Computational Sustainability: Computing for a Better World

Computational Sustainability: Computing for a Better World

[MUSIC] So yeah so this is sort of
a large scale effort that Carla Gomes actually
the director for it’s called computational
sustainability. And it fits nicely
into workshop, and it thinks actually is also
involved in this effort. So it’s an expedition in
computing by the NSF that started this effort, actually it
was already about five years or seven years ago that they
gave us a large grant, a $10 million grant. To start to fuel the
computational sustainability, and I’ll say more
about that soon. But then recently about two
years ago, they gave us a second expedition grant, another ten
million to build a network. And we call that compsys net, computational sustainability
network. And it involves about
125 researchers, there’s some pictures
of the people. Dozens of institutions,
academic institutions, but also NGOs and
government agencies. So this relays to the question, how do you actually change
things in the real world? Well you half to get
other agency involved to implement the ideas and police used and the things you
do in as a research effort. So, let me just. So at the higher level,
what conditions they build is proposing is to use
computer science methods, artificial intelligence methods,
to help sustainability efforts. And we think of
sustainability as a broad effort balancing environmental,
economic, and societal needs. So, we work in
various countries, like actually Carla
is right now, she’s actually just coming back
from the Amazon rain forest. She’s working with local
government there to plant hydro dams. I’ll say a little bit
about that, but there’s a tension between economic
development and sustainability goals of preserving the rain
forest, the Amazon rain forest. And you have to balance
these interests, so that’s a balancing act. And the idea is to use computer
design optimization techniques for trying to get the right
balance, or at least to advice policymakers on what to
write the best balance is. So we actually have three
sort of general trusts, balancing socioeconomic needs. We have for example a project in
Africa on poverty mitigations, helping herders people that herd various animals and
often water is the big issue. They find watering points so we have worked on the placement,
of the optimal placement, of watering points in Africa,
in some very dry areas. Conservation and by diversity,
is the second point of emphasis. Bird conservation,
wildlife conservation, we’ll get some more examples
of that in a moment. And a third thrust is
Scientific discovery, and we’re focusing
there on accelerating a materials discovery for
renewable energy. And that’s for example,
for solar cells. You want to find better
solar cell materials. And we found material scientists are working
very hard on that but we’re complementing that with
new computer science methods. So all of these areas we
work with domain experts and we complement it with
computer scientists. So in computer science we
look at a sort of a three. Computational thrusts
which are big data, machine learning, constraint
reasoning, optimization, [INAUDIBLE] actually
our own background and then third, multi-agent systems
where we use game theory, citizen science and
crowd sourcing. And we are integrating those
three different themes in a computation
sustainability network. So to actually one
thing that’s sort of nice to do when we have to
set up the expeditions, you could write ambitious proposals,
there’s a competition in the US. A very ambitious proposal from
all over the US came in, but you could really think about
how would I set something as interdisciplinary as really how would I create
that field and so we came up with the notion of introducing
our research projects or IRP’s. That had to consist of teams,
normally about two or three computer scientists,
and two or three and sometimes five or six can be
quite large of domain exports, ecologists, applied economists,
by diversity experts. Different areas,
other disciplines. Put them together in one team,
have several students involved. Often the students are the
driving force in collaborations because they have the most time
to work on the problem and they sort interfacing
faculty members. And this core teams then take
on specific challenges and is always looking for
what is interesting, both from the computer
science perspective and from the application perspective,
and that’s working very well. So some projects,
not all projects can flourish, but a good set of projects
have really flourished. And those, we keep supporting,
and ones that don’t work well, we replaced by new efforts, so we have literally dozens
of efforts, we’ll use that. So here’s a map view
because we have to think about how do we get
computer scientists involved. And especially how do you
get PhD students involved. And there’s actually
sort of a challenge, but I think we did find
a way to do it. Ultimately your students have to
go in the academic job market. They’re trained as
computer scientists, they have the computer
science PhD. If they say, well,
I did buy a diversity In the job interview for
a computer science department. [LAUGH] They won’t get the job. [LAUGH] So we thought carefully
about how can the student present himself and what we
developed is this notion that, well computer science is
really about abstractions. And it cuts across domain. So what we actually have,
is we have things like, we call them subway lines, because taken
from the subway line picture. But what we have is the students
work on things like pattern decomposition in big data. They work on agents
mechanism design, citizen science/crowdsourcing,
stochastic, probabilistic inference,
and optimization. These are different colors. Subway lines here. So the student’s specialty may be contribution
to Probabilistic Inference, Optimization, and
Stochastic Inference. Now, how does this come
together with sustainability. Well, you look at
the subway line. That basically is
a general computational theme that cuts across a number
of sustainability projects. So if I look at
the large scale temporal, reasoning and prediction,
it’s the black line here. It brings together you know
issues in power grid management and wind and
solar forecasting even in poverty management because of
the temporal special aspect. So we connect these
various projects through these computational themes. It’s not an artificial
connection, it’s really how
computer science work. Computer science abstracts away. So it’s almost a bit like
mathematics in a sense, it provides a general
computational model. So the students comes out as an
expert in general optimization, stochastic reasoning,
game theory. That really relates to
computer science departments. And then illustrates their
work by multiple projects and we found that works really
well because then you see impact in a number of different
sustainability projects. So it’s interesting because for
the domain experts, the projects themselves
are the key thing. So they work, the domain
actuary work on the single I do by diversity or something. And or I do property
management in Africa. So for the domain actuary
that’s enough, for the computer scientist it’s much
better to have several projects and contribute to
a range of them and have these computational themes. So that’s the separate
picture line. So here is a little example,
well actually let me go, so we’re connecting
wildlife reserves and we work with the wildlife
organization and the National Conservation
Society in the US. They actually buy
these parts of lands to connect wildlife
reserves in the US. And we did optimization work to
find what is the cheapest route, what is the cheapest set of. Lots of traders we
considered the cost and also efficiency of
the wildlife corridor. So what’s the best way
to design a corridor. And our computational
methods which were you know, standard work. Reasonably standard optimization
by the first VS made a huge difference for
the conservation people, because they had
never used this tool. So that’s one of the rewarding
aspects of all this work. You can actually do quite
a bit very quickly that the people in these separate
field have no hope of doing. One minute left. Let me just,
I guess, let me just. Yeah, so beautiful, let me just.>>[LAUGH]
>>So Carla wasn’t just in the Amazon,
so [INAUDIBLE] a project. Let me just actually [INAUDIBLE]
I wanna get the right pictures here. So one other project, we’re
analyzing hydro dam, China is pushing very hard to build
a large number of hydro dams, hydro power in
the Amazon rain forest. And so what we are looking
at is trade offs, 500 potential locations for
the hydro dams. And we’re looking at the trades,
so the water flows that way. [LAUGH] And the water comes
out of the mountains, and the dots here are various
hydrodams that are proposed. So we just finished, or
we’re in the middle actually of the project,
of finding the tradeoffs. The risk to biodiversity,
the seismic risks, and the cost, and the amount of electricity
the dam produces. Find an optimal set of dams to
design so we’re working with a colleague who just went there
talking to the local government to advise them as to what
is the best dam placement. But computing to create an
optimum we’re good at that, so we actually did that for the whole Amazon rainforests and
thousands of tradeoffs. Okay so time is over. Boy, let me just switch
to the last slide. We do bird conservation, we do
accelerating material discovery. [LAUGH]
>>[LAUGH]>>There’s almost nothing we don’t do.>>[LAUGH]
>>But let me go to the last picture. So we wanna flow both ways. Sustainability science, computation science, we have
a pathway in both directions, that’s very important to
have real impact, and we have lots of people
we work with, thank you.>>[APPLAUSE]

Daniel Ostrander

Related Posts

Leave a Reply

Your email address will not be published. Required fields are marked *