Seven Ph.D. candidates make the transition from student to colleague

At a graduation ceremony on commencement weekend, doctoral students and faculty advisors participated in the ancient ritual in which the hood is placed on the student’s academic regalia to mark the awarding of the Ph.D. degree. 

Professor David Williamson, ORIE’s Associate Director for Graduate Studies, noted that since research is the act of “doing something that was not known to be doable, casting light into formerly dark places,” the students “have doubtless known both the thrill and the joy of discovery, as well as the gloom and self-doubt caused by hard work that led only to a seeming dead end." He said that the faculty are “enormously proud of our graduates,” who “arrived as our students, but are leaving as our colleagues.”  Seven new colleagues were honored at the ceremony.  Applications of their research range from information filtering to energy transmission and health care to web marketing and finance. 

Andrey Krishenik, a graduate of the Moscow Institute of Physics and Technology, completed his Cornell Ph.D. under the supervision of Professor Andreea Minca, with him at right. In his thesis, he explored two highly relevant aspects of financial mathematics.  He used game theory to determine that there is a lower bound on how much liquid net worth a company must have before its inability to sell assets due to a lack of buyers (as happened for some during the recent financial crisis) results in defaulting on its obligations.  He also developed a model that addresses the question of why it is in the interest of shareholders for a company to holdsignificant amounta of both cash and debt on their balance sheet at the same time, rather than using the former to pay off the latter and reduce taxes and interest. He will be joining the Guggenheim Partners Investment Company as a Quantitative Researcher.

James Davis, who grew up in the small New Jersey farming community of Vineland, completed a bachelor’s degree in mathematics and philosophy at Rutgers University-Camden NJ after transferring there from Cumberland Community College.  He had enrolled in the community college from an early career as a cook with an inquisitive mind and a high school degree.  His Ph.D. research, advised by Professor Huseyin Topaloglu and Professor David Williamson, uses customer choice models - specialized mathematical descriptions of how customers behave - as the basis for algorithms that automatically make choices about which items a retailer should optimally display to its customers.  His algorithms can be efficiently employed in settings where millions of items are available.  Davis, who particularly enjoys building things, will be an Assistant Professor at the University of Illinois at Urbana-Champaign.  At left, Professors Topaloglu and Williamson adjust his hood.

Jacob Feldman, from Bethesda MD, holds a bachelor’s degree in mathematics from Harvey Mudd College.  His thesis focuses on questions similar to those James Davis addressed, determining which assortment of items a retailer should present to customers, though with a different customer choice model and with additional constraints, for example to forestall cannibalizing sales of other, more profitable products.   Such problems are known to become forbiddingly difficult to compute to optimality as the number of products grows, and so are typically approached with heuristic methods whose results cannot be guaranteed.  Under the supervision of Professor Huseyin Topaloglu, Feldman developed a way to find upper bounds on the optimal expected revenue, thereby quickly determining how close candidate solutions are to the best possible, which he shows through computational experimentation to be quite close indeed for certain heuristic approaches.  At right, Topaloglu places his hood.  Jake will join the faculty of the Olin Business School at Washington University in St. Louis as an Assistant Professor. 

Nicholas James, from Tallahassee FL, received a bachelor’s degree in mathematics from the University of Florida.  Under the supervision of OR field member David Matteson, he developed data analysis algorithms to deal with situations in which the properties of the observations in a statistical time series may change at points in time.  For example, economic events, such as the recent subprime mortgage crisis, can drastically change market behavior and therefore data series that depend on market behavior. However standard methods are based on the assumption that the underlying properties of the series are similar throughout, ignoring such change points. 

Using as few mathematical assumptions as possible, the algorithms James developed correctly partition data into consistent segments.  James demonstrated the statistical and computational advantages of his new methods by using novel mathematical proofs and creating a new software package.  His methods have been implemented at Twitter, where he spent an internship.  James is joining Google as a Software Engineer.  Professor Matteson was unable to attend the ceremony; Professor David Shmoys placed James’ hood and congratulates him at right.  

Ravi Kumar, from a suburb of Delhi, India, has Bachelor’s and Master’s degrees in mechanical engineering, the former from the Indian Institute of Technology in Delhi and the latter from the State University of New York at Buffalo.  At Buffalo, he did research on estimating the motion of tumors, a concern in radiation therapy.   In the four years between his earlier degrees, he worked on various aspects of electrical substation design and operation.  His Cornell thesis research, under the supervision of Professor Mark Lewis, deals with efficient use of energy in the operation of telecommunications networks.   In particular, he used queuing theory and stochastic dynamic programming to develop and analyze models and methods that determine near-optimal policies for scheduling transmission of different traffic streams over time and for adjusting transmission rates in response to traffic congestion.   He showed that these policies can computed efficiently and can lead to substantial savings in energy cost without sacrificing quality of service. At left, Professor Topaloglu congratulates Kumar.  He placed Kumar’s hood on behalf of Professor Lewis, who was unable to attend. 

Eric Cao Ni grew up in Hangzhou, China, an ancient city said to be the eastern terminus of the Silk Road.  He holds a double degree in engineering and economics from the National University of Singapore.   Under the supervision of Professor Shane Henderson, he developed a framework for the efficient performance of supercomputers and cloud computers in carrying out the computationally intensive algorithms necessary to make optimal decisions in an uncertain environment.   His research is in the area of simulation optimization, in which choices are made among competing strategies, evaluating each of which may entail a complex simulation – for example planning the location and deployment of a fleet of ambulances in a city where the frequency and arrival times of emergency calls are hard to predict.  Using techniques from probability theory, he has shown how such high performance computing resources can be used to remarkably increase the speed at which the simulation-based algorithms operate, thereby expanding the range of problems that can be solved within a given computing budget.  He will join an investment bank in London as a Quantitative Strategist. 

Xiaoting Zhao, from Fujan, China, graduated from Smith College in 2009 with a degree in physics and economics, later receiving a baccalaureate graduate certificate in mathematics and statistics from Smith.  Her thesis develops models for filtering the information in systems such as arXiv.org, the online repository of scientific articles established by Cornell physicist Paul Ginsparg and now primarily housed at Cornell.  With more than 1 million full-text articles, growing at the rate of 8000 new articles per month, discovering the most relevant recent information at arXiv.com is a challenge for scientists.  They can be aided by an information filtering system that automatically pre-processes incoming articles and selects some to forward to the scientist.  Under the supervision of Professor Peter Frazier, Zhao developed novel policies for filtering and ranking articles in a way that balances exploration, i.e. expanding the range of articles provided in order to gain better insight into those evolving preferences with exploitation, i.e. providing articles that respond to past user preferences.  At right, Professor Frazier congratulates Zhao, who is joining the Dynamic Sciences Lab at the enterprise software company Infor as a (data) Scientist.