Computational Sustainability: Computational Methods for a Sustainable Environment, Economy, and Society

1
Computational SustainabilityComputational
Methods for a SustainableEnvironment, Economy,
and Society
Bowdoin
Carla P. Gomes (Lead PI Cornell University)
Thomas Dietterich (PI Oregon State
University) Jon Conrad (Co-PI Cornell
University) John Hopcroft (Co-PI Cornell
University)
2
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.

Gro Brundtland Norwegian Prime Minister Chair of
WCED
UN World Commission on Environment and
Development,1987.
3
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.
• At the current rates of human destruction of
  natural ecosystems, 50 of all species of life on
  earth will be extinct in 100 years.

Global Warming
130 countries
Nobel Prize with Gore 2007
4
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.
5
I Research Goals and AgendaII Our TeamIII
Project Management, and Collaboration PlanIV
Knowledge Transfer, Education, and OutreachV
Value-Added as an Expedition
Outline
6
I Research Goals and AgendaII Our TeamIII
Project Management, and Collaboration PlanIV
Knowledge Transfer, Education, and OutreachV
Value-Added as an Expedition
Outline
7
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.

8
Sustainability Themes

9
Sustainability Themes
I Conservation and Biodiversity
E.g. Wildlife Corridors
II Balancing Socio-economic Demands and the
Environment
E.g. Policies for harvestingrenewable resources
III Renewable Energy
E.g. Biofuels
Ethanol Refinery
10
Computational Challenges

11
Challenges in Constraint Reasoning and
Optimization Conservation and Biodiversity
Wildlife Corridors
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
Connection Sub-graph Problem
Given a graph G with a set of reserves
Find a sub-graph of G that contains the
reserves is connected with cost
below a given budget and with
maximum utility
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.
12
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 .
Glacier Park
Real world instance Corridor for grizzly bears
in the Northern Rockies, connecting Yellowston
e Salmon-Selway Ecosystem Glacier Park
Salmon-Selway
Yellowstone
(12788 nodes)
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
Interdisciplinary Research Project
(IRP)Wildlife Corridors (Amundsen, Conrad,
Gomes, Selman postdoc 2 Ph.D. students )
13
Additional Levels of Complexity Stochasticity,
Uncertainty, Large-Scale Data Modeling

• Highly stochastic environments

• Multiple species (hundreds or thousands),
• with interactions (e.g. predator/prey).

• Spatially-explicit aspects within-species

• Different models of land acquisition
• (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 Machine Learning Dynamics and Optimization
14
Challenges in Dynamic Models and Optimization
Balancing Socio-Economic Demands and Environment
Economy
Key Issues in Dynamical Models multiple scales
and multistability
Increasing Complexity multiple renewable
resources interactions
IRP Rotational Management of Fishing
Grounds IRP Joint Public/Private Management for
Biodiversity
IRP Fire Management in Forests
15
Challenges in Highly Interconnected Multi-Agent
SystemsRenewable Energy
Energy Independence and Security Act(Signed into
law in Dec. 2007)
Ambitious mandatory goal of36 billion gallons of
renewable fuels by 2022 (five-fold increase from
current level)
IRP Large Scale Logistics Planning for
Biofuels
16
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 enoughfor 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!

IRP Impact of Biofuels Dynamic
Equilibrium Models
IRP Impact of Land-use on Climate
17
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,

18
Overall Research VisionTransformative Computer
Science ResearchDriven 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
Study computational problems as natural phenomena
? Science of Computation
Many highly interconnected components ? From
Centralized to Distributed Computational
Resource Economics
Multiple time scales ? From Statics to
Dynamics Dynamic Models
Large-scale data and uncertainty ? Machine
Learning, Statistical Modeling
Complex decision models ? Constraint Reasoning
and Optimization

19
Transformative Computer Science ResearchDriven
by Deep Research Challenges posed by
Sustainability
Design of policies to effectively manage Earths
naturalresources 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
Study computational problems as natural phenomena
? Science of Computation
Many highly interconnected components ? From
Centralized to Distributed Computational
Resource Economics
Multiple scales (e.g., temporal, spatial,
geographic) ? From Statics to Dynamics Dynamic
Models
Large-scale data and uncertainty ? Machine
Learning, Statistical Modeling
Science of Computation
Complex decision models ? Constraint Reasoning,
Optimization, Stochasticity
Science of Computation

20
I Research Goals and Agenda II Our TeamIII
Project Management, and Collaboration PlanIV
Knowledge Transfer, Education, and OutreachV
Value-Added as an Expedition
Outline
21
Multi-institutional, Multidisciplinary Research
Team 6 Institutions, 7 colleges, 12 departments
Bento Res. Env. Economics Cornell
Hopcroft CS Cornell
Zeeman Appl. Math Bowdoin
Gomes CS Cornell
Dietterich CS OSU
Mahowald Earth Atmos. Sci. Cornell
Yakubu Appl. Math Howard
Strogatz Appl. Math Cornell
Montgomery Res. Env.Economics OSU
Shmoys CS OR Cornell
Albers Res. Env. Econo. OSU
Institute for Computational Sustainability
Walker Biology Eng. Cornell
Chavarria HPC. PNNL
Rosenberg Conservation Biology Cornell
Wong CS OSU
Selman CS Cornell
Guckenheimer Appl. Math Cornell
DiSalvo Chemistry Cornell
Amundsen Conservation Planning Cons. Fund
Conrad Res. and Env. Economics Cornell
Barrett Res. Env. Economics Cornell
Sofia Biology Cornell
22
Cornell University, Ithaca, NY
Oregon State University Corvallis, OR
Tradition of public service as part of its
land-grant mission (private state university)
Tradition in agricultural extension service
research stations
1 ranked Forestry School
Several centers focusing on sustainability
research and dissemination results to community
and policy makers
.
Strong Ecosystem Informatics (IGERT, NSF Summer
Inst.)
Strong programs in computer science, fisheries
wildlife, atmospheric Sciences, oceanography, and
environmental sciences
Strong programs in computer science, agriculture,
natural resources, biology, renewable energy,
and environment
Provosts campus wide initiative in
Sustainability
Provosts initiative in Ecosystem Informatics
Bowdoin College Brunswick, ME
Conservation Fund Nationwide
Since 1985, the Fund and partners have
safeguarded over 6 million acres of wildlife
habitat, forests, and community greenspace.
Liberal arts undergraduate college
Culture of undergraduate research
Non-profit org. focused on conserving natural
resources.
Pacific Northwest National Laboratory
Howard University Washington, DC
Historically Black University
Department of Energy's (DOE's) laboratory focus
on environmental science, climate change,
remediation, and security.
Leading producer of African-American Ph.D.s
Focus on Science, Technology, Engineering, and
Mathematics
Access to HPC Systems
23
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!
24
I Research Goals and Agenda II Our TeamIII
Project Management, and Collaboration PlanIV
Knowledge Transfer, Education, and OutreachV
Value-Added as an Expedition
Outline
25
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 PIs Co-PIs
  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.

26
Interdisciplinary Research Projects (IRPs)The
Building Blocks of our Expedition
Seedling IRPs
 IRP Name  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
27
Nurturing New IRPs

Application/Synthesis Facilitators
28
I Research Goals and AgendaII Our TeamIII
Project Management, and Collaboration PlanIV
Knowledge Transfer, Education, and OutreachV
Value-Added as an Expedition
Outline
29
Institute Activities
Building Research Community
Research
Coordinating transformative synthesis
collaborations
Web Portal
Catalog Of Problems
Panels tutorials at major conferences
Interdisciplinary Research Projects (IRPs)
Host visiting Scientists
Annual symposium
ICS
Education
Outreach
Conservation Fund
Postdocs
Citizen Science Project
Research seminar series
Doctoral students
Honors projects
Cornell Cooperative Extension
Cornell Center for a Sustainable Future
Summer REU program targeting minority students
Computational Sustainability follow-up to State
of the Planet course
OSU Alliance for Computational Sustainability
30
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.

31
State of the Planet Course
Students unite to create State of the Planet
course - K.L. Rypien et al.,
Nature, Vol 447, July 2007, p775 -
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 hadnt 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
Biodiversity
Carbon and climate
Food water
32
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.
33
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

34
Additional Channels for Impact and Outreach
Community, policy makers, local and state
leaders
Community and Rural Development Institute
Lab. of Ornithology K. Rosenberg
Conservation Fund Ole Amundsen
Institute for Environment Energy
Policy Director A. Bento
CornellCooperative Extension A. Bento
Cornell Biofuels Lab L.Walker
Northeast Sun Grant Institute at
Cornell Director L. Walker
New York Sea-Grant Cornell Extension Jon Conrad
OSU IGERT Eco-System Informatics
Institutefor ComputationalSustainability
Cornell Center forSustainable Future Director
F. DiSalvo
USDA ForestrySciences Lab
African Food and SecurityNatural Resources
Mgmt.Prgm Director C. Barrett
EPA Western Ecology Division
Earth Atmospheric Sciences N. Mahowald
Howard UnA.Yakubu
Research community students
We will leverage a host of sustainability centers
and programs at our 6 institutions.
35
Computational Thinking into Sustainability
Our mission is to also educate other
sustainability researchers (e.g. resource
economists, environmental scientists, and
biologists) about the new methods and
opportunities computing and information science
brings to the field, in effect, introducing them
to computational thinking and modeling.
36
I Research Goals and AgendaII Our TeamIII
Project Management, and Collaboration PlanIV
Knowledge Transfer, Education, and OutreachV
Value-Added as an Expedition
Outline
37
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

38
Summary

• Expedition will lead to
• Foundational contributions to computer science
  driven by Sustainability questions.
• Societal and environmental impact --- development
  of computational methods to alleviate
  sustainability problems (e.g. significant
  opportunity for more efficient use of natural
  resources).
• Establishment of the field of Computational
  Sustainability.
• Integration of research, education, and outreach.
  New courses and seminars integrated with
  Interdisciplinary Research Projects.
• 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!