Parallelization of Horner Method: Implementation Strategy and Error Analysis
Résumé
Minimizing the computational cost of polynomial evaluation is a main problem in Computational Science. We present a modification of Horner’s method to exploit its advantages in the evaluation of sparse polynomials. We also propose a polynomial partitioning that allows to perform Horner’s method in parallel. In addition, we provide an analysis of the numerical error between the proposed parallel method and the classic algorithm.