Algorithmic Chernoff‐Hoeffding inequalities in integer programming

Abstract

Proofs of classical Chernoff-Hoeffding bounds has been used to obtain polynomial-time implementations of Spencer's derandomization method of conditional probabilities on usual finite machine models: given m events whose complements are large deviations corresponding to weighted sums of n mutually independent Bernoulli trials, Raghavan's lattice approximation algorithm constructs for 0-1 weights, and integer deviation terms in O(mn)-time a point for which all events hold. For rational weighted sums of Bernoulli trials the lattice approximation algorithm or Spencer's hyperbolic cosine algorithm are deterministic precedures, but a polynomial-time implementation was not known. We resolve this problem with an O(mn2 log y ) -t i m e algorithm, whenever the probability that all events hold is at least > 0. Since such algorithms simulate the proof of the underlying large deviation inequality in a constructive way, we call it the algorithmic version of the inequality. Applications to general packing integer programs and resource constrained scheduling result in tight and polynomial-time approximations algorithms. 0 1996 John Wiley & Sons, Inc.

How the result of graph clustering methods depends on the construction of the graph

“…Setting V ∼ Bin(n − 2, 4k n /(n − 1)), we certainly have 0 < k n /(n − 2) < 4k n /(n − 1) for n ≥ 3 and thus we obtain with a tail bound for the binomial distribution from Srivastav and Stangier (1996), which was first proved in Angluin and Valiant (1979),…”

Construction of Low-Discrepancy Point Sets of Small Size by Bracketing Covers and Dependent Randomized Rounding

“…It thus avoids the general, but more costly solution by Srivastav and Stangier [SS96]. The latter was a break-through from the theoretical point of view as it showed that randomized rounding for arbitrary linear constraints can be derandomized.…”

Solving Generalized Maximum Dispersion with Linear Programming

“…. , p this can be done as in the proof of Theorem 2.13 in [19]. For A 0 the Markov inequality helps.…”

“…, X i and then taking the expectation. With this observation the proof of Theorem 2.13 in [19] can be lifted to cover all events A 0 , A 1 , . .…”