back to table · edit · history · where entries came from · files
Title: Cuckoo hashing thresholds of random $k$-uniform hypergraphs-Definition: 'An $\ell$-orientable hypergraph is one whose edges can be assigned to- incident vertices so that no vertex receives more than $\ell$ edges. In the random- $k$-uniform hypergraph CITE{WikiHypergraph} with $n$ vertices and $m=\lfloor cn\rfloor$- edges, $c^*_{k,\ell}$ is the threshold density: as $n\to\infty$, the probability- of $\ell$-orientability tends to $1$ for $c<c^*_{k,\ell}$ and to $0$ for $c>c^*_{k,\ell}$- CITE{FKP}.'+Definition: An $\ell$-orientation assigns each edge to one of its vertices, with at+ most $\ell$ edges at any vertex. In the random $k$-uniform hypergraph CITE{WikiHypergraph}+ with $n$ vertices and $\lfloor cn\rfloor$ edges, $c^*_{k,\ell}$ is the high-probability+ threshold for $\ell$-orientability CITE{FKP}. Keywords: orientability threshold, blocked cuckoo hashing, multiple-choice hashing, balanced allocation
constraints: $\ell\geq1$ Comments:- comment-orientation: In hashing language, edges are keys and vertices are buckets,- so $c^*_{k,\ell}$ is the threshold for the number of keys per bucket in offline- cuckoo hashing CITE{WikiCuckoo} with $k$ choices and bucket capacity $\ell$ CITE{DGMMPR}.- The load factor, keys per slot, is $c^*_{k,\ell}/\ell$.+ comment-orientation: The high-probability threshold means that, as $n\to\infty$,+ $\ell$-orientability has probability tending to $1$ for $c<c^*_{k,\ell}$ and to+ $0$ for $c>c^*_{k,\ell}$. In hashing language, edges are keys and vertices are+ buckets, so $c^*_{k,\ell}$ is the threshold for the number of keys per bucket+ in offline cuckoo hashing CITE{WikiCuckoo} with $k$ choices and bucket capacity+ $\ell$ CITE{DGMMPR}. The load factor, keys per slot, is $c^*_{k,\ell}/\ell$. comment-conventions: 'The density is $c=m/n$, so the average hypergraph degree is $kc$. Dietzfelbinger, Goerdt, Mitzenmacher, Montanari, Pagh and Rink tabulate
Sign in to restore an earlier version.