Abstract. The aim of this article is to introduce the reader to the study of testing graph properties, while focusing on the main models and issues involved. No attempt is made to provide a comprehensive survey of this study, and specific results are often mentioned merely as illustrations of general themes.Keywords: Graph Properties, randomized algorithms, approximation problems.This survey was originally written for inclusion in [32].
The General ContextIn general, property testing is concerned with super-fast (probabilistic) algorithms for deciding whether a given object has a predetermined property or is far from any object having this property. Such algorithms, called testers, obtain local views of the object by making adequate queries; that is, the object is seen as a function and the tester gets oracle access to this function, and thus may be expected to work in time that is sub-linear in the size of the object.Looking at the foregoing formulation, we first note that property testing is concerned with promise problems (cf. [26,30]), rather than with standard decision problems. Specifically, objects that neither have the property nor are far from having the property are discarded. The exact formulation of these promise problems refers to a distance measure defined on the set of all relevant objects (i.e., this distance measure coupled with a distance parameter determine the set of objects that are far from the property). Thus, the choice of natural distance measures is crucial to the study of property testing. Secondly, we note that the requirement that the algorithms operate in sub-linear time (i.e., without reading their entire input) calls for a specification of the type of queries that these algorithms can make to their input. Thus, the choice of natural query types is also crucial to the study of property testing. These two general considerations will become concrete once we delve into the actual subject matter (i.e., testing graph properties).
Why Graphs?Let us start with an empirical observation, taken from Shimon Even's book Graph Algorithms [25] (published in 1979):
142Graph theory has long become recognized as one of the more useful mathematical subjects for the computer science student to master. The approach which is natural in computer science is the algorithmic one; our interest is not so much in existence proofs or enumeration techniques, as it is in finding efficient algorithms for solving relevant problems, or alternatively showing evidence that no such algorithms exist. Although algorithmic graph theory was started by Euler, if not earlier, its development in the last ten years has been dramatic and revolutionary.Meditating on these facts, one may ask what is the source of this ubiquitous use of graphs in computer science. The most common answer is that graphs arise naturally as a model (or an abstraction) of numerous natural and artificial objects. Another answer is that graphs help visualize binary relations over finite sets. These two different answers correspond to two types of models...