Abstract
Randomized Rounding (RR) is a simple and very efficient probabilistic method to convert a solution of a relaxed combinatorial optimization (CO) problem into an approximate solution to the original problem. Since its introduction by Raghavan and Thompson in [14], RR and variations of it have been successfully applied to many approximation algorithms for hard CO problems. Though RR can be applied to an even wider set of problems, often the performance guarantees of the RR based approximation algorithms are poor.