Title :
Bounds for codes via semidefinite programming
Author_Institution :
Univ. of Texas at Brownsville, Brownsville, TX
Abstract :
Delsarte´s method and its extensions allow to consider the upper bound problem for codes in 2-point-homogeneous spaces as a linear programming problem with perhaps infinitely many variables. In this paper two extensions of Delsarte´s method via semidefinite programming are considered. The first approach shows that using as variables power sums of distances this problem can be considered as a finite semidefinite programming problem. This method allows to improve some upper bounds. The second approach extends the Bachoc-Vallentin method for spherical codes. In particular, an extension of Schoenberg´s theorem for multivariate Gegenbauer polynomials has been proved.
Keywords :
codes; linear programming; polynomials; 2-point-homogeneous spaces; Bachoc-Vallentin method; Delsarte method; Schoenberg theorem; finite semidefinite programming problem; linear programming; multivariate Gegenbauer polynomials; spherical codes; Constraint optimization; Error correction codes; Functional programming; Hamming distance; Kernel; Linear matrix inequalities; Linear programming; Polynomials; Symmetric matrices; Upper bound;
Conference_Titel :
Information Theory and Applications Workshop, 2009
Conference_Location :
San Diego, CA
Print_ISBN :
978-1-4244-3990-4
DOI :
10.1109/ITA.2009.5044951