Mark Lewis Is now a full professor

Cornell Provost Kent Fuchs has approved the promotion of Mark Lewis to the rank of Professor.  Lewis joined ORIE as an Associate Professor in 2005 after serving as an Assistant Professor at the University of Michigan.

Lewis received undergraduate degrees  in mathematics and political science in 1992 from Eckerd College and went on to receive an M.S. in theoretical statistics from Florida State University and a Ph.D. in industrial and systems engineering from Georgia Institute of Technology, in 1998.  After a post doctoral year in the Centre for Operations Excellence at the University of British Columbia, he joined the Industrial and Operations Engineering Department of the University of Michigan as an Assistant Professor.

ORIE Director Adrian Lewis said "I am delighted that Mark has joined the ranks of the full professors in ORIE.   He has provided undergraduates, graduate students and the academic community with fundamental insights into stochastic modeling and analysis, and has built strong ties with the world of applied operations research as well.  He has also worked hard to enhance the experience of underrepresented minorities among the College of Engineering's faculty, our incoming PhD students, and his professional colleagues."

Lewis said "I am honored by the promotion and look forward to continuing to grow with my colleagues and students."  He noted that "even before I arrived at Cornell I held the School of ORIE in the highest esteem, so you can imagine how excited I was when I was asked to join the faculty in 2005.  Since arriving, the School,  the faculty and indeed the College of Engineering have exceeded my expectations both professionally and personally." 

At Cornell, Lewis teaches undergraduate and graduate classes on stochastic processes, as well as a graduate course on queueing theory and an advanced graduate course on stochastic dynamic programming. 

Research Activities

With support from the National Science Foundation through a Presidential Early Career Award for Scientists and Engineers and two other grants, as well as a grant from MITRE Corporation, Lewis conducts research on a variety of problems, many arising from the random arrival over time of customers, product parts, vehicles or telecommunications signals to be serviced by one or more facilities of limited capacity, with possibly random service times.  Such problems fall into the domain of queueing theory, which is a branch of stochastic processes.

Among the problems Lewis has analyzed are several that involve multiple facilities (i.e. workers or machine servers) operating in parallel and the ability for the service recipient to switch or be switched among them.  

Such problems arise in hospitals (where patients may be assigned to beds in different specialty units or floors in order to balance the workload of cross-trained nurses), at call-centers (where operators may handle calls of different types or even for different clients), and in so-called reconfigurable manufacturing systems (where jobs in process may be rerouted to balance workload or alleviate congestion due to machine failures).   In such systems, routing may be done by individual workers, by system operators, or even by customers, who may jump queues or abandon the system, e.g. by hanging up the phone, altogether.   Proper design and control of such systems requires an understanding of their dynamics.

Although queueing theory is now more than 100 years old and covers a huge array of models and applications, analysis of these new problems has required new insights.

Lewis has attacked these and other such problems using the tools of a subject called Markov decision processes (also known as stochastic dynamic programming), which deals with finding optimal policies for processes that move from state to state governed by transition probabilities and the actions of decision makers.  As the problems get increasingly complex and the number of possible states and actions grows, the possibility of computing optimal decisions (often in the form of policies that specify what action to take in each state) is limited by the so-called "curse of dimensionality" that can outrun the capacity of the fastest available computers.  

In the face of this limitation, Lewis has been able to provide new theoretical insights that in turn lead to methods - heuristics - that provide approximate solutions, and to simple management strategies.  (Other approaches to the curse of dimensionality researched in ORIE include simulation and approximate dynamic programming.) 

Lewis has also worked on models representing vehicular traffic, repair versus replacement in maintaining machines, cash management, revenue management, wireless ATM networks, and the provisioning of airline meals. 

Theory and practice

In common with several other ORIE professors, Lewis enjoys working on a mix of theoretical and applied problems.  "One challenges me technically, while the other keeps me grounded," he has said.  His interaction with non-profit corporation MITRE exemplifies this range.  As a result of advising Master of Engineering projects, with MITRE as client, that dealt with planning the evacuation of a population center in the event of a large scale catastrophic event,  Lewis received a grant from MITRE to develop new theory and methods of reallocating resources over time and across municipalities during such eventualities.   

Professional leadership

Lewis is vice chair of the Applied Probability Society of INFORMS, the Institute for Operations Research and the Management Sciences.  He will become chair of the organization in 2012, succeeding ORIE professor Shane Henderson.  He is on the editorial board of several journals.  He is founder (in 2004) and current president of the INFORMS Minority Issues Forum.