Computational Sustainability Computational Methods for a Sustainable Environment

Computational Sustainability: Computational Methods for a Sustainable Environment, Economy, and Society Carla P. Gomes Thomas Dietterich Jon Conrad John Hopcroft (Lead PI Cornell University) (PI Oregon State University) (Co-PI Cornell University) Bowdoin

Sustainability and Sustainable Development The 1987 UN report, “Our Common Future” (Brundtland Report): § Raised serious concerns about the State of the Planet. § Introduced the notion of sustainability and sustainable development: Sustainability: “development that meets the needs of the present without compromising the ability of future generations to meet their needs. ” § Stated the urgency of policies for sustainable development. UN World Commission on Environment and Development, 1987. Gro Brundtland Norwegian Prime Minister Chair of WCED 2

Follow-Up Report: Intergovernmental Panel on Climate Change (IPCC) (2007) ”There are no major issues raised in Our Common Future for which the foreseeable trends are favourable. ” Erosion of Biodiversity Examples: • The biomass of fish is estimated to be 1/10 of what it was 50 years ago and is declining. Global Warming +130 countries • At the current rates of human destruction of natural ecosystems, 50% of all species of life on earth will be extinct in 100 years. [Nobel Prize with Gore 2007] 3

Vision Computer scientists can — and should — play a key role in increasing the efficiency and effectiveness of the way we manage and allocate our natural resources, while enriching and transforming Computer Science. 4

Outline I Research Goals and Agenda II Our Team III Project Management, and Collaboration Plan IV Knowledge Transfer, Education, and Outreach V Value-Added as an Expedition 5

Outline I Research Goals and Agenda II Our Team III Project Management, and Collaboration Plan IV Knowledge Transfer, Education, and Outreach V Value-Added as an Expedition 6

Overall Goal of our Expedition To establish a new field, Computational Sustainability: Focused on computational methods for balancing environmental, economic, and societal needs for a sustainable future. To inject Computational Thinking into Sustainability that will provide: – new insights into sustainability questions; – new challenges and new methodologies in Computer Science (Analogous to Computational Biology) We will create the Institute for Computational Sustainability (ICS), – to perform and foster research in Computational Sustainability – to establish a vibrant research community, reaching far beyond the participating members of this Expedition. 7

Sustainability Themes 8

Sustainability Themes I Conservation and Biodiversity E. g. Wildlife Corridors II Balancing Socio-economic Demands and the Environment E. g. Policies for harvesting renewable resources III Renewable Energy Fuel distributors Farmers Energy crops Water quality E. g. Biofuels Soil quality Food supply Local air pollution Consumers Energy market Environmental impact Biodiversity Ethanol Refinery Gasoline producers Non-energy crops Social welfare Economic impact 9

Computational Challenges 10

Challenges in Constraint Reasoning and Optimization: Conservation and Biodiversity: Wildlife Corridors link core biological areas, allowing animal, seed, and pollen movement between areas; Typically: low budgets to implement corridors. Computational problem Connection Sub-graph Problem Given a graph G with a set of reserves: Find a sub-graph of G that: Connection Sub-Graph - NP-Hard Worst Case Result --- Real-world problems possess hidden structure that can be exploited allowing scaling up of solutions Science of Computation. contains the reserves; is connected; with cost below a given budget; and with maximum utility 11

Real world instance: Glacier Park Corridor for grizzly bears in the Northern Rockies, connecting: Yellowstone Salmon-Selway Ecosystem Glacier Park Salmon-Selway (12788 nodes) Yellowstone Scaling up Solutions by Exploiting Structure: Typical Case Analysis Identification of Tractable Sub-problems Streamlining for Optimization Static/Dynamic Pruning 5 km grid (12788 land parcels): minimum cost solution 5 km grid (12788 land parcels): +1% of min. cost Our approach reduced corridor cost from $1 Billion to $10 Million [Conrad et al. 2007] Our Expedition is partnering with the Conservation Fund which works with US Fish & Wildlife Service and the Nature Conservancy on preserving habitat in the Northern Rockies . Interdisciplinary Research Project (IRP): 12 Wildlife Corridors (Amundsen, Conrad, Gomes, Selman + postdoc + 2 Ph. D. students )

Additional Levels of Complexity: Stochasticity, Uncertainty, Large-Scale Data Modeling • Highly stochastic environments • Multiple species (hundreds or thousands), with interactions (e. g. predator/prey). IRP Nisource: Conservation Fund (14 states) • Spatially-explicit aspects within-species • Different models of land acquisition IRP Joint Public/Private Management for Biodiversity (e. g. , purchase, conservation easements, auctions) typically over different time periods • Dynamical models Dynamics of species; Movements and migrations; Themes - Transformative Synthesis: Optimization and Machine Learning Dynamics and Optimization IRP Native Plant Habitat Recovery Victoria (Australia) Combinatorial Auctions Outreach Cornell’s e. Bird Citizen Science 10 M+ bird observations since 2002, by general public IRP Collective Inference on Markov Models for Modeling Bird Migration Ruby-throated Hummingbird Sightings Negative observations Positive observations 13

Challenges in Dynamic Models and Optimization : Balancing Socio-Economic Demands and Environment: Economy No n-li n ear dy n am ics Example of a Biological Growth Function F(x): Logistics map: x t+1 = r xt (1 - xt), r is the growth rate Key Issues in Dynamical Models: multiple scales and multistability Increasing Complexity: multiple renewable resources interactions We are interested in problems that involve policy decisions (e. g. when to open/close a fishery ground over time). Uncharted territory: Combinatorial optimization problems with an underlying dynamical model. Transformative Synthesis: New Class of Hybrid Dynamic Optimization Models IRP: Rotational Management of Fishing Grounds IRP: Joint Public/Private Management for Biodiversity IRP: Fire Management in Forests 14

Challenges in Highly Interconnected Multi-Agent Systems: Renewable Energy Independence and Security Act (Signed into law in Dec. 2007) Ambitious mandatory goal of 36 billion gallons of renewable fuels by 2022 (five-fold increase from current level) IRP Large Scale Logistics Planning for Biofuels Fuel distributors Farmers Non-energy crops Energy crops Food supply Consumers Energy market Water quality Soil quality Biofuel Refineries Environmental impact Biodiversity Local air pollution Social welfare Economic impact 15

Realistic Computational Economic Models Current approaches limited in scope and complexity § E. g. based on general equilibrium models (e. g. , Nash style) § Strong convexity assumptions to keep the model simple enough for analytical, closed-form solutions (unrealistic scenarios) Limited computational thinking Transformative research directions § More realistic computational models in which meaningful solutions can be computed § Large-scale data, beyond state-of-the-art CS techniques § Study of dynamics of reaching equilibrium — key for adaptive policy making! Fuel distributors Farmers Gasoline producers Non-energy crops Food supply Energy crops Energy market Water quality Soil quality Biodiversity Consumers Environmental impact Local air pollution Social welfare Economic impact IRP Impact of Biofuels: Dynamic Equilibrium Models IRP Impact of Land-use on Climate 16

Science of Computation Our approach: The study computational problems as natural phenomena in which principled experimentation, to uncover hidden structure, is as important as formal analysis Science of Computation, Our team has a track record of making compelling scientific discoveries using such an approach. Small world phenomenon Pioneered science of networks [Watts & Strogatz] Phase Transitions in Computation Heavy tails in Computation Led to interactions between CS, statistical physics, and math Led to randomization and portfolios in combinatorial search [Selman & Kirkpatrick] [Gomes et al. ] 17

Overall Research Vision: Transformative Computer Science Research: Driven by Deep Research Challenges posed by Sustainability Design of policies to manage natural resources translating into large-scale optimization and learning problems, combining a mixture of discrete and continuous effects, in a highly dynamic and uncertain environment increasing levels of complexity Complex decision models Constraint Reasoning and Optimization Co mp uta t ion Large-scale data and uncertainty Machine Learning, Statistical Modeling ien ce of Multiple time scales From Statics to Dynamics: Dynamic Models Sc Many highly interconnected components; From Centralized to Distributed: Computational Resource Economics Distributed Highly Interconnected Components / Agents ing y as re exit Inc mpl Co Study computational problems as natural phenomena Science of Computation Dynamics Data and Uncertainty Constraint Satisfaction and Optimization Complexity levels in Computational Sustainability Problems 18

Transformative Computer Science Research: Driven by Deep Research Challenges posed by Sustainability an S sf Dy yn orm a n th Op am esi tive tim ics s: iza an tio d n Tr Guckenheimer Strogatz Zeeman. PI, W Yakubu. AA Complex decision models Constraint Reasoning, Optimization, Stochasticity Resource Economics, Environmental Sciences & Engineering ive at m : or is sf es on an th ti Tr Syn miza ing rn ti Op Lea & Dynamical Models Large-scale data and uncertainty Machine Learning, Statistical Modeling ty Multiple scales (e. g. , temporal, spatial, geographic) From Statics to Dynamics: Dynamic Models rsi Many highly interconnected components; From Centralized to Distributed: Computational Resource Economics Gomes PI, W, Hopcroft Co-PI Selman. Co-PI, Shmoys. Co-PI e div Bio Study computational problems as natural phenomena Science of Computation Constraint Reasoning & Optimization Re n En ewab erg le y Design of policies to effectively manage Earth’s natural resources translate into large-scale decision/optimization and learning problems, combining a mixture of discrete and continuous effects, in a highly dynamic and uncertain environment increasing levels of complexity Albers. Co-PI, W, Amundsen, Barrett, Bento, Conrad. Co-PI, Di. Salvo, Mahowald. W, Montgomery. Co-PI, W Rosenberg, Sofia. W, Walker. AA Environmental & Socioeconomic Needs Transformative Synthesis: Dynamics & Learning Science of Computation Data & Machine Learning Dietterich. PI Wong. Co-PI Chavarria. H 19

Outline I Research Goals and Agenda II Our Team III Project Management, and Collaboration Plan IV Knowledge Transfer, Education, and Outreach V Value-Added as an Expedition 20

Multi-institutional, Multidisciplinary Research Team 6 Institutions, 7 colleges, 12 departments Bento Res. & Env. Economics Cornell Shmoys CS & OR Cornell Strogatz Appl. Math Cornell Mahowald Earth & Atmos. Sci. Cornell Dietterich CS OSU Gomes CS Cornell Zeeman Appl. Math Bowdoin Hopcroft CS Cornell Yakubu Appl. Math Montgomery Howard Res. & Env. Economics OSU Albers Res. & Env. Econo. OSU Institute for Computational Sustainability Walker Biology & Eng. Cornell Chavarria HPC. PNNL Rosenberg Conservation Biology Cornell Guckenheimer Appl. Math Cornell Sofia Biology Cornell Di. Salvo Chemistry Cornell Barrett Res. & Env. Economics Cornell Amundsen Conrad Conservation Res. and Env. Planning Economics Cons. Fund Cornell Selman CS Cornell Wong CS OSU 21

Cornell University, Ithaca, NY Oregon State University Corvallis, OR Tradition of public service as part of its land-grant mission (private & state university) Several centers focusing on sustainability research and dissemination results to community and policy makers Strong programs in computer science, agriculture, natural resources, biology, renewable energy, and environment Provost’s campus wide initiative in Sustainability Bowdoin College Brunswick, ME Liberal arts undergraduate college Culture of undergraduate research Howard University Washington, DC Historically Black University Leading producer of African-American Ph. D. s Focus on Science, Technology, Engineering, and Mathematics Tradition in agricultural extension service & research stations #1 ranked Forestry School Strong Ecosystem Informatics (IGERT, NSF Summer Inst. ) Strong programs in computer science, fisheries & wildlife, . oceanography, and environmental atmospheric Sciences, sciences Provost’s initiative in Ecosystem Informatics Conservation Fund Nationwide Since 1985, the Fund and partners have safeguarded over 6 million acres of wildlife habitat, forests, and community greenspace. Non-profit org. focused on conserving natural resources. Pacific Northwest National Laboratory Department of Energy's (DOE's) laboratory: focus on environmental science, climate change, remediation, and security. 22 Access to HPC Systems

Institutional Support Strong institutional support: • From top level management (president, provost) • Researchers eager to share data, problems, and anxious for new computational models to solve their sustainability problems • Cornell Center for a Sustainable Future (CCSF) with organizational support and matching funds (5% non-federal money). Institute for Computational Sustainability And this space! 23

Outline I Research Goals and Agenda II Our Team III Project Management, and Collaboration Plan IV Knowledge Transfer, Education, and Outreach V Value-Added as an Expedition 24

Integrated Management Institute for Computational Sustainability: Director Carla P. Gomes (PI Cornell University) Associate Director David Shmoys (Co-PI Cornell Univeristy) Deputy Director Thomas Dietterich (PI Oregon State University) Deputy Director Mary Lou Zeeman (PI Bowdoin University) Plan, coordinate, and evaluate exciting and aggressive research, education, and outreach program. Disseminate computational sustainability. Executive Committee: PI’s + Co-PI’s Meets once a month. Advisory Board: Prominent researchers in fields related to computational sustainability, both from the application side (2 people) and the computer science side (2 people). Will also invite an NSF representative. Meets once a year. Collaboration Plan: Through Interdisciplinary Research Projects (IRPs, shared personnel), executive committee meetings, exchange students and faculty, three-day annual meetings, meetings at regular CS conferences. 25

Interdisciplinary Research Projects (IRPs): The Building Blocks of our Expedition Seedling IRPs IRP Name Faculty Team 1. Wildlife Corridors for Grizzly Bears Amundsen, Conrad, Gomes, Selman, Shmoys 2. Biofuels Bento, Gomes, Mahowald, Shmoys, Strogatz, Walker, Wong 3. Bird Conservation Rosenberg, Conrad, Dietterich, Gomes, Hopcroft, Strogatz, Zeeman 4. Native Plant Habitat Recovery in Victoria, Australia Dietterich, Gomes, Selman 5. Joint Public/Private Management for Biodiversity Amundsen, Montgomery, Dietterich, Gomes, Hopcroft 6. Fire Management in Forests Albers, Conrad, Guckenheimer, Selman 7. Rotational Management of Fishing Grounds Conrad, Guckenheimer, Yakubu, Zeeman 8. Pastoral Systems in East Africa Barrett, Conrad, Wong 26

Nurturing New IRPs Application/Synthesis Facilitators 27

Outline I Research Goals and Agenda II Our Team III Project Management, and Collaboration Plan IV Knowledge Transfer, Education, and Outreach V Value-Added as an Expedition 28

Institute Activities Building Research Community Research Coordinating transformative synthesis collaborations Web Portal Interdisciplinary Research Projects (IRPs) ICS Education Postdocs Doctoral students Honors projects Research seminar series Summer REU program targeting minority students Computational Sustainability follow-up to State of the Planet course Panels & tutorials at major conferences Annual symposium Catalog Of Problems Host visiting Scientists Outreach Citizen Science Project Cornell Center for a Sustainable Future Conservation Fund Cornell Cooperative Extension OSU Alliance for Computational Sustainability 29

Education and Outreach Research on Computational Sustainability will significantly broaden the field of computer science and attract a new generation of students who traditionally may not have considered studying computer science --- thus contributing to a revitalization of CS education. Undergraduate and graduate students will be drawn into hands-on computational sustainability research by joining IRP-teams. The ICS will organize regular seminars and an annual symposium and summer school. 30

State of the Planet Course “Students unite to create State of the Planet course” - K. L. Rypien et al. , Nature, Vol 447, July 2007, p 775 - mentors Tom Eisner and Mary Lou Zeeman (co-PI) 250 students, 45 different majors. 95% recommend it to their peers: • “… the class … has started me thinking about career paths I hadn’t considered before. I find most of the lectures inspiring and hopeful; they leave me feeling that I can actually make a difference in the world. ” • “Amazing course! Very thought-provoking, interesting, varied, and exciting. Best course I've taken at Cornell so far. I have been recommending it to everyone. ” We will create a companion course on Computational Sustainability Renewable Energy Food & water Biodiversity Carbon and climate 31

Integration of Research and Outreach Example: Citizen Science at the Cornell Laboratory Of Ornithology § Increase scientific knowledge Gather meaningful data to answer large-scale research questions § Increase scientific literacy Enable participants to experience the process of scientific investigation and develop problem-solving skills § Increase conservation action Apply results to science-based conservation efforts Citizen Science empowers everyone interested in birds to contribute to research. 32

Additional Channels for Impact and Outreach Lab. of Ornithology K. Rosenberg Community, policy makers, local and state leaders Institute for Environment Energy & Policy Director: A. Bento Conservation Fund Ole Amundsen Cornell Biofuels Lab L. Walker Community and Rural Development Institute Cornell Cooperative Extension A. Bento Northeast Sun Grant Institute at Cornell Director: L. Walker New York Sea-Grant Cornell Extension Jon Conrad OSU IGERT Eco-System Informatics Institute for Computational Sustainability USDA Forestry Sciences Lab EPA Western Ecology Division Howard Un A. Yakubu Research community & students Cornell Center for Sustainable Future Director: F. Di. Salvo African Food and Security Natural Resources Mgmt. Prgm Director: C. Barrett Earth & Atmospheric Sciences N. Mahowald We will leverage a host of sustainability centers and programs at our 6 institutions. 34

Outline I Research Goals and Agenda II Our Team III Project Management, and Collaboration Plan IV Knowledge Transfer, Education, and Outreach V Value-Added as an Expedition 36

Value-Added as an Expedition § Fundamentally new intellectual territory for computer science § Unique societal benefits § Tremendous potential for recruiting new groups (including under-represented minorities) to computer science § Establishing the field of Computational Sustainability requires critical mass and visibility that cannot be achieved with piece-meal efforts § Our research proposal is fundamentally cross-disciplinary: – IRPs require large teams involving both domain scientists and computer scientists – Transformative syntheses are also across disciplines in CS § NSF is the agency best positioned to lead this initiative 37

Summary Expedition will lead to: 1) Foundational contributions to computer science driven by Sustainability questions. 2) Societal and environmental impact --- development of computational methods to alleviate sustainability problems (e. g. significant opportunity for more efficient use of natural resources). 3) Establishment of the field of Computational Sustainability. 4) Integration of research, education, and outreach. New courses and seminars integrated with Interdisciplinary Research Projects. 5) Increasing diversity in computer science. Bringing in a new generation of students, traditionally not drawn to CS; also, broadening the public image of CS. Thank You! 38