Proofs: Kuratowski's theorem
These are the human proofs of Kuratowski's theorem, at full length.
Proof of Kuratowski's theorem full
This proof shows: statement.
We prove that a finite graph is planar if and only if it contains no subdivision of or . Graphs are simple: loops and parallel edges do not change planarity. We write for a subdivision of , and say that contains if some subgraph of is a . A model of in is a family of disjoint vertex sets (), the branch sets, each inducing a connected subgraph, with an edge of between and for each edge of ; has an minor exactly when it has a model of . The proof follows the route of Diestel, Sections 4.4 and 3.2, with the minimal counterexample of Thomassen.
The easy direction
A subgraph of a planar graph is planar: delete points and arcs from a drawing. If contains or and were planar, a drawing of the subdivision would give a drawing of or , by reading each subdivided path as one arc. But has edges, and is triangle-free with edges, so neither is planar by the edge bounds from Euler's formula. So a planar graph contains neither subdivision.
From minors to subdivisions
Lemma 1 (a center in a tree). Let be a finite tree and let labels be placed on vertices of (several labels may sit on one vertex). If no edge of leaves at least 2 labels on each side of , then there is a vertex such that each component of carries at most one label.
Proof. Each edge has a side with at most one label. Orient away from such a side (if both sides have at most one label, orient it either way). Start anywhere and follow the oriented edges. A walk along oriented edges in a tree cannot use an edge twice (it would have to go back against its orientation) and cannot close a cycle, so it never repeats a vertex and stops at a vertex with no outgoing edge. For each edge , the edge points to , so the side of , which is the component of that contains , has at most one label.
Call such an a center. For a center and each label on a vertex , the paths meet pairwise only in : two of them lie in different components of , or one is the trivial path .
Lemma 2. (a) If has a minor, then contains . (b) If has a minor, then contains or .
Proof. Let be a model of in . For each edge of fix an edge of with and . In , put the label on for each neighbour of , and let be a spanning tree of .
Suppose first that every has a center . For each edge of , let be the path , the edge , and . Two such paths meet only in common ends: inside one tree , the parts toward different labels meet only in by the remark after Lemma 1, and different trees are disjoint. So the paths with the vertices form a in .
For each vertex has 3 labels, and no edge of can leave 2 labels on each side, so Lemma 1 gives a center in every tree. This proves (a).
For , if every has a center we get . Otherwise, by Lemma 1, some tree, say with labels , has an edge such that the component of that contains carries two labels, say 2 and 3, and the component that contains carries 4 and 5. Then the sets and are the branch sets of a model with these two sides: and are connected and joined by the edge ; contains and , so it is joined to and ; is joined to and in the same way; and each of is joined to each of because the form a model. So has a minor, and contains by (a).
Since a in gives an minor (contract the subdivided paths), Lemma 2 shows: contains or if and only if has a or minor. Minors of minors are minors, so if contains neither subdivision, neither does any minor of .
A contractible edge
A graph is 3-connected if it has more than 3 vertices and no set of at most 2 vertices separates it. In a 3-connected graph , if is a set of 3 vertices that separates , then each vertex has a neighbour in every component of : otherwise the other two vertices of would separate that component from the rest.
Lemma 3 (Thomassen). Every 3-connected graph other than has an edge such that is 3-connected.
Proof. The only 3-connected graph with 4 vertices is , so has vertices and each has at least 4. Suppose that no contraction is 3-connected. Let be an edge and the vertex of that replaces and . Then has a separating set of at most 2 vertices. If , then separates too, which is impossible. So contains , and for a vertex , since alone would give the separating set of . Then separates .
Among all edges and such vertices , choose them so that has a component with as few vertices as possible. The vertex has a neighbour in . By assumption is not 3-connected either, so as above there is a vertex such that separates , and each of has a neighbour in every component of . Since and are adjacent, they lie in at most one component of together with possibly ; so there is a component that contains neither nor . The vertex has a neighbour . Since and is adjacent to , lies in . The subgraph is connected and avoids , so . And , so , which contradicts the choice of , with the edge and the vertex .
Two facts about drawings
We use two standard facts about plane graphs.
(F1) In a 2-connected plane graph, every face is bounded by a cycle of the graph (Diestel, Proposition 4.2.6). The region bounded by a cycle is homeomorphic to an open disk with the cycle as its boundary circle, by the Jordan-Schönflies theorem (Mohar and Thomassen, Section 2.1). So a new vertex placed in a face bounded by a cycle can be joined to any set of vertices of by arcs inside the face that meet only at the new vertex.
(F2) For each face of a plane graph, there is a drawing of the same graph with the same faces in which is the outer face (project the drawing onto the sphere and back from a point inside ). So for each edge there is a drawing with on the boundary of the outer face. If with , and if and are planar (or , are planar and share at most one vertex ), then is planar: draw both parts with (or ) on the outer face, put the drawing of the second part inside a face of the first next to (or at ), and delete if it is not an edge of (Diestel, Section 4.4).
The main proof
Suppose the theorem is false, and let be a non-planar graph that contains no and no , with as small as possible. Then has no isolated vertex, and every graph with a smaller value of and no , is planar. Graphs with at most 4 vertices are planar (they are subgraphs of ), so .
is 2-connected. Otherwise for proper subgraphs that share at most one vertex, with no edge between and , each with a vertex outside the other. Each has a smaller value of and no or , so it is planar, and is planar by (F2). This is a contradiction.
is 3-connected. Suppose that separates . Write with , each with a vertex outside , and no edge between and ; put the edge , if it exists, into . Let . Each component of inside has a neighbour at and one at , since is 2-connected; so contains an - path whose inner vertices lie outside . The graph has fewer vertices than and at most one edge that lacks, while has at least two edges not in ; so . If contained a or , then replacing the edge in it, if it is used, by would give one in . So and are planar, and is planar by (F2), a contradiction. So is 3-connected, and since .
The contraction. By Lemma 3, has an edge with 3-connected. Let be the vertex of that replaces and . Since contains no or , it has no or minor (by Lemma 2), so neither has its minor , so contains no or . It is smaller than , so it is planar. Fix a drawing of .
The graph is 2-connected, so by (F1) the face of that contains the point of is bounded by a cycle . The edges at run inside , so all neighbours of lie on . Let and . Both are subsets of , and , since is 3-connected. Let be the vertices of in their cyclic order on , and let be the path on from to that contains no other vertex of (indices mod ).
Case 1: for some . Erase the edges at that do not go to , and rename as ; the arcs from to divide into regions, and the region between the arcs to and is bounded by the cycle . Put inside and join it to and to each vertex of by arcs in , which is possible by (F1). The result is a drawing of , a contradiction.
Case 2: . Let . Split into three disjoint paths, each containing exactly one of ; they are connected and pairwise joined by edges of . With and , these are five branch sets of a model in : and are adjacent, and both are adjacent to . So has a minor and contains or by Lemma 2, a contradiction.
Case 3: otherwise. We find and , four different vertices that alternate around . If some is not in , it lies inside a unique path , between and . Since Case 1 fails, some lies outside , and then alternate around (also when ). If instead , then because Case 2 fails, so with ; since Case 1 fails, and are not consecutive in the cyclic order of , so each of the two arcs of between them contains a vertex of , say and , and alternate. In both cases the four arcs of between consecutive vertices of (in this cyclic order), the edges , , , and form a with sides and : reaches , , by edges, and each of , reaches by an edge and , along . This is a contradiction.
All cases are contradictory, so no counterexample exists.