INFORMS Logo
 

Optimization Techniques I

Session: MC33
Date/Time: Monday 14:45-16:15
Type: Contribute
Sponsor:
Track:
Cluster:
Room: Windsor South
Chair: Arne S. Drud
Chair Address: ARKI Consulting & Develop. A/S, Bagsvaerdvej 246A, Bagsvaerd, 2880 , Denmark
Chair E-mail:

MC33.1 A General Pre-Processor for GAMS Models Arne S. Drud --- ARKI Consulting & Develop. A/S, Bagsvaerdvej 246A, Bagsvaerd, 2880 , Denmark (adrud@arki.dk)
We describe a general purpose pre- and post-processor for GAMS models, independent of and usable with most GAMS solvers. The pre-processor removes variables, aggregates or disaggregates constraints based on user options, adds domain bounds for nonlinear models, checks scaling, etc. We report experiments with various GAMS solvers.

MC33.2 Modeling & Experimentation for Multiple Response Optimization Daniel R. Rand, Kevin J. Dooley --- Univ. of MN, 805 28th St. NW, Rochester, MN 55901 , (rand0030@maroon.tc.umn.edu)
An algorithm for multiple response optimization will be described that satisfies 2 objectives: to formulate an aggregate response function when individual responses have already been modeled as functions of control variables and to expand response surface search techniques to pursue a collective optimum (for multiple responses) with efficient experimentation.

MC33.3 MProbe: Estimating the 'Shape' of Nonlinear Functions John W. Chinneck --- Carleton Univ., Systems & Comp. Eng., 1125 Colonel by Dr., Ottawa, Ontario, K1S 5B6 , Canada (chinneck@sce.carleton.ca)
The 'shape' of nonlinear functions is crucial in NLP: are they convex or concave or almost so or both? MProbe provides an estimate of shape in a specified region. A function that is analytically nonlinear may be simpler, e.g., linear, in the region of interest.

MC33.4 A Deterministic Method for the Global Optimization of Heat Exchanger Networks Juan M. Zamora, Ignacio E. Grossmann --- Carnegie Mellon Univ., Dept. of Chem. Eng., Pittsburgh, PA 15213-3890, (jz29@andrew.cmu.edu)
A rigorous global optimization method is presented for the synthesis of heat exchanger networks with no stream splitting. We propose convex relaxations of the heat transfer equation and use linear convex envelopes for concave costs. These are used within a B&B search that involves MINLP subproblems with contaction steps.


INFORMS Online

Questions on membership, subscriptions and the like should go to INFORMS Customer Service. 
Questions/comments of a general nature about this Web site should go to Editor, IOL. 
Copyright © Institute for Operations Research and the Management Sciences