FEASIBLE INTERIOR POINT METHOD FOR LINEAR COMPLEMENTARITY PROBLEM
No Thumbnail Available
Date
2024
Authors
CHALEKH RANDA
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The purpose of this study is to solve the linear complementarity problem (LCP). To this end, we
divided this work into two parts. In the first half, we provide an interior-point algorithm based on
two novel parametric kernel functions. Proving that, under suitable assumptions, we guarantee the
existence and uniqueness of the solution to the LCP. Afterwards, we show that a certain choice of
the barrier degrees of our functions coincide with the best-known iteration bound for large-update
methods. Finally, we offer some numerical results that prove the utility of the proposed algorithm. In
the second part, we employ a smoothing-type algorithm and assuming a reasonable assumption, we
can describe the LCP as an NP-hard absolute value equation (AVE). So we must rewrite AVE as a set
of smooth equations and introduce two smoothing functions. Then, we show that the method is welldefined
when the singular values of the matrix related to AVE exceed one and that it is convergent
under the same assumption. We also demonstrate the numerical efficacy of this algorithm using these
two functions.