FIELDS INSTITUTE FOR RESEARCH IN MATHEMATICAL SCIENCES
Industrial Optimization Seminar
at 5:00 p.m.
the Fields Institute, 222 College St., Toronto
The inaugural meeting of the Fields Industrial Optimization Seminar
took place on November 2, 2004. The seminar meets in the early evening
of the first Tuesday of each month. Each meeting is comprised of
two related lectures on a topic in optimization; typically, one
speaker is a university-based researcher and the other is from the
private or government sector. The series welcomes the participation
of everyone in the academic or industrial community with an interest
in optimization theory or practice, expert or student . Please
subscribe to the Fields mail list to be
informed of upcoming seminars.
The Fields Institute makes a video record of this seminar through
FieldsLive. If you make a presentation
to the Seminar, the Institute will be video-recording the presentation
and will make the video record
available to the public.
Talks streamed live at FieldsLive
Adrian Nachman, University of Toronto
A Variational Problem Arising in Conductivity Imaging from Interior
Imaging electric conductivity of tissue is both desirable and challenging.
The classical Electric Impedance Tomography Problem seeks to determine
the conductivity from measurements of voltages and currents at the
boundary; it has spurred deep and far-reaching mathematical developments.
The ill-posedness of the problem is now well understood, and places
severe limitations on the resolution that can be achieved. We will
discuss one approach to overcome these limitations: using interior
current density data obtainable by a method pioneered by Joy, Scott
and Henkelmann at the University of Toronto which makes use of Magnetic
Resonance Imagers in a novel way.
We only require knowledge of the magnitude |J| of one current for
a given voltage f on the boundary. We show that the corresponding
electric potential is the unique solution of a constrained minimization
problem with respect to a weighted total variation functional defined
in terms of the physical data. Working with the dual variational problem
leads naturally to an alternating split Bregman algorithm, for which
we prove convergence. The dual problem also turns out to yield a novel
method to recover the full vector field J from knowledge of its magnitude,
and of the voltage f on the boundary. Time permitting, we will discuss
the corresponding problem for anisotropic conductivities.
The results presented are from joint work with Nicholas Hoell, Robert
Jerrard, Amir Moradifam, Alexandru Tamasan and Alexander Timonov.
Experimental results are joint work with Nahla Elsaid, Michael Joy,
Weijing Ma, and Tim DeMonte.
Douglas C. White, Emerson Process Management
Process Plant Optimization in Real Time: Energy and Environmental
Many industrial plants produce products worth millions of dollars
per day and continuous financial optimization of their operations
is obviously attractive. Applications of real time optimization technology
in the process industries have been attempted for at least the last
fifty years; sometimes successfully, sometimes not. In this presentation
there will be a short review of the history of these attempts and
some of the lessons learned. With the global increase in energy costs
and environmental regulations, a current focus is real time optimization
of the very complex energy systems in these large industrial sites
and the associated environmental impact. Optimization has to fit within
the overall control hierarchy and structure at the site which creates
special system requirements. The problem structure, issues and current
status of these applications is presented as well as open questions
that are topics for future developments.
Laura Sanita, University of Waterloo
Finding small stabilizers for unstable graphs
A vertex v of a graph G is called inessential if there exists a maximum
matching in G that exposes v. G is said to be stable if the set of
its inessential vertices forms a stable set. Stable graphs play a
key role in network bargaining games where we are given a set of players
represented as vertices of a graph G, and a set of possible deals
between players represented by the edges of G. Kleinberg and Tardos
[STOC 08] defined the notion of a balanced outcome for a network
bargaining game, and proved that a balanced outcome exists if and
only if the correspondent graph G is stable. This connection motivates
the optimization problem of finding a minimum cardinality stabilizer
of a given unstable graph G, that is a subset of edges F such that
G \ F is stable. In this talk we prove some structural results about
this problem and develop efficient approximation algorithms for sparse
graphs. Joint work with A. Bock, K. Chandrasekaran, J. Koenemann,
and B. Peis.
Ritchie (Yeqi) He, Royal Bank of Canada
An Improved Model for Calculation of Debt Specific Risk VaR with
Initially introduced in the 1996 Amendment of Basel Accord, the specific
(or spread) risk of a debt portfolio (DSR) is the risk due to changes
of idiosyncratic credit spreads (bond spreads or CDS spreads) related
to individual entities. Financial institutions are allowed to use
internal models to calculate the Value-at-Risk (VaR) of DSR. Internal
models usually calculate portfolio DSR PnL based on an assumption
that idiosyncratic credit spreads follow a tractable closed-form joint
distribution such as multi-variate normal or student's t-distribution.
This assumption may not give a satisfactory approximation to the joint
distribution because the marginal distribution of idiosyncratic credit
spreads usually has fat tails. To better model fat tails, we propose
an improved Monte Carlo-based model to calculate the DSR VaR. In the
proposed model, the marginal distribution is modeled by a normal kernel
distribution with Pareto tails, and the dependence structure of idiosyncratic
credit spreads are captured by a student's t copula. For Pareto tails,
the calibration of shape parameters and scale parameters involves
a series of density fitting problems, which are solved by the maximum
likelihood estimation. Numerical results show that, the proposed model
is capable to generate more accurate distributions, and consequently
the quality of estimation of the DSR VaR is improved.
Joint work with Meng Han, Royal Bank of Canada.
Please click here for Dr. He's presentation
September 24, 2013
|3:15 - 4:05
Jonathan Briggs (Canada Pension Plan Investment Board) Video
A Portfolio Construction Toolkit
In the practitioner world, the value of portfolio construction is
often viewed with skepticism - a skepticism born of an honest assessment
of the dubious forecasting power of the inputs and the opacity of
the process. Despite all our wonderful mathematical gymnastics, if
we dont know the nature what is consumed, how can we possibly
truthfully convey the nature of what we create? Now suppose we knew
the distributions, the dynamics, the Information Ratio (IR) and the
interrelationships between a multifactor model and its returns, and
further we could disentangle each twist and turn of the raw factors
as they are transformed into trades and holdings? Well, then maybe
we really could demonstrate the value of portfolio construction.
4:05 - 4:55 p.m.
Bogie Ozdemir (Sun Life Financial Group) Video
Capital and Business Mix Optimization
Basel III amounts to a climate change in the banking industry. It
increased the capital requirements significantly - especially for
certain businesses (most notably Capital Markets) and decreased the
acceptable forms of capital. Capital has become a scarce resource
under Basel III, putting significant downward pressure on ROE. In
this new environment, banks will need to change their business mixes,
exit or shrink capital heavy businesses and adjust their operating
models, while meeting income targets. During this course correction
their ROE and Income Targets will be challenged further as some re-balancing
of operating models may compromise short term income to improve ROE
in future years. Subject to more onerous capital requirements under
Basel III banks will need to increase the efficiency of capital utilization
and place greater emphasis on optimizing capital allocation and business
mix across their operations. In this presentation, we will discuss
how to establish an optimization framework incorporating both economic
and regulatory capital.
Back to top