Least-Index Resolution of Degeneracy in Linear Complementarity Problems.
Abstract
This study centers on the circling phenomenon associated with degeneracy in linear complementarity problems and presents an easily implemented technique for resolving it. With certain exceptions, the device is to use the least-index for selecting the variable to leave the basic set. The results of this report pertain only to linear complementarity problems involving P-matrices or positive semi-definite matrices. With this restriction, it is shown that inclusion of the least-index pivot selection rule insures finiteness for the principal pivoting method of Dantzig and Cottle, Lemke's algorithm, and Cottle's parametric principal pivoting method. It is shown that for circling to occur in the principal pivoting method, the matrix must have order at least four, and for Lemke's algorithm it must be at least three. Examples are given showing that these bounds are sharp. Finally, Murty's version of Bard's method is extended from P-matrices to the positive semi-definite case. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1979
- Accession Number
- ADA080113
Entities
People
- Yow-yieh Chang
Organizations
- Stanford University