On the Decidability of Distributed Tasks with Output Sets under Asynchrony and Any Number of Crashes

Abstract

This paper studies the decidability of task problems, i.e., distributed problems expressed as sets of distributed tasks. Specifically, we introduce a new class of task problems called Set of Output Sets (SOS) problems. An SOS problem $\Pi_O$ is defined by a set $O$ (called SOS), and requires that the set of sets of distinct output values produced across all executions corresponds exactly to $O$. We then demonstrate that this class of problems is decidable: there is a procedure determining whether any SOS problem is solvable asynchronously under $f$ crashes. The decision rule is as follows. Every SOS problem is solvable when $f=0$. For $f > 0$, an SOS problem is solvable if and only if the graph $G=(O,\subset)$ is connected. In this graph, each vertex is an output set in $O$, and two vertices are linked by an edge whenever one output set includes the other. One of the surprising implications of our results is that, replacing validity by a completeness property (which guarantees that all output sets of size at most $k$ are produced), $k$-set agreement is solvable under any number of crashes $f \geq 0$ for $k>1$, and unsolvable under $f>0$ crashes only for $k=1$ (consensus). Finally, we study a novel family of problems called $d$-disagreement, which requires the system to always produce $d$ different output values, and we show that its implementability condition is related to the harmonic series.