Graphs (MTAT , 4 AP / 6 ECTS) Lectures: Fri 12-14, hall 405 Exercises: Mon 14-16, hall 315 või N 12-14, aud. 405

Transcription

1 Graphs (MTAT , 4 AP / 6 ECTS) Lectures: Fri 12-14, hall 405 Exercises: Mon 14-16, hall 315 või N 12-14, aud. 405 homepage: (contains slides) For grade: three tests (during or after semester)

2 Ã Ò Ö ½ Ø ÖÜ Ø Û Ð Ø Ø ÛÓÙÐ ÖÓ Ú ÖÝ Ö Ü¹ Ó ÓÒÒ Ö ØÙÖÒ ØÓ Ø ÒÒ Ò ØÐÝ

3 Ö ÔÖ ÒØ Ø ÓÒ Ó Ø Ã Ò Ö Ö ÔÖÓ ¹ Ö Ô Ð Ð Ñ ÙÐ ÖËÙ Û Ð ÑÔÓ Ð

4 Ø ÒÝ ÔÓ ÒØ Ò Ò Ð ØÖ Ð ÖÙ Ø Ø ÙÑ Ó ÙÖ¹ ÓÛ Ò ØÓÛ Ö Ø Ø ÔÓ ÒØ ÕÙ Ð ØÓ Ø ÙÑ Ó Ö ÒØ Ì Ö Ø ÙÑ Ó Ø Ð ØÖ Ð ÔÓØ ÒØ Ð Ö Ò ÖÙ Ø ÑÙ Ø Þ ÖÓº ÖÓÙÒ Ã Ö Ó Ð Û ½ µ ÙÖÖ ÒØ ÓÛ ÒÛ Ý ÖÓÑ Ø Ø ÔÓ Òغ

5 ÖØ ÙÖ ÝÐ Ý ½ µàóû Ñ ÒÝ ÓÑ Ö ÓÐ Ò Ë Ö Ò À ¾Ò ¾ Ö Ø Ö

6 ÖÑÝ ½ µàóû Ø Ø ÔÓ Ð ØÓ ØÖ Ò ÔÓÖØ ÍË ÖÓÑ ËÓÚ Ø ÍÒ ÓÒ ØÓ Ø ÖÒ ÙÖÓÔ ÙÔÔÐ Í Ò Ø Ö ÐÛ Ý Ý Ø Ñ ÙÔ ØÓ ½ ¼¼¼ ÓÖ ² ÙÐ Ö ÓÒ Ø Ñ º ØÓÒ

7 (Undirected) graph is a Î triple Eµ, where Î is the set of vertices (also denote Î µ); is the set of edges (also denote µ). E P Î µ is the incidency mapping. For all ¾, E µ must have 1 or 2 elements. In this course we only consider finite graphs.

8 ½ ¾ Ú ½ Ú ½ Ú ¾ ¾ Ú Example: Let Î Ú ½ Ú ¾ Ú Ú, ½ ¾ and E µ Ú ½ Ú ¾ Ú ¾ Ú Ú ¾ Ú Ú Ú Ú Ú Ú ¾ A drawing may illustrate a graph. But a graph itself is still the triple Î Eµ.

9 ½ Ú ½ Ú ¾ µ ¾ Ú ¾ Ú µ Ú ¾ Ú µ Ú Ú µ Ú Ú µ Ú ¾ Ú ¾ µ Ú ½ Ú ½ Ú ¾ ¾ Ú Directed graph consists of the sets of vertices Î and edges, and the incidency mapping E Î Î. The edges of a directed graph may also be called arcs. Example: let Î Ú ½ Ú ¾ Ú Ú, ½ ¾ and E µ

10 Let Î Eµ be a graph. Notations: If Ú ¾ E µ, then Ú and are incident. If there exists, such that E µ Ú ½ Ú ¾, then Ú ½ and ¾ are adjacent. Ú If Ú ½ Ú ¾, then Ú ½ and Ú ¾ are the endpoints of E µ. Ú Denote also ½ Ú¾. Let Î Eµ be a directed graph. Notations: If E µ Ú ½ Ú ¾ µ, then Ú ½ the end vertex of. is the start vertex and Ú ¾

11 Ú ½ ½ Ú ¾ ¾ ¾ is a multiple edge, if there exists ¼ ¾ Ò, such E µ E that µ. ¾ is a loop, if E µ ½. ¼ loop Ú multiple edges A simple graph is a graph without loops and multiple edges. Ú

12 ½ Ú ½ Ú ¾ µ ¾ Ú ¾ Ú µ Ú ¾ Ú µ Ú Ú µ In a directed simple graph, we may assume Î Î. To an edge ¾, where E µ Ú ½ Ú ¾ µ, corresponds Ú ½ Ú ¾ µ ¾ Î Î. Example: let Î Ú ½ Ú ¾ Ú Ú, ½ ¾ and E µ We may assume that Ú ½ Ú ¾ µ Ú ¾ Ú µ Ú ¾ Ú µ Ú Ú µ.

13 ½ ¾ In an undirected simple graph we may also assume Î. Î To an ¾ edge, E µ Ú where Ú ¾ ½, corresponds ½ Ú ¾ µ Ú ¾ Ú ½ µ Î Î. Ú Example: let Î Ú ½ Ú ¾ Ú Ú, ½ ¾ and E µ Ú ½ Ú ¾ Ú ¾ Ú Ú ¾ Ú Ú Ú We may assume that Ú ½ Ú ¾ µ Ú ¾ Ú ½ µ Ú ¾ Ú µ Ú Ú ¾ µ Ú ¾ Ú µ Ú Ú ¾ µ Ú Ú µ Ú Ú µ.

14 Úµ ¾ Ú ¾ E µ ¾ E µ Ú Ú ½ Ú Ú ¾ Ú The degree of a vertex Ú in the graph Î Eµ is the number of edges incident to it (the loops count twice). Denote Úµ. 1 5 ½ 3 3 ¾

15 ε Let be undirected simple graph. Î Let ½ Ú Ò. The adjacency matrix of is a Ò Ò matrix Ú A, where If Ú Ú µ ¾, then ½. If Ú Ú µ ¾, then ¼. The adjacency matrix is symmetric and its main diagonal contains zeroes. Example Ú ½ Ú ½ Ú ¾ ¾ Ú

16 Theorem. An undirected simple graph contains an even number of vertices of odd degree. Proof. ε. Count the ones in the adjacency matrix of Their number is ¾. Their number is È Ú¾Î Úµ. These two quantities are equal. Hence the sum of degrees of all vertices is even. The sum of integers is even if an even number of summands are odd. Similarly, any undirected graph contains an even number of vertices of odd degree.

17 ¾ E µ Ú½ ¾ Ú Ò ¾ E µ ¾ E µ ÚÒ Proof. Let Î Ú ½ Ú Ò. Consider the sets ¾ Ú ½ ¾ E µ ¾ Ú ¾ ¾ E µ ¾ E µ Ú¾... Each ¾ belongs to exactly two of those sets. Hence È Ú¾Î Úµ ¾, which is even.

18 È Úµ ھΠIn a directed graph Î Eµ we define for a vertex Ú Ú; its indegree the number of edges ending in Úµ outdegree Úµ the number of edges starting in Ú. Similar theorem: Úµ. È Ú¾Î

19 A walk in the graph ε (from vertex Ü to vertex Ý) is a sequence È Ü Ú ¼ ½ Ú ½ ¾ Ú ¾ Ú Ú ½ Ú Ý The number is the length of the walk È. Denote È. Let Ü È Ý denote that È is a walk from Ü to Ý. A path is a walk where all vertices are distinct (only Ú ¼ Ú and may coincide). A walk is closed if Ú ¼ Ú. A closed path is a cycle. A graph is connected if there is a walk between each two of its vertices. The distance Ù Úµ between vertices Ù Ú ¾ Î is the length of the shortest walk connecting them.

20 Ú Ú ½ Ú Ú ¾ Ú Ú Examples: Ú Walk: ½ Ú ¾ Ú Ú Ú ¾ Ú Ú Path: ½ Ú ¾ Ú Ú Closed Ú walk: ½ Ú ¾ Ú Ú ½ Ú Ú Ú ½ Ú Cycle: ½ Ú ¾ Ú Ú Ú ½ ½ Ú µ ¾, Ú ½ Ú ¾ µ ½, Ú ½ Ú ½ µ ¼. Ú

21 Theorem. If the degree of each vertex of a graph is at least ¾, then this graph contains a cycle. Proof. Loops and multiple edges are cycles. Assume ε is simple. Let Ú ½ ¾ Î. Exists Ú ¾ ¾ Î, such that Ú ½ Ú ¾. Exists Ú ¾ Î, such that Ú ½ Ú ¾ Ú. This path is simple. Let Ú ½ Ú ¾ Ú be a simple path. There exists Ú ½ ¾ Î, such that Ú ½ Ú ½ and Ú Ú ½. If Ú ½ Ú for some ¾ ½ ¾, then we have a cycle. Otherwise we have a longer simple path Ú ½ Ú ¾ Ú Ú ½. The length of a simple path is bounded by Î.

22 Theorem. A simple graph, where the degree of each vertex is at least ¾, has a cycle of length at least ½. Proof. Let Ü ¼ Ü ½ Ü Ð be an open path of maximal length in this graph. Ü ¼ Ü ½ Ü ¾ Ü All neighbours of Ü ¼ are located in this path. Let Ü be the neighbour of Ü ¼ with maximal index. Then. Ü ¼ Ü ½ Ü Ü ¼ is a cycle of length Ü Ü Ð ½ ½.

23 Ú½ Ú Ú¾ Ú Ú½ Ú¾ Ú Ú½ Ú¾ Ú The subgraph ε of a graph is a graph ¼ µ, where ¼ Î ¼ Î, ¼ and for all ¾ holds ¼ Î ¼. Denote ¼. E µ Î ¼ A µ subgraph is induced (by the Î set ), if the set ¼ Î ¼ is as large as possible, ¼ Î i.e. µ ¾ holds for ¼ E µ ¼ ¼ all ¾. Example: The connected components of a graph are its maximal connected subgraphs.

24 More notions: An edge of a graph is bridge if its removal increases the number of connected components. A vertex of a graph is cut vertex if its removal (together with its incident edges) increases the number of connected components.

25 A homomorphism from ½ Î ½ ½ µ to ¾ Î ¾ ¾ µ is a mapping Î ½ Î ¾, such that Ü Ý ¾ Î ½ are adjacent iff ܵ ݵ ¾ Î ¾ are adjacent. Example: Homomorphism is monomorphism, if it s injective. Homomorphism is isomorphism, if it s bijective. Graphs ½ and ¾ are isomorphic (denote ½ ¾ ), if there exists an isomorphism between them.

26 Usually we identify isomorphic graphs with each other. For example, let ½ Î ½ ½ µ and ¾ Î ¾ ¾ µ be ½ ¾ ½ Î ¾ ¼ ½ Î ½ Î ¼ ½ ½ µ graphs. We say that is a subgraph of, if there exist and ¾, such that ¼ ¼ ½ ¾. µ ½ Î ½ is an induced subgraph of ¾ iff there exists a monomorphism from ½ to ¾.

27 Graph isomorphism problem: Find whether two given simple graphs are isomorphic. The graphs are represented by e.g. their adjacency matrices. No polynomial-time algorithm for that task is known. In the following, let us see some polynomial-time verifiable necessary conditions for the isomorphism of two graphs.

28 Isomorphic graphs have equally many vertices and equally many edges. The degree sequence of the Ú graph Ú Ò µ ½ is the non-decreasing sequence of Ú values µ Ú Ò ½ µ. For example, the degree sequence of the graph is ½ ½ ¾ ¾ µ. Isomorphic graphs have the same degree sequences.

29 Ú ½ Ú ½ Ú ¾ ¾ Ú Ü ½ ¼ ¼ ½ Let be a square matrix and a unit matrix of the same dimensions. The characteristic polynomial (of the variable Ü) of the matrix is Ø Ü µ The characteristic polynomial of a simple graph is the characteristic polynomial of its adjacency matrix. E.g.: ¼ ½ Ü ½ ½ Ø ¼ ½ Ü ½ ¼ ½ ½ Ü Ü Ü ¾ ¾Ü ½

30 Two simple graphs are isomorphic iff the first adjacency matrix can be transformed to the second by permuting its rows and columns in the same way. Such permutation does not change the elements on the main diagonal. Such permutation contains an even number of swaps of rows or columns, hence it does not change the determinant. Isomorphic graphs have the same characteristic polynomial.

31 ½µ Ò Ò ¾ An edgeless graph is a graph without edges. A null graph Ò of vertices is denoted Ç by Ò Æ or Ò. A complete graph is a simple graph with an edge between each pair of vertices. A complete graph of Ò vertices is denoted by à Ò. Proposition. Graph Ã Ò has edges. Graph ε is bipartite, if Î can be partitioned to two sets Î ½ and Î ¾ (i.e. Î ½ Î ¾ Î and Î ½ Î ¾ ), such that the endpoints of any edge belong in different parts. (More generally: a graph is -partite if its vertices can be partitioned into parts such that all edges are between different parts.)

32 A bipartite simple graph with parts of vertices Î ½ and Î ¾ is complete bipartite if there is an edge between each Ú ½ ¾ Î ½ and Ú ¾ ¾ Î ¾. Let Ã Ñ Ò denote the complete bipartite graph with Î ½ Ñ and Î ¾ Ò. Proposition. Ã Ñ Ò has ÑÒ edges. Theorem. A graph is bipartite all its cycles are of even length. Proof µ. A cycle goes a number of times from the first part to the second and the same number of times from the second part to the first. Proof. ε Assume is connected. Otherwise consider each connected component separately. In the following we ll colour the vertices of black and white.

33 Pick a vertex Ú ¼ ¾ Î and colour it white. Let Ù be a coloured vertex that has uncoloured neighbours. Let Ú be one of such neighbours. Colour it with the opposite colour to Ù. Remember that the colour of Ù was used to choose the colour of Ú. Denote it Ú Ù. Repeat the previous paragraph, until one of the following happens there appear adjacent vertices Ü and Ý of the same colour; we run out of vertices to colour.

34 Ú ¼ Ú¼ If such vertices Ü and Ý appear then Ü Ý we have a cycle of odd length Ü Ú ¼ Ý Ü. Otherwise (we run out of vertices) the black vertices form one part and white vertices the other part of vertices of the bipartite graph.

35 Some history: 1735/6: Euler considered the Königsberg bridges question. 1750: Planar graphs: Euler s polyhedron formula. Proven 1794 by Legendre. Knight s tour problem (on chessboard): known hundreds of years. Solutions by Euler (1759), Vander- monde (1771). 1845: Kirchhoff s circuit laws. 1852: Earliest known mention of the four-color problem, in a letter from De Morgan to Hamilton. 1857: Icosian game (Hamilton). 1874: counting trees (Cayley). 1889: counting labeled trees.

36 1930: Necessary and sufficient criterions for planarity (Kuratowsky). 1976: Proof of the four-color theorem (Appel and Haken). Möbius s puzzle to his students ( ½ ¼): There once was a king with five sons. In his will he stated that after his death the sons should divide the kingdom into five regions, such that the boundary of each region should have a frontier line in common with each of the other four regions. Can the terms of the will be satisfied?

37 Tartu Akadeemiline Meeskoor võtab vastu uusi lauljaid. Proovid toimuvad T 18:30 ja N 18:30 endises EPA klubis (Veski 6, Kassitoomel). Uute lauljate vastuvõtt algab 7. septembrist. Eriti oodatud on tenorid.