Lamar IE Optimization Team
The Lamar IE Optimization team studies a variety of optimization problems. Our primary mission is to provide high quality optimization tools to external and internal customers to solve difficult optimization problems with heuristic and optimal solution procedures. Our team is focused on designing high quality algorithms and software implementations to find solution to difficult problems.
Our team is currently working on using emerging low cost computer technology (GPU and others) to solve challenging research problems. The GPU computing platform can dramatically improve the speed of heuristic searches. Our group is new with an official kickoff date of January 16, 2009. Beyond GPU computing, our group has experience solving industry problems with traditional tools (CPlex, COIN-OR, and heuristics developed in C++).
Faculty
Technology
Our team has over 10 year experience in industry in software development and optimization.
Our current research focus is on using GPU technology to accelerate meta heuristics for combinatorial and nonlinear optimization. We are currently using CUDA/C++ for GPGPU software development.
Beyond GPU technology, our team has access and development skills with a wide range of OR tools including CPLEX, COIN-OR, AMPLE, and Arena. For displaying solutions and developing user interactive environments, members of our team have used C++, Visual Basic, Map Windows GIS, haptics, and Open GL. For databases, our team members have used Oracle, SQL Sever, and Access. Most of our software development is on a PC platform (or cluster of PC) with the goal of developing low cost / high performance optimization tools. We have a wide range of computer resource in the industrial engineering department including many high end NVIDIA graphics cards to support our activities.
Collaboration
Our team is always looking for interesting research and consulting projects. For internal customers and basic research collaboration, we provide basic consulting services for free as a service. Please contact Dr. James Curry , Dr. Weihang Zhu, or Dr. Alberto Marquez.
Funded Projects
1. Reverse Logistics for Hazardous Waste Processing, Sponsored by Texas Hazardous Waste Research Center, 09/2008 ~ 08/2009.
2. Procurement Modeling and Optimization for a paper mill (06/2008 ~ 11/2008).
Unfunded and Student Projects
1. Tabu search for QAP on GPU platform.
This project developed a GPU based Tabu search algorithm for the QAP problem. The computational results showed a speedup of 40 by using GPU hardware. A paper based on this research was accepted to IJPR.
2. Nonlinear optimization on a GPU platform.
This project developed several GPU based optimization heuristics in a unified software platform. The heuristics include Genetic Algorithms, Pattern Search, and Ant Colony. Computational results showed a speedup of 20 to over 100. Pattern search was found to be particular effective in this hardware environment in combination with a meta-heuristic.
3. Parallel machine scheduling on a GPU platform.
This project is currently active with a MS thesis student actively working on the topic.
4. Route flexibility in cellular manufacturing systems.
This project is currently active with a doctoral student actively working on the topic.
5. Incinerator scheduling using a SA-LP heuristic.
This doctoral research project developed a method for efficiently scheduling waste feeds to an industrial hazardous waste incinerator. The research developed a simulated annealing linear programming heuristic to solve the problem. Operating data from an industrial incinerator was used to test the algorithm. The algorithm proved effective on scheduling commercial size problems. Computational experiments indicate the problem is less challenging if a single permit limit is more binding than the individual lance limits. The algorithm was implemented in Microsoft Visual C++TM with COIN-OR CLPTM employed to solve the linear program. Microsoft AccessTM was employed to store the problem and output schedules so that the algorithm can be deployed to a generic incinerator.
Publications as a Team
- Weihang Zhu, J. Curry, A. Marquez, SIMD Tabu Search for Quadratic Assignment Problem with Graphics Hardware Acceleration, accepted by International Journal of Production Research.
- Weihang Zhu and James Curry. Particle Swarm with Graphics Hardware Acceleration and Local Pattern Search on Bound Constrained Problems, 2009 IEEE Swarm Intelligence Symposium.
- Weihang Zhu, J. Curry, Multi-walk Parallel Pattern Search on a GPU Computing Platform, International Conference of Computational Science, Baton Rouge, LA, USA, 2009
- Weihang Zhu, James Curry, Anjali Mishra and Victor Zaloom (2009). A study of Ant Colony-Based Parallel Machine Scheduling with Graphics Hardware Acceleration, Proceedings of the ASME 2009 International Manufacturing Science and Engineering Conference, October 4-7, 2009, West Lafayette, Indiana, USA.
- Weihang Zhu and James Curry (2009). Parallel Ant Colony for Nonlinear Function Optimization with Graphics Hardware Acceleration, Proceedings of IEEE System, Man and Cybernetics Conference, Oct 11~14, 2009, San Antonio, TX, USA.
Additional Publications
A set covering formulation for agile capacity planning within supply chains. J.K. Cochran, A. Marquez. International Journal of Production Economics. Vol. 95 (2005) pp 139-149.
Two-stage simulation optimization for agile manufacturing capacity planning: Marquez, A., Cochran, K. Shunk D. International Journal of Production Research. Vol. 41 06/2003. 1181-1197
A Cross Entropy Algorithm for the Knapsack Problem with Setups, M. Caserta, E. Quiñonez and A. Márquez, Submitted to "Computers and Operations Research" (Accepted January 2006/in Press)
W. Li, H. Peng, Weihang Zhu, D. Sheng, J. Chen, An Immune-Tabu Hybrid Algorithm for Thermal Unit Commitment Problem in Power Plant Optimization, accepted by Journal of Zhejiang University – Science A
Weihang Zhu, A Methodology for Building up an Infrastructure for Haptically Enhanced Computer-Aided Design Systems, Transaction of the ASME: Journal of Computing and Information Science in Engineering, Issue 4, 2008
Curry, J. and Peters, B. A. (2005). Rescheduling parallel machines with stepwise increasing tardiness and machine assignment stability objectives. International Journal of Production Research, 43 (15), 3231-3246.
Smith, J.S., Peters, B.A, Curry, J. and Gupta, D., “Prototype software model for designing intruder detection systems with simulation,” Proceedings of SPIE ’98, Orlando, FL, April, 1998
Source Code
We are currently working on providing samples of the software developed by our lab to aid other researchers.
Current Students
Doctoral Students
Pavan Masvekar - He is a current doctoral student working on multi-objective optimization (Fall 2009 - ).
MS Thesis
Anjali Mishra-- She is currently working on GPGPU based heuristics for parallel machine scheduling (Spring 2007—Anticipated Graduation Fall 2009).
Former Students
Doctoral Students
Carol Schulte—She developed scheduling procedures for hazardous waste incinerators. The scheduling algorithms minimized the amount of on hand inventory subject to regulatory restrictions on emissions. She used a SA-LP algorithm implemented in C++ and Coin-OR to solve the scheduling problem (Fall 2006– Graduation Spring 2008). She is currently working for McNesse State University in Lake Charles, La.MS Thesis
Pavan Masvekar— He studied the impact of driver hour restriction on the vehicle routing problem. (Spring 2007 - Graduation Fall 2008)
Prashant Mehta—He studied the impact of multiple factors on the real rate of return of mortgage based securities. (Spring 2007- Fall 2008)
Jovan Hill—He studied the impact of vendor selection for a global supply base. (Spring 2007 –Spring 2008)
"The views and opinions expressed in this document are strictly those of the author (Dr. James Curry) and do not necessarily reflect the views or opinions of the State of Texas, the Regents or officials of The Texas State University System, the Lamar University Administration, and Lamar University colleges or departments, or any recognized Lamar University organization. Comments on the contents of this document should be directed to the author(s)."
