Problem 27: Modeling of Distributed Computation

From Open Problems in Sublinear Algorithms
Revision as of 01:50, 7 March 2013 by Krzysztof Onak (talk | contribs) (updated header)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
Suggested by Paul Beame
Source Kanpur 2009
Short link

MapReduce has recently inspired two distributed models of computation in the theory community. One is the MUD model of Feldman et al. [FeldmanMSSS-10]. In this model they assume that every worker has at most a polylogarithmic amount of space available, which in total gives at most $\tilde O(n)$ space, where $n$ is the input size (in the order of at least terabytes). The other model of computation, due to Karloff et al. [KarloffSV-10], assumes that each of $n^{1-\epsilon}$ workers has at most $n^{1-\epsilon}$ space, where $\epsilon$ is a fixed positive constant. This totals to $n^{2-2\epsilon}$ space in the entire system. Can one design an interesting and practical model that only uses $n^{1+o(1)}$ space/resources?