Interesting tasks are scarce. Yet, they are essential as an investigation material, if we are to understand the structure of the tasks world. We propose a new collection of families of tasks called 0-1 Exclusion tasks, and show that families in this collection are interesting.
A 0-1 Exclusion task on n processors is specified by a sequence of n *** 1 bits b (1),b (2),...,b (n *** 1). For participating set of size k , 0 < k < n , each processor is to output 0 or 1 but they should not all output b (k ). When the participating set is of size n , then they should all output neither all 0's nor all 1's. A family of tasks, one for each n , is created by considering an infinite sequence of bits b (k ), k = 2,3,..., such that the sequence that specifies instant n , is a prefix of the sequence that specifies the n + 1'st instance.
Only one family in the collection, the one specified by b (1) = b (2) = ...= 1, was implicitly considered in the past and shown to be equivalent to Set- Consensus. In this initial investigation of the whole collection we show that not all of its members are created equal. We take the family specified by b (1) = 1,b (2) = b (3) = ... = 0, and show that it is read-write unsolvable for all n , but is strictly weaker than Set-Consensus for n odd.
We show some general results about the whole collection. It is sandwiched between Set-Consensus from above and Weak-Symmetry-Breaking from below. Any Black-Box of n ports that solves a 0-1 Exclusion task, can be used to solve that task for n processors with ids from unbounded domain.
Finally we show an intriguing relation between Strong-Renaming and the 0-1 Exclusion families, and make few conjectures about the implementations relationships among members of the collection, as well as possibly tasks outside it.
[1]
Eli Gafni,et al.
Immediate atomic snapshots and fast renaming
,
1993,
PODC '93.
[2]
Eli Gafni,et al.
A simple algorithmically reasoned characterization of wait-free computation (extended abstract)
,
1997,
PODC '97.
[3]
Achour Mostéfaoui,et al.
Exploring Gafni's Reduction Land: From Omegak to Wait-Free Adaptive (2p-[p/k])-Renaming Via k-Set Agreement
,
2006,
DISC.
[4]
Maurice Herlihy,et al.
The topological structure of asynchronous computability
,
1999,
JACM.
[5]
Nir Shavit,et al.
Atomic snapshots of shared memory
,
1990,
PODC '90.
[6]
Eli Gafni,et al.
Three-Processor Tasks Are Undecidable
,
1999,
SIAM J. Comput..
[7]
Eli Gafni.
A simple algorithmic characterization of uniform solvability
,
2002,
The 43rd Annual IEEE Symposium on Foundations of Computer Science, 2002. Proceedings..
[8]
Soma Chaudhuri,et al.
Agreement is harder than consensus: set consensus problems in totally asynchronous systems
,
1990,
PODC '90.
[9]
Eli Gafni,et al.
Generalized FLP impossibility result for t-resilient asynchronous computations
,
1993,
STOC.
[10]
Eli Gafni,et al.
Immediate Atomic Snapshots and Fast Renaming (Extended Abstract).
,
1993,
PODC 1993.
[11]
Sergio Rajsbaum,et al.
New combinatorial topology upper and lower bounds for renaming
,
2008,
PODC '08.
[12]
Hagit Attiya,et al.
Renaming in an asynchronous environment
,
1990,
JACM.
[13]
Maurice Herlihy,et al.
Subconsensus Tasks: Renaming Is Weaker Than Set Agreement
,
2006,
DISC.
[14]
Nancy A. Lynch,et al.
Impossibility of distributed consensus with one faulty process
,
1985,
JACM.