David Shmoys Is a co-Principal Investigator on Cornell's Sustainable Computing Grant
Sustainability is defined in a 1987 United Nations Report as "development that meets the needs of the present without compromising the ability of future generations to meet their needs." If Operations Research (OR) is defined as "the application of mathematics and computation to the optimal allocation of scarce resources, over time and under uncertainty," then OR has a role to play in dealing with growing sustainability concerns.
David Shmoys, an ORIE professor since 1989, recognizes this role and has joined with more than a score of researchers from six institutions in a major effort to promote Computational Sustainability as a new field of research and applications. He is co-Principal Investigator on a $10 million National Science Foundation (NSF) grant to pursue "computational methods for a sustainable environment, economy, and society."
The funding will support a ambitious research agenda spanning conservation, biodiversity, renewable energy and the "balancing of socio-economic demands and the environment," according to grant documents. The team has identified an initial eight interdisciplinary research projects as "building blocks" for what NSF calls an "Expedition in Computing." Shmoys is particularly involved in two of them, one dealing with large scale logistics planning for biofuels such as ethanol and the other with designing affordable migration corridors to be acquired by conservation organizations for use by wildlife such as grizzly bears in the Northern Rockies. The grant provides for more such interdisciplinary projects to be added in the future, "nurtured" by designated "application/synthesis facilitators" who represent specific application areas and computational approaches, according to the grant proposal.
Shmoys notes that ORIE is involved in the "transformation of raw data into quantifiably modeled decisions." The availability of data for models has increased dramatically with the ubiquity of information technology. This trend led to the incorporation of Information Engineering into the mission and name of the School and is an important component of Computational Sustainability as well.
"The essence of what that NSF sought in 'Expedition' proposals is to push the limits of computation in a fundamental way," says Shmoys. "For our proposal we coined the term 'Computational Sustainability' to connect a wide swath of problems -- in areas like biodiversity, energy issues, and ecological resource management -- for which mathematical models are inherently computational in nature and are of a scale beyond what can be solved by currently known computational methods." He acknowledges that there is already a well-established OR presence in areas such as forestry and fisheries, but that Computational Sustainability differs in providing "equal emphasis on computational tools and on the models that relate to the areas of application."
In addition to his research participation, Shmoys has a key role in the administration of a new institute that the grant makes possible, The Institute for Computational Sustainability. He is Associate Director of the Institute, housed on the 5th floor of Upson Hall -- which connects to the 4th floor of Rhodes that houses part of ORIE. The Institute is headed by Carla Gomes, who has a joint appointment in Computer Science, Computing and Information Sciences, and Applied Economics and Management. The Institute provides a management structure for the use of the grant, and is expected to be a continuing entity beyond the unusually long 5 year duration of the grant. Gomes notes that Shmoys "has already been playing a major role in the NSF Expedition with his active involvement, from the development of the vision through the formulation of the proposal to the establishment of the institute."
The biofuels logistics project that Shmoys and others are working on addresses the ambitious goal for renewable fuels mandated by the 2007 Energy Independence and Security Act. Accomplishing the goal, a five-fold increase from the current level by 2022, entails modeling spatial, technological, and economic elements and then determining optimal values for a myriad of decision variables, taking into account the dynamic and stochastic elements of the problem as well as the diverse array of players - economic agents - such as households, landowners, ethanol producers, petroleum refiners, and food producers. The resulting computational problem lies beyond the scope of current methods, and so is typically dealt with using small scale models, equilibrium arguments, and simple algorithms. However, as the grant proposal points out, an optimization based approach "can make the difference between economic viability and failure."
|An example of possible migration corridors connecting the habitat areas shown in light green.|
The wildlife migration problem, which is significant in the sustainable preservation of species, entails the design of conservation corridors, such as those contemplated among three disconnected grizzly bear habitats in Idaho, Wyoming, and Montana. It has already been shown that deciding which of some 13,000 land parcels should be acquired for the corridors can be informed by a network-based computational model that consumes more than 10,000 hours of computer time. Preliminary results indicate that the model can yield dramatic reductions in land acquisition costs. The grant will permit Shmoys and others to generalize this approach to additional species and adding levels of complexity to the model, taking into account the high random variability of the habitats, the multiple species, the dynamics of the species and their movements and migrations.
The increasing levels of complexity involved in the management of Earth's natural resources pose a challenge to Computer Science and Operations Research as disciplines, according to Shmoys. In particular, complex decision models require advances in the mathematics of constraint reasoning, optimization and stochasticity, areas that are of interest to Shmoys and some of his Institute collaborators. Meeting these and other challenges -- such as the study of computational problems as natural phenomena; dealing with highly interconnected problem components via distributed computation; handling multiple time, space and geographic scales by using dynamic models; and using machine learning and statistical modeling techniques to cope with uncertainty and large volumes of data -- mean that work on Computational Sustainability has the potential to transform basic research in the disciplines, according to grant documents.
Noting that the new Center is highly multidisciplinary, Gomes says that "obviously there is a whole community at Cornell interested in sustainability problems, but many researchers in fields other than Computer Science and Operations Research are not fully comfortable with computational thinking." On the other hand, "sustainability is largely uncharted territory for CS and OR, and the nature of the problems in the area, particularly dealing with future outcomes, leads to challenging computational issues."
In addition to establishing the new field of Computational Sustainability and the "development of computational methods to alleviate sustainability problems," the grant program's "Expedition" incorporates education and outreach components, and a mission to increase diversity by "bringing in a new generation of students not traditionally drawn to the underlying disciplines."
The Center for Computational Sustainability complements the work of another University-wide organization, the Cornell Center for a Sustainable Future (CCSF). Gomes says that Professor Frank DiSalvo, who heads CCSF, "has been supportive from the outset of our efforts leading to the new grant."
For additional information, click here.