2016-2018 Operations Research Seminar

Welcome to the Web page of the SFU Operations Research Seminar Series. We are associated with:
CORDS (Centre for Operations Research and Decision Sciences), and
The Department of Mathematics, Simon Fraser University.
Our aim is to meet and discuss Operations Research topics.

Unless noted the talks will be at 3:30 on Thursday in Room 3040, SFU Surrey.
Please contact Tamon Stephen if you would like to speak.

Date Speaker Title and Abstract
Mar. 15th

SUR 3040
Sandy Rutherford

Complex Systems Modelling Group

Simon Fraser University
Control of an HIV Epidemic among Injection Drug Users: Simulation Modeling on Complex Networks

HIV remains a serious public health problem in many marginalized communities. We develop a network model of the HIV epidemic affecting injection drug users and female sex workers in the Downtown Eastside neighborhood of Vancouver, Canada, calibrated using data from public health surveillance and cohort studies. Many HIV positive individuals are unaware of their status and strategies for testing are an important part of HIV response programs. Upon diagnosis, HIV patients enter a continuum of care, involving both engagement and retention in treatment. We explored potential epidemic control strategies through simulation: reduced syringe sharing during injection drug use, reduced time to diagnosis, reduced time to initiation of treatment following diagnosis, and improved retention in treatment. We find that syringe sharing, HIV testing, and retention in treatment significantly impact HIV prevalence. Close connections between syringe sharing and sexual networks deserve attention as important avenues for rapid HIV transmission.
Mar. 8th

via COCANA (Kelowna)

Tim Hoheisel

McGill University
Epi-convergent Smoothing with Applications to Convex-Composite Functions

Further details available from the COCANA Website.
Mar. 1st

SUR 3040
Winfried Grassmann

Computer Science

University of Saskatchewan
Discrete Event Systems as Basis of Discrete-time Queues, Including Multiserver Queues and Tandem Queues

A number of papers have appeared describing how to calculate the distribution of the number of elements in a discrete-time GI/G/1 queue in a statistical equilibrium. Essentially, the process is converted to a discrete-time Markov chain and solved as such. It turns out that this method also works for discrete-time discrete event systems in a statistical equilibrium. They, too, can be converted into discrete time Markov chains, and solved. The problem is that due to the curse of dimensionality, the number of states is huge, and to solve the equilibrium equations, iterative methods should be used, such as Gauss-Seidel. Still, every effort must be made to reduce the dimensionality of the problem as much as possible. One way to do this is to embed the process at the points in time where events occur. Using this method, we could solve systems of reasonable size on our laptop, including GI/G/c queues and tandem queues. For some models, further reductions in the dimensionality are possible. One example is the applications of the distributional Littleā€™s law to the GI/G/1 queue. Possible extensions of this method are discussed.
Feb. 22nd

*SUR 3040*
David Stanford

Statistics and Actuarial Science

Western (Ontario) University
The Affine Accumulating Priority Queue: a model which prioritises based upon acuity and waiting time

Previous accumulating priority queues (APQ) in the literature have assumed that all arriving customers accumulate priority credits over time starting from an initial value of 0. The affine APQ model introduces a new element in terms of an initial class-dependent credit level, from which the accumulated priority grows linearly over time. In this presentation, we consider a two-class APQ, for which class-1 customers receive a positive initial credit upon arrival. (Without loss of generality the class-2 customers continue to have an initial credit of 0.) We assess the impact of the initial priority score upon the waiting time distribution for the lower class of customers, and insights regarding the higher priority class waiting time distribution as well. Numerical examples will be presented to illustrate the trends we observe.

This is joint work with Maryam Mojalal, Richard Caron, Peter Taylor and Ilze Ziedins.
Jan. 25th

via COCANA (Kelowna)

*SUR 2746*
Paulo J. S. Silva

Using the spectral proximal gradient method to decompose a matrix as the sum of a sparse plus a low-rank term

Further details available from the COCANA Website.
Dec. 11th



*SUR 2710*
Bolong He

M.Sc. project presentation

Senior Supervisor: T. Stephen
Analysis of Firefighter Absences and Hiring Schedule Optimization at the Surrey Fire Department

We study staffing issues at the Surrey Fire Department with a view to understanding and optimizing the annual hiring cycle for full-time firefighters. This project begins with a discussion of a previous model used by the Fire Department which predicts absences based on seasonally adjusted historical data and then optimizes the hiring cycle based on a simulation. We extend the analysis of the data to include the age cohort as a variable and compare short-term and long-term absences. We then use time series to predict future absences and use these predictions along with additional constraints to optimize the hiring schedule.
Dec. 4th



*SUR 3040*
Timothy Yusun

Ph.D. thesis defence

Senior Supervisor: T. Stephen
On the Circuit Diameters of Polyhedra

we develop a framework to study the circuit diameters of polyhedra. The circuit diameter is a generalization of the combinatorial (edge) diameter, where walks are permitted to enter the interior of the polyhedron as long as steps are parallel to its circuit directions. Because the circuit diameter is dependent on the specific realization of the polyhedron, many of the techniques used in the edge case do not transfer easily. We reformulate circuit analogues of the Hirsch conjecture, the d-step conjecture, and the nonrevisiting conjecture, recovering the edge case equivalences in the circuit case. To do this we adapt the notion of simplicity to work with circuit diameter, and so we define C-simplicity and wedge-simplicity. Then, we prove the circuit 4-step conjecture, including for unbounded polyhedra, by showing that the original counterexample U4 to the combinatorial analogue satisfies the Hirsch bound in the circuit case, independent of its realization. This was the first known counterexample to Hirsch, and several families of counterexamples are constructed from U4. In particular, the unbounded Hirsch conjecture could still hold in the circuit case. We also use computational methods to study Q4, the bounded counterpart to U4, and give two realizations with different circuit diameters. It remains open whether Q4 is circuit Hirsch-sharp; however, we are able to lower the distance bound for at least one direction between the two far vertices of Q4. Finally, we present some auxiliary results involving representations of polyhedra and circuit calculations.
Oct. 26th

via COCANA (Kelowna)

Hamid Afshari

University of Manitoba and UBC Okanagan
Improving the Resilience of Energy Flow Exchanges in Eco-Industrial Parks: Optimization Under Uncertainty

Further details available from the COCANA Website.
Sept. 28th

via COCANA (Kelowna)

Gord Lovegrove

UBC Okanagan
Models of Bicycle Rider Perceptions Related to Safety and Comfort

Further details available from the COCANA Website.
Sept. 16th

*Saturday, 8:30-4:00*


Details of the Fall 2017 West Coast Optimization Meeting here.
Sept. 14th

via COCANA (Kelowna)

Scott Lindstrom

University of Newcastle (Australia)
Douglas-Rachford Method for Non-Convex Feasibility Problems

Further details available from the COCANA Website.

Archives of the 2006-07, 2007-08, 2008-09, 2009-10, 2010-11, 2011-12, 2013-14, 2014-15, 2015-16 and 2016-17 SFU Operations Research Seminars.

Last modified September 12th, 2017.