Date of Award


Degree Type


Degree Name

Doctor of Philosophy


Applied Mathematics


Dhavide Aruliah

Second Advisor

Rob Corless

Third Advisor

Laureano Gonzalez-Vega


The purpose of this thesis is to provide the mathematical tools for computing simple isolated roots of bivariate (and univariate) polynomial equations described by their values or some of the derivatives at some nodes. The polynomial root-finding problem has a long and rich history in computational mathematics, mainly because of its frequent applications in engineering. It may be the case that the polynomial system as it appears naturally in the application formulation is not presented in the familiar monomial basis. The approach we take here avoids conversion between bases. This is important because converting from the basis in which the polynomials are given to the monomial basis (or any other basis) is likely to degrade the conditioning of the problem. The root-finding algorithm presented in this thesis is a hybrid of a resultant-type algorithm and eigenvalue techniques. Here, we use some new formulations for the Bézout matrix and the companion matrix pencil associated with a pair of bivariate polynomial equations described by their values. These two matrix constructions are the essential tools to compute all the simple isolated roots of a bivariate polynomial system. We use the Bézout matrix to eliminate one of the variables and then we construct the companion matrix pencil to convert our polynomial root-finding problem to a generalized eigenvalue problem. We then use conventional software to compute the eigenvalues. A complete chapter of this thesis is devoted to studying the application of the presented algorithm in Computer Aided Geometric Design. There are also other geometric applications for which our root-finding algorithm can apply. The applications we have introduced here are merely representative. Our experiments show that the algorithm presented here is numerically stable. While this thesis has not been primarily concerned with efficiency, the algorithms presented here are efficient enough to suggest that the effort involved in making them competitive will be worthwhile.



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.