L. Babai, L. Kucera, Canonical Labellings of Graphs in Linear Average Time, IEEE Symp. on Foundations of Computer Science (1980), 39-46.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....two regular graphs of the same valence and order. Nevertheless, due to its ef ciency and simplicity, color re nement is successfully used as the basis of almost all practical implementations of isomorphism tests. This can be explained by results of Babai, Erd os, Selkow [3] and Babai, Ku ccera [4] showing that it succeeds almost always . A more powerful variant of the algorithm colors k tuples according to their isomorphism type and iteratively re nes this coloring according to the colors occurring among the neighbors of a tuple. This method is called the k dimensional Weisfeiler Leman ....

L. Babai and L. Kucera. Canonical labelling of graphs in linear average time. In Proceedings of the 20th Annual IEEE Symposium on Foundations of Computer Science, pages 39-46, 1979.

....evidence for the expressive power of C 2 , which far exceeds that of L 2 . Immerman and Lander [6] show that the C 2 theory of a finite graph (which is actually axiomatized a single sentence of C 2 ) exactly characterises the stable colouring of that graph. By results of Babai and Kucera [1] it follows that almost all finite graphs are characterized up to isomorphism by their C 2 theory almost all in the sense of asymptotic probabilities: the proportion of graphs with vertices 0; n Gamma 1 having this property tends to 1 as n goes to infinity. For more information on ....

L. Babai, L. Ku cera, Canonical labellings of graphs in linear average time, IEEE Symp. on Foundations of Computer Science (1980), pp. 39-46. 15

....2 goes far beyond that of L 2 also for finite structures. Most notably, the stable colouring of graphs is C 2 definable [10] Natural derived classes of graphs that are definable in C 2 are the classes of colour class size l for each natural number l 1. By results of Babai and Kucera, [2], the class of colour size 1 graphs is dense in the class of all finite graphs, which further implies that almost all finite graphs are characterized even up to isomorphism by their C 2 theory. In this condensed version we explicitly treat the case with counting quantifiers, that also is the ....

L. Babai, L. Kucera, Canonical Labellings of Graphs in Linear Average Time, IEEE Symp. on Foundations of Computer Science (1980), 39-46.

....fact that these are at least as hard as GRAPH ISOMORPHISM (ISO) the problem of testing whether two graphs are isomorphic. In practice, ISO is not a hard problem (e.g. see [McK] Indeed, on average over all graphs, and even over regular graphs, isomorphism is known to be testable in linear time [BK], Ku] Furthermore, there is strong evidence that ISO is not NPcomplete, else the polynomial time hierarchy would collapse to Sigma p 2 = Pi p 2 = AM ( GMW] Nevertheless, ISO has stubbornly resisted attempts to place it in polynomial time. At present the best algorithm for general graphs ....

L. Babai and L. Kucera, Canonical labelling of graphs in linear average time, Proc. 20th IEEE FOCS, 1979, 49--46.

....and vice versa. In [14] they used the fact that whp the r = d3 log 2 ne largest degrees are unique and that furthermore, the remaining vertices have distinct adjacency relationships with these r vertices. This particular result was strengthened by Karp [109] Lipton [137] and Babai and Kucera [13]. In particular, Babai and Kucera describe an algorithm which runs in O(n 2 ) expected time on G n;1=2 . The algorithm describes a way of canonically labelling all but an extremely small proportion of the graphs on vertex set [n] At a given stage the algorithm will have produced an ordered ....

L.Babai and L.Kucera, Canonical labelling of graphs in linear average time, Proceedings of the 20th Annual IEEE Symposium on the Foundations of Computer Science (1979) 39-46.

....but is not known to be polynomial. Nevertheless, graph isomorphism is rarely difficult in practice, as has been profoundly demonstrated by the efficient nauty system [ McKay, 1990 ] Furthermore, it has been shown that, on average, graph isomorphism is in linear time using even naive methods [ Babai and Kucera, 1979 ] The second catch is that even after detection is complete, computing the full symmetry breaking predicate appears to be intractable. However, there is generally no reason to generate the full symmetry breaking predicate. We can generate a partial symmetry breaking predicate without affecting ....

L. Babai and L. Kucera. Canonical labelling of graphs in linear average time. In Proceedings of the Twentieth IEEE Conference on Foundations of Computer Science, pages 46--49, 1979.

....show that Omega Gamma n) variables are needed for first order logic with counting to distinguish a sequence of pairs of graphs G n and H n . These graphs have O(n) vertices each, have color class size 4, and admit a linear time canonical labeling algorithm. This contrasts sharply with results in [10, 27] where it is shown that two variables suffice to identify all trees and almost all graphs, and that three variables suffice to identify all graphs of color class size 3. Research supported by NSF grant CCR 8709818. y Research supported by NSF grant CCR 8805978 and Pennsylvania State ....

....dimensional Weisfeiler Lehman method (k dim W L) In the late seventies and early eighties, this method was developed by many researchers, including Faradzev, Zemlyachenko, Babai, and Mathon. With k = 1, this method gives a linear time graph isomorphism algorithm that works for almost all graphs [10]. Furthermore, the fastest known general graph isomorphism algorithms make use of this method with k = O( p n) 11] It had been conjectured that this method would provide a polynomial time graph isomorphism test at least for graphs of bounded valence. Valence is a synonym for degree. Our ....

[Article contains additional citation context not shown here]

L'aszl'o Babai and Ludek Kucera, "Canonical Labelling of Graphs in Linear Average Time," 20th IEEE Symp. on Foundations of Computer Science (1979), 39-46.

....We next define the languages L k (resp. C k ) consisting of the formulas of first order logic in which only k variables occur (resp. L k plus counting quantifiers ) We then study which sets of graphs are characterized by certain L k s and C k s. It follows by a result of Babai and Kucera [4] that the language C 2 characterizes almost all graphs. We also show that C 2 characterizes all trees. In Section 1.9 we give a simple O[n k log n] step algorithm to test if two graphs G and H on 1 Computer Science Dept. University of Massachusetts, Amherst, MA 01003. Research supported by NSF ....

....1. Describing Graphs: a First Order Approach to Graph Canonization 19 1. var(T k ) 8 : 2 if k = 1 3 if 2 k 3 k if k 3 2. var(S k ) 8 : 2 if 1 k 2 3 if 3 k 6 dk=2e if k 6 Babai and Kucera have proved the following result about stable colorings of random graphs: Fact 1.8. 4 [4] There exists a constant ff 1 such that if G is chosen randomly from the set of all labeled graphs on n vertices then ProbfG has two vertices of the same stable colorg ff n : Corollary 1.8.5 Almost all finite graphs are characterized by C 2 . It is easy to see that Fact 1.8.4 fails for ....

Laszlo Babai and Ludek Kucera (1980), Canonical Labelling of Graphs in Linear Average Time," 20th IEEE Symp. on Foundations of Computer Science, 39-46.

No context found.

L. Babai, L. Kucera, Canonical Labellings of Graphs in Linear Average Time, IEEE Symp. on Foundations of Computer Science (1980), 39-46.

No context found.

L. Babai and L. Kucera. Canonical labelling of graphs in linear average time. In Proceedings of the Twentieth IEEE Conference on Foundations of Computer Science, pages 44-49, 1979.

No context found.

L. Babai and L. Kucera. Canonical labelling of graphs in linear average time. In Proceedings of the Twentieth IEEE Conference on Foundations of Computer Science, pages 44-49, 1979.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC