The discovery, more than a decade ago, of the relation between Distributed-Computing (DC) and Algebraic-Topology (AT) raised the specter of requiring checking task solvability to be intimately connected to expertise in AT. Yet, in the area of Centralized Algorithms proving a problem to be NP or PSPACE complete requires more algorithmic expertise than complexity one. In analogy, we show that in DC the equivalent of polynomial-time reductions, is read-write reductions. We define the notion of read-write reduction between distributed tasks, and show that all interesting known read-write impossible tasks can be proven impossible via read-write reduction to a task called Symmetry-Breaking (SB). Discovering a read-write reduction requires solely algorithmic expertise.
KeywordsComparison Version Input Tuple Processor Output Output Tuple Latin American Symposium
Unable to display preview. Download preview PDF.
- 8.Gafni, E.: DISC/GODEL presentation: R/WReductions, DISC 2004 (2004), http://www.cs.ucla.edu/~eli/eli/godel.ppt
- 11.Gafni, E., Kouznetsov, P.: Two Front Agreement with Application to Emulation and Robustness (to appear)Google Scholar
- 13.Fredrickson, N., Lynch, N.: Electing a Leader in a Synchronous Ring. J. ACM (January 1987)Google Scholar
- 16.Gafni E., (in preparation)Google Scholar
- 17.Ostrovsky, R.: Private communication to the authorGoogle Scholar