Decision-Making Patent to Solve Unsolvable Problems
By Office of Communication
Posted on March 24, 2012, March 24, 2012

UT Arlington industrial engineers have patented an innovative method that can obtain optimal decisions for a broad class of real-world problems not previously solvable.

Bill Corley and Jay Rosenberger, professor and associate professor in the Industrial & Manufacturing Systems Department, were recently issued a patent entitled "System, Method and Apparatus for Allocating Resources by Constraint Selection."

Linear programming is a mathematical description of a vast number of decision problems occurring throughout the business and scientific worlds. Solving these problems allows an organization to maximize profit, minimize costs or allocate resources.

"Linear programming is the most widely used computational model in the business and scientific worlds," Corley said. "It will now become much more important. That's the bottom line. We drastically improved over 60 years of research for computing with this ubiquitous decision model."

In logistical applications, linear programming solutions are needed to transport materials and people efficiently. In the telecommunications industry they can route either data or cars in transit to their destinations in the quickest manner. Linear programming also is used in statistics and the sciences as a computational tool.

The newly patented approach to solving linear programming problems uses a process called Constraint Optimal Selection Techniques, or COSTs, to reduce the number of calculations needed to make an optimal decision thousands of times faster for large decision problems with huge numbers of solution variables and restrictions on these variables.

"It will allow faster decision-making in today's high-speed, high-tech, ever-accelerating world," Corley said.

Rosenberger added that new approach expands the applications of linear programming. It can solve enormous problems for which previous methods could take months of computer time.

"It will also give answers to currently unsolvable nonlinear decision problems by approximating them with enormous linear programming problems," Rosenberger noted.

Engineering Dean Jean-Pierre Bardet said the patent represents the kind of research that benefits not just the industrial engineering discipline but any scientific and engineering field in search of optimal solutions for complex problems.

"It becomes the ultimate computing tool," Bardet said.

The patent is an example of the kind of research that's under way at The University of Texas at Arlington, a comprehensive research institution of 33,439 students in the heart of North Texas. Visit www.uta.edu to learn more information.

Article provided courtesy of UT Arlington

UT Arlington
News, Education