e*orts…to develop ‘ﬂipped them to become more proﬁcient in classroom’ experiences, in which lecture- T!" "#$%&'($) and adoption of online helping students learn. style material is delivered increasingly learning technologies, and other e*orts to He also foresaw a welcome deemphasis online before class, while classroom time enhance teaching, continued briskly as the on pure lecturing and greater use of “more is spent by students and faculty actively fall semester began. Some highlights: active forms of learning, such as problem- engaging together to develop strategies President emeritus Derek Bok, a champion solving, socratic discussion, and group for solving the types of problems students of improved pedagogy, publicly embraced projects”—all productive in “giving stu- will encounter in their careers.” the potential for education technology to dents a deeper understanding of the sub- HILT has also underwritten digital enhance the classroom in a September ad- ject and helping them to develop the skills infrastructure, including the new “vid- dress marking the six-hundredth anniver- of critical thinking and analysis that facul- eo-capture studio” in Widener Library. sary of the University of St Andrews. The ties have long considered the most impor- There, faculty members can use high-def- collaborative nature of preparing an online tant goals of a liberal education.” inition production equipment or experi- course, he said, makes it “less intuitive On-campus curriculum and digital facili- ment with inserting online lectures, learn- and more a product of conscious delibera- ties. Harvard School of Public Health, as- ing exercises, demonstrations, and visual tion” in deciding how best to present each sisted by $300,000 in planning funds from materials into their courses. The Faculty concept and in what order. Moreover, he the Harvard Initiative for Learning and of Arts and Sciences (FAS) and HarvardX, noted: Teaching (HILT)—apparently one in a se- the online learning program, will manage For hundreds of years, only stu- ries of such grants—and substantial phil- the studio. HILT funded FAS’s construc- dents have known much about anthropic support, tion of the “Sci-

Peter K. Bol OMMUNICATIONS how well or badly they are being announced a whole- Box” in the Science C taught.…In an online course, every sale revision of its Center, too: a ﬂex- statement professors make, every master ’s degree ible, open lecture, question they ask, and every answer programs and a new lab, and discussion students give can be recorded. By doctorate in public- space—with built- examining this material, instruc- health leadership. in telecommunica- tors can discover much more about It aims to deploy tions and adjacent what their students are learning “more case-based 3-D printer and easily and what material is giving and ﬁeld-based ‘real tools—in which them trouble. This change alone will world’ learning op- di*erent modes of

make professors more aware of their portunities” and teaching and learn- AND AFFAIRS PUBLIC MITCHELL/HARVARD STEPHANIE

64 N$#"+,"- - D"."+,"- 2013 Reprinted from Harvard Magazine. For more information, contact Harvard Magazine, Inc. at 617-495-5746