performance - What are the "new hash functions" in cuckoo hashing? -


i reading cuckoo hashing pagh , rodle , can't understand meaning of paragraph:

it may happen process loops, shown in fig. 1(b). therefore number of iterations bounded value “maxloop” specified in section 2.3. if number of iterations reached, rehash keys in tables using new hash functions, , try once again accommodate nestless key. there no need allocate new tables rehashing: may run through tables delete , perform usual insertion procedure on keys found not @ intended position in table.

what mean using new hash functions?
in insert algorithm table resized. supposed have "pool" of hash functions use somehow? how create pool?

yes, they're expecting new hash functions, say. fortunately, don't require pile of new algorithms, different hashing behavior on current data set.

take @ section 2.1 of paper, , appendix a. discusses construction of random universal hash functions.

a simple, illustrative example, suppose you've got normal hash function like, operates on blocks of bytes. we'll call h(x). want use produce family of new, different hash functions h_n(x). well, if h(x) good, , requirements weak, can define h_n(x) = h(concat(n,x)). don't have nice strong guarantees behaviors of h_n(x), you'd expect of them different.


Comments

Popular posts from this blog

angularjs - ADAL JS Angular- WebAPI add a new role claim to the token -

php - CakePHP HttpSockets send array of paramms -

node.js - Using Node without global install -