Research Activities

 

 

 

My research interests are in the field of Operations Research, with a focus on ergodic theory and discrete optimization (analysis and heuristics).   In particular, I enjoy applying the techniques of Operations Research to a variety of fields and applications.  To achieve this, I have participated in a variety of projects.  The following is a list of projects that I am currently involved in. 

 

Projection Pursuit

The objective of this research is to study the problem of finding an optimal projection of data contained in a high dimensional space to a lower dimensional space.  The data is partitioned into sets, where the sets represent the origin of the data (where the data came from).  Optimality, in this case, will preserve the separation of the data.  The Bayes Classifier is used to classify the data in the low dimensional space. 

 

Hybrid Local Search Algorithms

This research develops hybrid local search algorithms for approaching discrete optimization problems.  The idea is to merge convergent local search algorithms (e.g., simulated annealing) with non-convergent algorithms (e.g., pure local search, Monte Carlo search).  Thereby, creating algorithms that combine heuristic procedures that guarantee long term convergence (globally optimal solutions) with heuristic procedures that guarantee reasonable finite-time performance (locally optimal solutions). 

 

Collaborators:  Sheldon H. Jacobson

 

Discrete Manufacturing Process Design Optimization*

The goal of this research is to construct efficient and effective optimization algorithms to identify optimal/near-optimal manufacturing process designs, where the finished unit meets certain geometric and microstructural specifications, and is produced at minimum cost.  Computer simulation models of the manufacturing process designs are provided by the Materials Process Design Branch of the Air Force Research Laboratory, Wright Patterson Air Force Base (Dayton, Ohio, USA).

 

Collaborators:  Derek Armstrong, Sheldon H. Jacobson, Kelly Sullivan and Allan Johnson

Papers: Vaughan, D.E., Jacobson, S.H. and Armstrong, D. (2000)

 

Simultaneous Generalized Hill Climbing Algorithms*

This research focuses on approaching sets of related discrete optimization problems (i.e., overlapping traveling salesman problems, machine shop scheduling problems with similar constraints).  The aim of the research is to develop an algorithm that can approach the whole set in one application.  Information gained while optimizing over one discrete optimization problem is used to initialize in the subsequent discrete optimization problem.

 

Collaborators:  Sheldon H. Jacobson

Papers: Vaughan, D.E. and Jacobson, S.H. (2001b, 2001c)

 

Formulating the Meta-Heuristic Tabu Search

The goal of this project is to mathematically formulate probabilistic tabu search to be used in conjunction with other heuristics (i.e., simulated annealing).  The framework allows the algorithm to be modeled using nonstationary Markov chain theory.  This framework allows already existing convergence theories to embody applications of tabu search. 

 

Collaborators:  Sheldon H. Jacobson

Papers: Vaughan, D.E. and Jacobson, S.H. (2001a)

 

Optimal Search Strategy Problem*

This research models optimal strategies for conducting military searches using a set of platforms (i.e., helicopters, planes) as the well-known traveling salesman problem.  Local search algorithms as well as simultaneous generalized hill climbing algorithms are used to approach these problems.

 

 Collaborators:  Darrall Henderson, Sheldon H. Jacobson

 Papers:  Henderson, D., Vaughan, D.E., Jacobson, S.H. and Wakefield, R. (2001a)

               Vaughan, D.E., Henderson, D. and Jacobson, S.H. (2001)

 

Optimal Earthmoving Vehicle Routes

This research develops a mathematical model for minimizing the cost of operating large capacity vehicles to level project sites prior to a construction project.  The model is shown to be NP-hard and approached using heuristics.  This research will aid in future collaboration between the field of operation research and the field of construction engineering. 

 

Collaborators:  Darrall Henderson, Ron Wakefield, Sheldon H. Jacobson

Papers:  Henderson, D., Wakefield, R., Vaughan, D.E. and Jacobson, S.H. (2001b)

              Henderson, D., Vaughan, D.E., Jacobson, S.H., Wakefield, R. and Sewell, E.C. (2001)

 

Analysis of the Urn Problem

This research uses probability theory as well as computational techniques to approach variations of the well-known urn problem. 

 

Collaborators:  John E. Kobza, Sheldon H. Jacobson

Papers:  Vaughan, D.E., Kobza, J. and Jacobson, S.H. (2001)

 

Hybrid Neighborhood Functions

This research develops hybrid neighborhood functions for approaching discrete optimization problems.  Particular focus is on the traveling salesman problem.  So far a hybrid neighborhood function has been developed that combines the 2-opt and 4-exchange neighborhood functions.

 

Collaborators:  Sheldon H. Jacobson, Darrall Henderson

Papers:  Vaughan, D.E., Henderson, D., Jacobson, S.H. (2001)

 

 

 

*This work is supported in part by the Air Force Office of Scientific Research (F49620-01-1-0007, F49620-98-1-0432) and the National Science Foundation (DMI-9907980).