Extractors with weak random seeds

• Authors: Ran Raz
• Venue: Proceedings of the thirty-seventh annual ACM symposium on Theory of computing
• Year: 2005

We show how to extract random bits from two or more independent weak random sources in cases where only one source is of linear min-entropy and all other sources are of logarithmic min-entropy. Our main results are as follows: A long line of research, starting by Nisan and Zuckerman[14], gives explicit constructions of seeded-extractors, that is, extractors that use a short seed of truly random bits to extract randomness from a weak random source. For every such extractor E, with seed of length d, we construct an extractor E′, with seed of length d′=O(d), that achieves the same parameters as E but only requires the seed to be of min-entropy larger than (1⁄2+δ) •d′ (rather than fully random), where δ is an arbitrary small constant. Fundamental results of Chor and Goldreich and Vazirani [6,21] show how to extract Ω(n) random bits from two (independent) sources of length n and min-entropy larger than (1⁄2δ) • n, where δ is an arbitrary small constant. We show how to extract Ω(n) random bits (with optimal probability of error) when only one source is of min-entropy (1⁄2+δ) •n and the other source is of logarithmic min entropy.3 A recent breakthrough of Barak, Impagliazzo and Wigderson[4] shows how to extract Ω(n) random bits from a constant number of (independent) sources of length n and min-entropy larger than δ n, where δ is an arbitrary small constant. We show how to extract Ω (n) random bits (with optimal probability of error) when only one source is of min-entropy δ n and all other (constant number of) sources are of logarithmic min-entropy.A very recent result of Barak, Kindler, Shaltiel, Sudakov and Wigderson[5] shows how to extract a constant number of random bits from three (independent) sources of length n and min-entropy larger than δn, where δ is an arbitrary small constant. We show how to extract Ω(n)/ random bits, with sub-constant probability of error, from one source of min-entropy δ n and two sources of logarithmic min-entropy.In the same paper, Barak, Kindler, Shaltiel, Sudakov and Wigderson[5] give an explicit coloring of the complete bipartite graph of size 2n x 2n with two colors, such that there is no monochromatic subgraph of size larger than 2δn x 2 2δn, where δ is an arbitrary small constant. We give an explicit coloring of the complete bipartite graph of size 2n x 2n with a constant number of colors, such that there is no monochromatic subgraph of size larger than 2δn x n5.We also give improved constructions of mergers and condensers. In particular, We show that using a constant number of truly random bits, one can condense a source of length n and min-entropy rate δ into a source of length Ω (n) and min-entropy rate 1-δ, where δ is an arbitrary small constant.We show that using a constant number of truly random bits, one can merge a constant number of sources of length n, such that at least one of them is of min-entropy rate 1-δ, into one source of length Ω(n) and min-entropy rate slightly less than 1-δ, where δ is any small constant.