UofL Discrete Math Workshop 2008 1 Online Degree Ramsey Theory posed by Illinois REGS (2007) problem 1 presented by Lesley Wiglesworth LATEX byadamjobson For a family of graphs F closed under subgraphs, and a graph G from F, define a game (G, F) played by Builder and Painter. The game begins with Builder presenting an edge and Painter coloring the edge red or blue. On each turn thereafter, Builder adds another edge to the red-blue colored graph present (adding vertices if needed), so that the new graph, viewed uncolored, is a member of F. Painter must then color the new edge red or blue. Builder wins if a monochromatic version of G appears as a subgraph. Grytczuk, HaÃluszczak, and Kierstead [1] showed that for any k-colorable graph G, builder has a winning strategy for the game in which F is the set of all k-colorable graphs. Some futher questions related to this were presented at the Illinois REGS last year as problem number 28. In the course of studying it, they introduced the online degree-ramsey number of a graph, odr(g) =min{k Builder wins (G, S k )} where S k is the family of all graphs with maximum degree k. They characterized when odr(g) 3 and showed that odr(k 1,m )=m, 4 odr(c n ) 5 and odr(c n )=4forn even, 337 n 514, and n 689. Question: odr(c 5 )=5? Question: When is odr(g) 4? Question: Is it true that odr(g) f( (G)) for some polynomial f? The Illinois group has not yet published their results, but slides of presentations by both Doug West and Kevin Milans are available online at http://www.math.uiuc.edu/~west/pubs/olramt.pdf and http://www.math.uiuc.edu/~milans/publications/odr/odr-slides.pdf respectively. [1] J.A. Grytczuk, M. HaÃluszczak, H.A. Kierstead; On-line Ramsey Theory. Electronic Journal of Combinatorics 11 (2004), no. 1, #R57.
UofL Discrete Math Workshop 2008 2 problem 2 Hadwiger Conjecture for Inflations of the Petersen Graph posed by Bjarne Toft (2008) presented by André Kézdy Problem: Determine whether all inflations of the Petersen graph satisfy Hadwiger s conjecture. Toft posed this problem informally at the end of his plenary talk at the 2008 Cumberland Conference in Nashville, TN 2008. He mentioned it because he believes that Hadwiger s conjecture is false and that insufficient effort has been given to disproving the conjecture. Inflations of the Petersen graph seem to him like a good place to search for counterexamples, especially in light of the counterexamples to Hajós conjecture. Notes: 1. To inflate a graph is to replace each vertex by a complete graph (for a formal definition of this, see for example the otherwise unrelated article [2]). 2. One version of Hadwiger s conjecture states that every graph with chromatic number t contains a minor of K t. The conjecture is not hard to prove for t =1, 2, 3, 4andis equivalent to the Four Color Theorem when t = 5. Robertson, Seymour and Thomas [4] have shown that the t =6isalsotrue. Fort 7 the conjecture is still open. 3. Hajós (1950) conjectured that every graph with chromatic number t contains a subdivision of K t. Catlin [1] gave counterexamples to Hajós conjecture. For example, inflating each vertex of C 5 with a triangle produces a graph with chromatic number 8 but no subdivision of K 8. [1] Catlin, P. A., Hajós graph-coloring conjecture: variations and counterexamples. J. Comb. Th. (B) 26 (1979), 268-274. [2] Prowse, Anton; Woodall, Douglas R. Choosability of powers of circuits. Graphs Combin. 19 (2003), no. 1, 137 144. [3] Jensen, Tommy R.; Toft, Bjarne Graph coloring problems. Wiley-Interscience Series in Discrete Mathematics and Optimization. A Wiley-Interscience Publication. John Wiley & Sons, Inc.,New York, 1995. [4] Robertson, Neil; Seymour, Paul; Thomas, Robin, Hadwiger s conjecture for K 6 -free graphs. Combinatorica 13 (1993), no. 3, 279 361. [5] Toft, Bjarne A survey of Hadwiger s conjecture. Surveys in graph theory (San Francisco, CA, 1995). Congr. Numer. 115 (1996), 249 283.
UofL Discrete Math Workshop 2008 3 λ-labellings of Graphs posed by Jerrold Griggs and Roger K. Yeh (1992) posed by Georges and Mauros (2003) problem 3 presented by Adam Jobson An L n1,n 2,...,n k -labelling of a graph G is a function f : V (G) Z + under which vertices at distance exactly i get mapped to numbers at least n i apart. This comes out of the problem of trying to allocate frequencies to radio stations so that stations which are close to each other physically have frequencies far enough apart to avoid interference. It is easy to talk about L p -labellings and L 1,1,...,1 -labellings (and by extension, L p,p,...,p - labellings). Thus the first interesting case is the L 2,1 -labelling. The span of a labelling is the maximum number assigned to a vertex. The smallest span of all L n1,n 2,...,n k -labellings of G is denoted by λ n1,n 2,...,n k (G). Notice that an L 1 -labelling of G can be thought of as a proper coloring of G and that λ 1 (G) =χ(g) 1. It is easy to show that λ k (G) =k (χ(g) 1) and that λ 1,1 (G) =χ(g 2 ) 1. Recall that χ(g) (G)+1 and by Brooks Theorem, for G connected, χ(g) = (G) + 1 G is a complete graph or an odd cycle, where (G) is the maximum degree of a vertex in G. Thus for the L 2,1 case, Griggs and Yeh in 1992 [4] conjectured the following: Conjecture : For all G with maximum degree, wehaveλ 2,1 (G) 2. The current best result is by Gonçalves [2] which states that λ 2,1 (G) 2 + 2. I believe Griggs mentioned in his talk at the 2008 Cumberland Conference that desired result has been shown for very large n (say, 10000), but I can find no citation of this result, even in his recent survey paper [3]. If the result were true, then for = 3 we would have λ 2,1 (G) 9. In [1], Georges and Mauro investigated the situation for 3-regular graphs. They obtained the desired result for certain generalized Petersen graphs and made the following conjecture: Conjecture: For G connected, 3-regular, and not the Petersen graph, λ 2,1 (G) 7. [1] J. P. Georges, D. W. Mauro: On generalized Petersen graphs labelled with a condition at distance two, Discrete Math. 259 (2002), 311-318. [2] D. Gonçalves: On the L(p, 1)-labellings of graphs, Discrete Math and Theor. Comp. Sci. AE (2005), 81-86. [3] J.R. Griggs, D. Král: Graph labellings with variable weights, a survey, Disc. Appl. Math., to appear. [Preprint obtained from http://www.math.sc.edu/ griggs/] [4] J. R. Griggs, R. K. Yeh: Labeling graphs with a condition at distance two. Siam J. Discrete Math. 5 (1992), 586-595.
UofL Discrete Math Workshop 2008 4 [5] R. K. Yeh: A survey on labeling graphs with a condition on distance two. Discrete Math. 306 (2006), 1217-1231.
UofL Discrete Math Workshop 2008 5 problem 4 Matchings extend to Hamiltonian cycles in hypercubes posed by Ruskey and Savage (1993) presented by Tim Brauch A matching is a set of edges such that no two edges share a vertex. A hypercube or n-cube is an n-dimensional analogue of the square or cube with 2 n vertices (usually placed placed in R n with each coordinate 0 or 1) with two vertices adjacent if they differ in exactly one coordinate. A Hamiltonian cycle is a cycle that visits each vertex exactly once. Question: Does every matching of the hypercube extend to a Hamiltonian cycle? Or perhaps even easier, the following weaker version can be explored. Question: Does every matching of the 5-cube extend to a Hamiltonian cycle? Ruskey and Savage first posed this problem in [3] in 1993. It has been shown to be true for the n-cube where n 4. A similar, but weaker result was proved by Fink (see [1]) who showed that every perfect matching of a hypercube extends to a Hamiltonian cycle. The perfect matching version was proposed by Kreweras in [2]. What makes this problem nice is that the symmetries of the hypercube allow for minimizing the number of cases to consider. Also, any matching that can be extended to a perfect matching need not be considered. The first step of extending such a matching would be to extend it to a perfect matching then apply the result of Fink. For example, in the 2-cube, there is only one non-perfect matching (up to ismorphism). Simply select any edge. In the 2-cube the only non-perfect matching can be extended to a perfect matching easily by picking the edge opposite the selected edge. The 3-cube becomes a little more interesting. A maximal, non-perfect matching must include three edges. Any matching with two edges can be extended to a perfect matching (this is not difficult to show, look at faces). A perfect matching contains four edges and there are only 2, up to isomorphism (this can be shown by counting the edges included in the matching on each face). To show it for n =3itsuffices to show it is true for matchings of size 3. [1] J. Fink, Perfect matchings extend to Hamilton cycles in hypercubes, J. Comb. Theory, Ser. B 97 (2007), no. 6, 1074 1076. [2] G. Kreweras, Matchings and Hamiltonian cycles on hypercubes, Bull. Inst. Combin. Appl. 16 (1996) 87 91. [3] F. Ruskey and C. D. Savage, Hamiltonian Cycles which Extend Transposition Matchings in Cayley Graphs of S n, SIAM Journal on Discrete Mathematics 6 (1993), no. 1, 152 166.
UofL Discrete Math Workshop 2008 6 Distinct Sums Modulo n posed by Hunter Snevily (1999) problem 5 presented by Jake Wildstrom It was shown by Marshall Hall [3] that, given an abelian group of order n and (not necessarily distinct) elements a 1,...,a n thereof, there is a permutation b 1,...,b n of elements of the group such that each a i + b i is distinct if and only if P n i=1 a i =0. Itappearstobe the case, however, that if the set {b i } and sequence {a i } have fewer than n elements, such a permutation always exists. This motivated Hunter Snevily to present a conjecture [5] in the particular case of G = Z n :givenpositiven and k with k<nand sequence of integers a 1,a 2,...,a n, there is a permutation π S k such that all of a π(i) + i are distinct modulo n. Snevily s conjecture has been proven for prime n by Alon using the Combinatorial Nullstellensatz [1]; a variant approach also using the Combinatorial Nullstellensatz by Kézdy and Snevily[4] proved the conjecture whenever k n+1 2. A more general form of the question, also investigated by Kézdy and Snevily, is that of how many permutations yield distinct sums. For a a vector of k elements of Z n,wedenote by Φ(n, a) the number of permutations π S k such that a π(i) + i are all distinct. We now ask how many permutations there are for the most restrictive choice of a: thatis,whatis N(n, k) =min a Z k n Φ(n, a). Snevily s conjecture would, with this definition, reduce to the statement that N(n, k) > 0whenk<n. Question: What properties does the function N(n, k) possess? Kézdy and Snevily present several conjectures about the behavior of N(n, k): The function is monotone in n: N(n, k) N(n +1,k). The function is monotone in k up to n: N(n, k) N(n, k +1)whenk<n 1. If n is significantly larger than k, N(n, k) =d k 2 e!b k 2c!. This is known to be an upper bound on N(n, k), since Φ(n, 0 d k 2 e 1 b k 2 c ) achieves this value. Note that the monotonicity conjectures, together with the trivial statement N(n, 1) = 1 for all n, imply Snevily s conjecture. Most recent work along the lines of this problem has considered restrictions on the choice of a: papers by Dasgupta et. al [2] and Sun [6] are both more general than the traditional problem, expanding into a larger class of underlying groups than Z n, and more restricted, as they demand that terms in a be distinct and of odd order. Current unpublished results for this problem include simple results when n is very large relative to k, and one result for the special case k = n 1: If n>k(k 1), then N(n, k) N(n+1,k). If n>k 2 1, then N(n, k) =N(n+1,k). If n>k(k + 1), then N(n, k) N(n, k +1).
UofL Discrete Math Workshop 2008 7 N(n, n 2) N(n, n 1). These results can probably be improved significantly. In particular, it is hoped that the last result can generalize to cases other than k = n 1. [1] N. Alon, Additive Latin transversals. Israel J. Math. 117 (2000), 125 130. [2] S. Dasgupta; G. Károlyi; O. Serra; B. Szegedy, Transversals of additive Latin squares, Israel J. Math. 126 (2001), 17 28. [3] M. Hall, A combinatorial problem on abelian groups, Proc.Amer.Math.Soc.3 (1952), 584 587. [4] A. Kézdy and H. Snevily, Distinct Sums Modulo n and Tree Embeddings, Combin. Probab. Comput. 11 (2002), 35 42. [5] H. Snevily, The Cayley Addition Table of Z n, Amer. Math Monthly 106 (1999), 584 585. [6] Z. Sun, On Snevily s conjecture and restricted sumsets, J. Combin. Theory, Ser. A 103 (2003), 291 304.