Theorem (Bichromatic K6 has Monochromatic K3)

Given any coloring of the edges of a Complete Graph with two colors, there is always a monochromatic subgraph.

Proof:

Select any vertex , and suppose our two colors were red and blue. Since and there are two colors, by the Pidgeonhole Principle, either has red edges or blue edges. Suppose it has , such that are the incident vertices of those red edges.

If there is a red edge between any two of then we have . If not, then must have blue edges between them, which forms .

Therefore we have a monochromatic in either case.

"\\usepackage{tikz-cd}\n\\begin{document}\n\\begin{tikzpicture}[\n vertex/.style={circle, draw, fill=white, inner sep=2pt, font=\\small},\n red_edge/.style={red, thick},\n blue_edge/.style={blue, thick},\n highlight_red_edge/.style={red, ultra thick},\n node distance=2cm,\n every node/.style={font=\\small}\n]\n\n % Place nodes in a circle\n \\foreach \\i in {1,...,6} {\n \\node[vertex] (\\i) at ({360/6 * (\\i - 1)}:3cm) {\\i};\n }\n\n % Draw edges and color them\n % Group 1: Highlighted Red K3 (1-2-5)\n \\draw[highlight_red_edge] (1) -- (2);\n \\draw[highlight_red_edge] (2) -- (5);\n \\draw[highlight_red_edge] (5) -- (1);\n\n % Group 2: Other Red Edges\n \\draw[red_edge] (1) -- (3);\n \\draw[red_edge] (2) -- (4);\n \\draw[red_edge] (3) -- (6);\n \\draw[red_edge] (5) -- (6);\n\n % Group 3: Blue Edges (the rest)\n \\draw[blue_edge] (1) -- (4);\n \\draw[blue_edge] (1) -- (6);\n \\draw[blue_edge] (2) -- (3);\n \\draw[blue_edge] (2) -- (6);\n \\draw[blue_edge] (3) -- (4);\n \\draw[blue_edge] (3) -- (5);\n \\draw[blue_edge] (4) -- (5);\n \\draw[blue_edge] (4) -- (6);\n\n\\end{tikzpicture}\n\\end{document}"123456
source code

Ramsey’s Theorem

For any positive integers , there so that for any and any red-blue coloring of the edges of a complete graph , there is either a red or a blue .

In particular, we defined the smallest such number as the Ramsey Number, .

Proof:

We proceed by induction on . If or , then any coloring of a is red or blue. Indeed

Assume is finite for all . For we will show not only that is finite, but that

Take that

and color . Consider vertex . Then

Indeed, has at least that many edges. By the Pidgeonhole Principle (recall this is the same technique we used in proving the previous theorem), either it has

  • red edges
  • or blue edges. WLOG, let have that many red many edges to the other vertices .
graph LR;
	v((v)) o--o S[[S]];
	v o--o S;
	v o--o S;
	v o--o S;
	v o--o S;
	v o--o S;

To be more precise, + \omega(K_{p}

Thus, . This lets us apply the inductive hypothesis, such that either contains

  • red
    • In this case, we form . Where , the red Clique of .
  • blue
    • We get blue Either way, we get our desired result.

Computing Ramsey Numbers

  • by symmetry
    • any red edge gives red
    • or we can color entirely blue
    • Consider . We can have red , but we cannot have blue , so this is not a lower bound.
  • .
    • We can show that . This is due to in the proof of Ramsey’s Theorem.
    • To show it cannot be smaller, then we would need to show an invalid coloring on . Indeed, the following is a coloring of where there is no red or blue.
"\\usepackage{tikz-cd}\n\\begin{document}\n\\begin{tikzpicture}[\n vertex/.style={circle, draw, fill=black, inner sep=2pt},\n red_edge/.style={red, thick},\n blue_edge/.style={blue!80!cyan, thick},\n node distance=2cm\n]\n\n % Place 5 nodes in a circle, starting at 90 degrees (top)\n \\foreach \\i in {1,...,5} {\n \\node[vertex] (\\i) at ({90 + 360/5 * (\\i - 1)}:2cm) {};\n }\n\n % Group 1: Red Edges (The Pentagon)\n \\draw[red_edge] (1) -- (2);\n \\draw[red_edge] (2) -- (3);\n \\draw[red_edge] (3) -- (4);\n \\draw[red_edge] (4) -- (5);\n \\draw[red_edge] (5) -- (1);\n\n % Group 2: Blue Edges (The Star)\n \\draw[blue_edge] (1) -- (3);\n \\draw[blue_edge] (3) -- (5);\n \\draw[blue_edge] (5) -- (2);\n \\draw[blue_edge] (2) -- (4);\n \\draw[blue_edge] (4) -- (1);\n\n\\end{tikzpicture}\n\\end{document}"
source code
    • The upper bound is found by showing any red-blue coloring of a has either a red or blue . We know that . So, consider .
    • Pick some . If has at least red edges, then as is at least a , if any internal edge is red, we have red (include ). If not, then all edges must be blue. But then we at least have a blue , as desired.
    • If has at least blue edges, then we apply the same logic. More importantly, since , then we are guaranteed to have red or blue .
      • If red we are done.
      • If blue , add , giving us blue we are done.
      • Note that we can do the same for the previous case with .
    • Otherwise, each vertex must have blue and red edges.
    • However, this violates the Handshake Lemma. The total number of red edges is , so any coloring of the graph means there must be at least one vertex of blue and edges.
    • We see that . So it suffices to show that .
      • The lower bound requires to do some construction. In particular, we arrange vertices in a circle and d raw a red edge between two if they are separated by steps.
      • Leave the other edges blue, and symmetry makes this easy.
  • The other known Ramsey Numbers. See here.

Theorem (Upper Bound On Ramsey Numbers)

Proof: Induct on . If or , then . Now assume the inequality holds for smaller . Then

and we are done.

Theorem (Lower Bound on Ramsey Numbers)

If , then .

Proof:

We note that . Combined with the upper bound, this says the symmetric Ramsey Numbers are exponentially large.

The key idea is to randomly color . On average, how many chromatic ‘s are there in ? There are approximately many collections of vertices. Each has

probability of being monochromatic. So, the average is

If is much smaller than , then this is less than , so some coloring must have none.

Definition (Graph Ramsey Number)

For graphs , we define the graph Ramsey Number to be the minimum so that any red-blue coloring of has either a red copy of or a blue copy of .

Note that since are contained in complete graphs, then is finite.

For example, . To prove the lower bound, we need to show any coloring of contains no monochromatic . For the upper bound, one must show any edge coloring of has at least one monochromatic .

Theorem (Graph Ramsey Number Upper Bound)

Note that the RHS is shorthand for the Ramsey Number notation. It is equivalent to .

Proof:

Let . Any red-blue coloring of has either a monochromatic complete red graph on or monochromatic blue complete graph on . These contain a red copy of or blue copy of , by embedding in and likewise for and .

Theorem (Graph Ramsey Number: Tree, Complete)

If are integers with dividing and is a Tree with vertices then

Proof:

We first want to find a lower bound. In particular, we want to color without a red or blue . Note the that

We color by partitioning it into red and connect them with blue edges.

"\\usepackage{tikz-cd}\n\\begin{document}\n\\begin{tikzpicture}\n % Node style: Red circle with black border, black text\n \\tikzset{vertex/.style={circle, draw=black, very thick, fill=red, text=black}}\n % Edge style: Blue, thick line\n \\tikzset{edge/.style={blue, line width=3.5pt}}\n\n % Vertex Coordinates\n \\node[vertex] (TC) at (0, 1.5) {$K_{m-1}$}; % Top Center\n \\node[vertex] (TL) at (-3.5, 0.8) {$K_{m-1}$}; % Top Left\n \\node[vertex] (TR) at (3.5, 0.8) {$K_{m-1}$}; % Top Right\n \\node[vertex] (BL) at (-1.5, -1.8) {$K_{m-1}$};% Bottom Left\n \\node[vertex] (BR) at (1.5, -1.8) {$K_{m-1}$}; % Bottom Right\n\n % Edges\n \\draw[edge] (TL) -- (TC);\n \\draw[edge] (TL) -- (BL);\n \\draw[edge] (TC) -- (BL);\n \\draw[edge] (TC) -- (BR);\n \\draw[edge] (TC) -- (TR);\n \\draw[edge] (BL) -- (BR);\n \\draw[edge] (BR) -- (TR);\n\n\\end{tikzpicture}\n\\end{document}"Km¡1Km¡1Km¡1Km¡1Km¡1
source code

Indeed, we have no red , since

  • the only red edges are in
  • every vertex in a is in a Cycle and thus cannot be in a tree

Likewise, there are no blue complete bipartite graph (star graph with edges) since each vertex has at most a blue degree of

where there are other vertices, and other vertices in each red , such that we can only make other blue edges. Because , we cannot make .

For the upper bound, we need to show that any red-blue coloring of a has either a red or a blue . If any vertex has or more blue edges, we have a blue . Otherwise, consider subgraph , the graph of only red edges.

Let . Then as every vertex has less than blue edges,

Therefore by construction.

Lemma: Let be any tree on vertices and a graph with . Then contains a copy of .

The proof idea is to build on one vertex at a time form . We proceed by induction on . If , then it is trivial to embed a single point. Assume we can embed any tree on vertices. Let be a leaf of . Removing and its singular edge gives us . By the inductive hypothesis, we can embed in . We now need to add back by picking some to represent (if we can do this, then we have embedded in ). Since and , then we have at minimum vertices not paired with , indeed this is , and by adding , this is our tree with vertices, and we are done.

Back to the proof, implies that we have a red tree , completing the theorem.