Alternative proofs of the asymmetric Lovász local lemma and Shearer's lemma

Ioannis Giotis, Lefteris Kirousis, John Livieratos, Kostas I. Psaromiligkos, Dimitrios M. Thilikos

Research output: Contribution to journalConference articlepeer-review


We provide new algorithmic proofs for two forms of the Lovász Local Lemma: the Asymmetric version and Shearer's Lemma. Our proofs directly compute an upper bound for the probability that the corresponding Moser-type algorithms last for at least n steps. These algorithms iteratively sample the probability space; when and if they halt, a correct sampling, i.e. one where all undesirable events are avoided, is obtained. Our computation shows that this probability is exponentially small in n. In contrast most extant proofs for the Lovász Local Lemma and its variants use counting arguments that give estimates of only the expectation that the algorithm lasts for at least n steps. For the asymmetric version, we use the results of Bender and Richmond on the multivariable Lagrange inversion. For Shearer's Lemma, we follow the work of Kolipaka and Szegedy, combined with Gelfand's formula for the spectral radius of a matrix.

Original languageEnglish (US)
Pages (from-to)148-155
Number of pages8
JournalCEUR Workshop Proceedings
StatePublished - 2018
Event11th International Conference on Random and Exhaustive Generation of Combinatorial Structures, GASCom 2018 - Athens, Greece
Duration: Jun 18 2018Jun 20 2018

ASJC Scopus subject areas

  • General Computer Science


Dive into the research topics of 'Alternative proofs of the asymmetric Lovász local lemma and Shearer's lemma'. Together they form a unique fingerprint.

Cite this