Lecture 5 November 6, 2012

Similar documents
Chords in Graphs. Department of Mathematics Texas State University-San Marcos San Marcos, TX Haidong Wu

arxiv: v1 [math.co] 25 Sep 2016

Restricted cycle factors and arc-decompositions of digraphs. J. Bang-Jensen and C. J. Casselgren

On the tree cover number of a graph

GRAY CODES FAULTING MATCHINGS

Simpler Testing for Two-page Book Embedding of Partitioned Graphs

Graphs and Networks Lecture 5. PageRank. Lecturer: Daniel A. Spielman September 20, 2007

Discrete Applied Mathematics. The induced path function, monotonicity and betweenness

The Brauer Manin obstruction

arxiv: v1 [cs.dm] 27 Jun 2017 Darko Dimitrov a, Zhibin Du b, Carlos M. da Fonseca c,d

Partial cubes: structures, characterizations, and constructions

IN this paper we consider simple, finite, connected and

A Note on Arboricity of 2-edge-connected Cubic Graphs

Spanning Trees with Many Leaves in Graphs without Diamonds and Blossoms

G. Mahadevan 1 Selvam Avadayappan 2 V. G. Bhagavathi Ammal 3 T. Subramanian 4

arxiv: v1 [math.co] 10 Nov 2010

Edge-counting vectors, Fibonacci cubes, and Fibonacci triangle

Connectivity and Menger s theorems

Cubic graphs have bounded slope parameter

On oriented arc-coloring of subcubic graphs

Axiomatizing the Cyclic Interval Calculus

The Minimal Estrada Index of Trees with Two Maximum Degree Vertices

Vectors in Rn un. This definition of norm is an extension of the Pythagorean Theorem. Consider the vector u = (5, 8) in R 2

RESOLUTION OF INDECOMPOSABLE INTEGRAL FLOWS ON A SIGNED GRAPH

A generalized Alon-Boppana bound and weak Ramanujan graphs

Xihe Li, Ligong Wang and Shangyuan Zhang

Lesson 81: The Cross Product of Vectors

A generalized Alon-Boppana bound and weak Ramanujan graphs

Lecture 3. (2) Last time: 3D space. The dot product. Dan Nichols January 30, 2018

To pose an abstract computational problem on graphs that has a huge list of applications in web technologies

Conditions for Approaching the Origin without Intersecting the x-axis in the Liénard Plane

Linear Algebra and its Applications

ON THE PERFORMANCE OF LOW

Minimizing Intra-Edge Crossings in Wiring Diagrams and Public Transportation Maps

A Note on Johnson, Minkoff and Phillips Algorithm for the Prize-Collecting Steiner Tree Problem

Worst-case analysis of the LPT algorithm for single processor scheduling with time restrictions

Upper Bounds on the Spanning Ratio of Constrained Theta-Graphs

Complex Variables. For ECON 397 Macroeconometrics Steve Cunningham

Generalized Fibonacci cubes

Two-ended regular median graphs

Key words. partially ordered sets, dimension, planar maps, planar graphs, convex polytopes

Computing distance moments on graphs with transitive Djoković-Winkler s relation

Faster exact computation of rspr distance

We automate the bivariate change-of-variables technique for bivariate continuous random variables with

The Real Stabilizability Radius of the Multi-Link Inverted Pendulum

Minimal Obstructions for Partial Representations of Interval Graphs

c 2009 Society for Industrial and Applied Mathematics

Subcritical bifurcation to innitely many rotating waves. Arnd Scheel. Freie Universitat Berlin. Arnimallee Berlin, Germany

THE DIFFERENTIAL GEOMETRY OF REGULAR CURVES ON A REGULAR TIME-LIKE SURFACE

Online Stochastic Matching: New Algorithms and Bounds

Equivalence between transition systems. Modal logic and first order logic. In pictures: forth condition. Bisimilation. In pictures: back condition

Graphs and Their. Applications (6) K.M. Koh* F.M. Dong and E.G. Tay. 17 The Number of Spanning Trees

Decoder Error Probability of MRD Codes

A GAP PROGRAM FOR COMPUTING THE HOSOYA POLYNOMIAL AND WIENER INDEX OF NANO STRUCTURES

Applications of Isometric Embeddings to Chemical Graphs

Spring, 2008 CIS 610. Advanced Geometric Methods in Computer Science Jean Gallier Homework 1, Corrected Version

THE CATEGORY OF CGWH SPACES

Lecture J. 10 Counting subgraphs Kirchhoff s Matrix-Tree Theorem.

Lecture 11: June 27. Distance Sensitivity Oracles. Succinct Graph Structures and Their Applications Spring Lecturer: Merav Parter

The median function on median graphs and semilattices

The New (2+1)-Dimensional Integrable Coupling of the KdV Equation: Auto-Bäcklund Transformation and Non-Travelling Wave Profiles

Differentiation of Exponential Functions

Complexity of the Cover Polynomial

Classify by number of ports and examine the possible structures that result. Using only one-port elements, no more than two elements can be assembled.

Study on the Mathematic Model of Product Modular System Orienting the Modular Design

Decoder Error Probability of MRD Codes

Exercise 4. An optional time which is not a stopping time

Chapter 6 Momentum Transfer in an External Laminar Boundary Layer

Nonlinear parametric optimization using cylindrical algebraic decomposition

Sign-reductions, p-adic valuations, binomial coefficients modulo p k and triangular symmetries

MATH2715: Statistical Methods

CRITERIA FOR TOEPLITZ OPERATORS ON THE SPHERE. Jingbo Xia

MAT389 Fall 2016, Problem Set 6

Section 7.4: Integration of Rational Functions by Partial Fractions

Reaction-Diusion Systems with. 1-Homogeneous Non-linearity. Matthias Buger. Mathematisches Institut der Justus-Liebig-Universitat Gieen

The Heat Equation and the Li-Yau Harnack Inequality

Partial Hamming graphs and expansion procedures

Linear Strain Triangle and other types of 2D elements. By S. Ziaei Rad

Graph-Modeled Data Clustering: Fixed-Parameter Algorithms for Clique Generation

ACYCLIC WEAK CONVEX DOMINATION IN GRAPHS

All Good (Bad) Words Consisting of 5 Blocks

Palindromes and local periodicity

Gradient Projection Anti-windup Scheme on Constrained Planar LTI Systems. Justin Teo and Jonathan P. How

Existence of HCMU metrics on higher genus Riemann surfaces

A diagrammatic standardization theorem

ZAGREB INDICES AND ZAGREB COINDICES OF SOME GRAPH OPERATIONS

6.4 VECTORS AND DOT PRODUCTS

On the connectivity of the direct product of graphs

Math 144 Activity #10 Applications of Vectors

MEAN VALUE ESTIMATES OF z Ω(n) WHEN z 2.

Right Trapezoid Cover for Triangles of Perimeter Two

An Explicit Lower Bound of 5n o(n) for Boolean Circuits

On the difference between the revised Szeged index and the Wiener index

SUBJECT:ENGINEERING MATHEMATICS-I SUBJECT CODE :SMT1101 UNIT III FUNCTIONS OF SEVERAL VARIABLES. Jacobians

Packing chromatic vertex-critical graphs

AMS 212B Perturbation Methods Lecture 05 Copyright by Hongyun Wang, UCSC

EE2 Mathematics : Functions of Multiple Variables

Lecture Notes On THEORY OF COMPUTATION MODULE - 2 UNIT - 2

3.4-Miscellaneous Equations

Convex sets in lexicographic products of graphs

Transcription:

Hypercbe problems Lectre 5 Noember 6, 2012 Lectrer: Petr Gregor Scribe by: Kryštof Měkta Updated: Noember 22, 2012 1 Partial cbes A sbgraph H of G is isometric if d H (, ) = d G (, ) for eery, V (H); that is, H preseres all distances from G. Clearly, eery isometric sbgraph is an indced sbgraph. To see that the conerse implication is not tre, consider a path P 4 in the cycle C 5. Figre 1: A path P 4 in the cycle C 5 is indced bt not isometric sbgraph. A graph G is a partial cbe if it has an isometric embedding into Q n for some n. That is, G is isomorphic to an isometric sbgraph of Q n for some n. The smallest sch n (if it eists) is the isometric dimesion of G, denoted by dim I (G). Obseration 1 dim I (T ) = n 1 for eery tree T on n ertices. dim I (C 2n ) = n for eery n 2, with isometric embedding sch that eery direction is sed eactly twice, on antipodal edges. Eamples of partial cbes: the Desarges graph, the permtahedron, benzenoid graphs, linear etension graphs, hyperplane arrangement graphs, Fibonacci cbes, see Figre 2. 2 Djokoić-Winkler relation Let G = (V, E) be a bipartite graph. For E let W = { V ; d(, ) < d(, )} and W = { V ; d(, ) < d(, )}. Since G is bipartite, W W is a partition of V. Obseration 2 All ertices on eery shortest -path are in W (that is, I(, ) W ) if W. The Djokoić-Winkler 1 relation, denoted by Θ, is defined as follows. Definition 3 eθf for edges e = and f if f joins a erte in W with a erte in W. 5-1

111111 00011 00111 111110 011111 01001 01011 01010 01101 10011 00101 01110 10010 00110 10110 111100 110100 011010 110101 110111 100111 001111 000111 10001 10101 111000 011000 001010 100101 001011 11001 11010 01100 10100 110000 000011 001000 11000 11100 100000 (a) 000000 153624 (b) 100001 000001 153264 135624 315624 152364 153246 135264 315264 152346 125364 135246 315246 132564 312564 125346 123564 132546 312546 123546 (c) 123456 132456 312456 (d) Figre 2: Eamples of partial cbes: (a) the Desarges graph (also known as the generalized Petersen graph P (10, 3) or the middle leel graph of Q 5 ), (b) the permtahedron of order 4 (also known as the trncated octahedron), (c) the linear etension graph of a (now secret) poset, (d) a benzenoid graph. From Obseration 2 we obtain the following. Obseration 4 For eery bipartite graph G, the relation Θ is refleie and symmetric, if edges e, f both belong to a same shortest path in G, then e Θf. To see that the relation Θ does not hae to be transitie, consider K 2,3 on Figre 3(b). 1 The definition gien here is de to Djokoić [8]. Winkler [17] defined another relation which in bipartite graphs coincides with this definition. Hence it is commonly called Djokoić-Winkler relation. 5-2

a) W W b) W f 2 W f f 1 Θ e e Figre 3: (a) a partition of V to W and W, (b) K 2,3 with eθf 1, eθf 2, bt f 1 Θf 2. 3 Characterization of partial cbes A set S V is cone if I(, ) S for eery, S where I(, ) is the interal between and. That is, with eery two ertices S contains eery shortest path between them. Theorem 5 (Djokoić [8], Winkler [17]) Let G = (V, E) be a connected graph. following statements are eqialent. The 1. G is a partial cbe, 2. G is bipartite and W, W are cone for eery E, 3. G is bipartite and Θ is transitie (and conseqently, an eqialence). Proof 1) 2) Let f : V V (Q n ) be an isometric embedding. The sbgraph f(g) of Q n is clearly bipartite which implies that G is bipartite. If E then clearly f() f() = e i for some i. Assme withot loss of generality that f() i = 0. Since f(g) is an isometric sbgraph of Q n and distances in Q n are measred by the Hamming distance it follows that f() i = 0 for eery W. A shortest path between f( 1 ), f( 2 ) for 1, 2 W clearly contains at most one edge of each direction. Ths, it does not contain any edge in the direction i and all ertices on the shortest path hae the i-th coordinate eqal to 0. This means that the shortest path between f( 1 ) and f( 2 ) stays in f(w ) so the same is tre for its preimage - the shortest path between 1, 2 W in G. 2) 3) Let ()Θ(y), say W and y W. Symmetry of Θ implies that W y and W y. We claim that W = W y. Sppose w W \ W y. Then is on a shortest w-path and w W y. The coneity of W y and the fact that w, W y imply that W y, a contradiction. The claim implies that Θ is transitie. 5-3

3) 1) We proe this implication by constrcting an isometric embedding f of G to Q n where n is the inde of Θ, i.e. n = E / Θ. Let e i = i i be a representatie of the i-th class of Θ. The embedding f : V (G) Z n 2 is defined as follows. f() i = { 0 if Wi i 1 if W i i Note that eθe if and only if f(e) and f(e ) are of the same direction in Q n. For distinct, y V a shortest y-path has edges from distinct Θ-classes which means that f() f(y), so f is injectie. By the same reason, f maps shortest paths from G to shortest paths in Q n. Ths, f is an isometric embedding. Note that from the proof it follows that dim I (G) is the inde of Θ. Since the relation Θ can be fond and tested for transitiity in a polynomial time, we obtain the following. Corollary 6 Partial cbes can be recognized in polynomial time. In fact, they can be recognized in O(mn) time and O(n 2 ) space where n = V and m = E [1]. 4 Median graphs A set of medians for a triple of ertices, y, z is defined by Med(, y, z) = I(, y) I(y, z) I(z, ). A graph G is a median graph if Med(, y, z) = 1 for eery, y, z V (G). That is, eery triple has a niqe median, which is then denoted by med(, y, z). The class of median graphs incldes all trees, grids, hypercbes (by the following proposition). We will see that median graphs are in some sense graphs between trees and hypercbes. They occr in many at first seemingly nrelated areas sch as the stdy of soltions of 2-SAT [10], stable configrations of nonepansie networks [10], stable matchings in the well-known stable roommate problem [7], median semilattices [4]. Proposition 7 Q n is a median graph for eery n 1 with med(, y, z) = maj(, y, z) where maj is the coordinate-wise majority fnction. Proof If two ertices agree in a coordinate, say i = y i, then eery m Med(, y, z) has m i = i = y i since m belongs to a shortest y-path, so m = maj(, y, z). On the other hand, if m = maj(, y, z) then i = y i implies m i = i = y i, so m lies on a shortest path between eery pair of ertices. Eery median graph is bipartite. Indeed, for a contradiction consider a shortest odd cycle with a erte antipodal to an edge yz. Then Med(, y, z) =. Theorem 8 Eery median graph G = (V, E) is a partial cbe. 5-4

W y W z P y m P 1 m P 2 z P 3 U U Figre 4: (a) The sets U and U, (b) a shortest -path P in U and adjacent ertices from U, (c) the shortest paths between and y, and, y and. Proof For E let U = { W has a neighbor in W } and similarly U. Claim 9 If U then eery shortest -path belongs to U (that is, I(, ) U ). Let z be a neighbor of on a shortest -path P, see Figre 4(b). Clearly, d(z, y) = 2 and d(z, ) = d(, ) = d(y, ). Then m = med(z, y, ) mst be a neighbor of z lying on a shortest y-path. Therefore z U. Then we contine this argment for the net erte z on P ntil we reach. Claim 10 U indces an isometric sbgraph. Let, y U and m = med(, y, ). Let P 1 be a shortest m-path, P 2 a shortest ym-path and P 3 a shortest m-path, see Figre 4(c). By Claim 4, P 1 P 3 and P 2 P 3 are in U which means that P 1 P 2 also lies in U (i.e. d U (, y) = d G (, y)). Therefore, W and similarly W are cone and we may apply Theorem 5. Not eery partial cbe is a median graph, see Figre 5. y y z z Figre 5: Two eamples of partial cbes that are not median graphs 5 Mlder s cone epansion Let G = (V, E) be a connected graph and S V be a cone set. The epanded graph G = ep(g, S) is obtained as follows: for eery erte in S add a copy and join and by an edge, 5-5

insert an edge between new ertices wheneer is an edge of S. The set of new ertices in denoted by S (a copy of S), see Figre 6. G S S Figre 6: The epansion procedre. Since a sbgraph of a median graph indced by a cone set is clearly median, by considering all cases for triples in G it can be shown that if G is a median graph, then G is also a median graph. Interestingly, eery median graph can be obtained by this procedre 2 starting from a single erte. This shows s a tree-like strctre of median graphs. Theorem 11 (Mlder [14]) A graph G is a median graph if and only if it can be obtained from a single erte by a seqence of cone epansions. For a proof we refer to [14]. Figre 7: Eamples of cone epansions. 6 Eler-type formla Recall that ertices and edges can be seen as 0-faces and 1-faces, respectiely. Ths the following relation can been regarded as an Eler-type formla. Theorem 12 (Škrekoski [16]) Let G be a median graph, k = dim I (G), and let q i be the nmber of sbcbes of dimension i in G. Then, ( 1) i q i = 1 and k = ( 1) i iq i. 2 Originally, the cone epansion is described in a more general way. Howeer, it can be shown that this simplified ersion sffices. 5-6

Proof Let G = ep(g, S), q i be the nmber of copies of Q i in G, qi S be the nmber of copies of Q i in S. From the epansion procedre, obsere that q i = q i + qs i + qi 1 S for eery i 0 (where we define q 1 S = 0). Then, ( 1) i q i = ( 1) i q i + ( 1) i qi S + ( 1) i+1 qi S = 1 (1) i 1 }{{} = 0 since q 1 S = 0 by indction for G. Frthermore, applying indction for G and (1) for S gies s k = k + 1 = ( 1) i iq i + ( 1) i q S i = = ( 1) i iq i ( 1) i (iqi S (i + 1)qi S ) = = ( 1) i iq i ( 1) i iq S i ( 1) i iqi 1 S = i 1 ( 1) i q i. 7 Open problems Problem 1 (Eppstein [9]) Find a nonplanar cbic partial cbe other than the Desarges graph. (Here, cbic means 3-reglar.) Conjectre 13 (Brešar et al. [5]) Eery cbic partial cbe is Hamiltonian. Problem 2 Characterize Hamiltonian partial cbes (median graphs). A partial cbe G = (V, E) is Θ-gracefl if there is a bijection f : V {0,..., V 1} sch that for eery e 1, e 2 E : where f () = f() f(). e 1 Θe 2 if and only if f (e 1 ) = f (e 2 ) Conjectre 14 (Brešar, Klažar [6]) Eery partial cbe is Θ-gracefl. Remark This conjectre implies the gracefl tree conjectre. 8 Notes Median graphs were first stdied by Aann [2], Nebeský [15], and independently by Mlder [14]. There is an etensie literatre on median graphs, see [3, 13] for sreys. A good starting reference is also the book 3 of Imrich and Klažar [12]. 3 The second edition [11] is already aailable. 5-7

References [1] Arenhammer, Hagaer, Recognizing binary Hamming graphs in O(n 2 log n) time, Math. Syst. Theory 28 (1995), 387 395. [2] S. Aann, Metric ternary distribtie semi-lattices, Proc. Am. Math. Soc. 12 (1961), 407 414. [3] H.-J. Bandelt, V. Chepoi, Metric graph theory and geometry: a srey, Contemp. Math. 453 (2008), 49 86. [4] H.-J. Bandelt, J. Hedlíkoá, Median algebras, Discrete Math. 45 (1983), 1 30. [5] B. Brešar, S. Klažar, A. Lipoec, B. Mohar, Cbic inflation, mirror graphs, reglar maps, and partial cbes, Eropean J. Combin. 25 (2004), 55-64. [6] B. Brešar, S. Klažar, Θ-gracefl labelings of partial cbes, Discrete Math. 306 (2006), 1264-1271. [7] C. Cheng, A. Lin, Stable roommates matchings, mirror posets, median graphs and the local/global medan phenomenon in stable matchings, SIAM J. Discrete Math. 25 (2011), 72 94. [8] D. Djokoić, Distance presering sbgraphs of hypercbes, J. Combin. Theory Ser. B 14 (1973), 263 267. [9] D. Eppstein, Cbic partial cbes from simplicial arrangements, Electron. J. Combin. 13 (2006), #R79. [10] T. Feder, Stable networks and prodct graphs, Mem. Amer. Math. Soc. 116 No.555 (1995), 223 pp. [11] R. Hammack, W. Imrich, S. Klažar, Handbook of Prodct Graphs, Second Edition, CRC Press, 2011. [12] W. Imrich, S. Klažar, Prodct graphs, strctre and recognition, Wiley, New York, 2000. [13] S. Klažar, H. M. Mlder, Median graphs: characterization, location theory and related strctres, J. Combin. Math. Combin. Compt. 30 (1999), 103 127. [14] H. M. Mlder, The strctre of median graphs, Discrete Math. 24 (1978), 197 204. [15] L. Nebeský, Median graphs, Comment. Math. Uni. Carolin. 12 (1971), 317 325. [16] R. Škrekoski, Two relations for median graphs, Discrete Math. 226 (2001), 351-353. [17] P. Winkler, Isometric embedding in prodcts of complete graphs, Discrete Appl. Math. 7 (1984), 221 225. 5-8