Concentration inequalities for nonlinear matroid intersection

Abstract

In this work we propose new randomized rounding algorithms for matroid intersection and matroid base polytopes. We prove concentration inequalities for polynomial objective functions and constraints that have numerous applications and can be used in approximation algorithms for Minimum Quadratic Spanning Tree, Unrelated Parallel Machines Scheduling and Scheduling with Time Windows and Nonlinear objectives. We also show applications related to Constraint Satisfaction and Dense Polynomial optimization. Copyright © SIAM.