The operation point units of distributed constraint solvers

We propose a way to define the logic computation cost of operations to be used in evaluations of scalability and efficiency for simulated distributed constraint reasoning (DCR) algorithms. We also report experiments showing that the cost associated with a constraint check, even within the same algorithm, depends on the problem size. The DCR research has seen heated debate regarding the correct way to evaluate efficiency of simulated algorithms. DCR has to accommodate two established practices coming from very different fields: distributed computing and constraint reasoning. The efficiency of distributed algorithms is typically evaluated in terms of the network load and overall computation time, while many (synchronous) algorithms are evaluated in terms of the number of rounds that they require. Constraint reasoning evaluates efficiency in terms of constraint checks and visited search-tree nodes. We argue that an algorithm has to be evaluated from the point of view of specific operating points, namely of possible or targeted application scenarios. We then show how to report efficiency for a given operating point based on simulation, in particular we show how to tune the distribution used to generate the message latency costs as function of the logic computation unit. New real and simulated experiments show that the cost of a constraint check varies with the size of the problem. We also show how to select logic units for nogood-based algorithms, such that the unit is constant with the size of the problem,