Online Ramsey Games in Random Graphs Reto Spöhel, ETH Zürich Joint work with Martin Marciniszyn and Angelika Steger TexPoint fonts used in EMF. Read the TexPoint manual before you delete this box.: A A A A AA A A

Introduction Ramsey theory: when are the edges/vertices of a graph colorable with r colors without creating a monochromatic copy of some fixed graph F ? For random graphs: solved in full generality by Ł uczak/Ruci ń ski/Voigt, 1992 (vertex colorings) Rödl/Ruci ń ski, 1995 (edge colorings)

Introduction ‚solved in full generality‘: Explicit threshold functions p 0 (F, r, n) such that In fact, p 0 (F, r, n) = p 0 (F, n), i.e., the threshold does not depend on the number of colors r [except …] The threshold behaviour is even sharper than shown here [except …] We transfer these results into an online setting, where the edges/vertices of G n, p have to be colored one by one, without seeing the entire graph.

The online edge-coloring game Rules: one player, called Painter start with empty graph on n vertices edges appear u.a.r. one by one and have to be colored instantly (‚online‘) either red or blue game ends when monochromatic triangle appears Question: How many edges can Painter color? Theorem (Friedgut, Kohayakawa, Rödl, Ruci ń ski, Tetali, 2003): The threshold for this game is N 0 ( n ) = n 4/3. (easy, not main result of paper)

Our results Online edge-colorings: threshold for online-colorability with 2 colors for a large class of graphs F including cliques and cycles Online vertex-colorings [main focus of this talk]: threshold for online-colorability with r R 2 colors for a large class of graphs F including cliques and cycles Unlike in the offline cases, these thresholds are coarse and depend on the number of colors r.

The online vertex-coloring game Rules: random graph G n, p, initially hidden vertices are revealed one by one along with induced edges vertices have to be instantly (‚online‘) colored with one of r R 2 available colors. game ends when monochromatic copy of some fixed forbidden graph F appears Question: How dense can the underlying random graph be such that Painter can color all vertices a.a.s.?

Example F = K 3, r = 2

Main result (simplified) Theorem (Marciniszyn, S., 2006+) Let F be a clique or a cycle of arbitrary size. Then the threshold for the online vertex-coloring game with respect to F and with r R 2 available colors is i.e.,

Bounds from ‚offline‘ graph properties G n, p contains no copy of F  Painter wins with any strategy G n, p allows no r-vertex-coloring avoiding F  Painter loses with any strategy  the thresholds of these two ‚offline‘ graph properties bound p 0 (n) from below and above.

Appearance of small subgraphs Theorem (Bollobás, 1981) Let F be a non-empty graph. The threshold for the graph property ‚G n, p contains a copy of F‘ is where

Appearance of small subgraphs For ‚nice‘ graphs – e.g. for cliques or cycles – we have (such graphs are called balanced)

Vertex-colorings of random graphs Theorem ( Ł uczak, Ruci ń ski, Voigt, 1992) Let F be a graph and let r R 2. The threshold for the graph property ‚every r-vertex-coloring of G n, p contains a monochromatic copy of F‘ is where

Vertex-colorings of random graphs For ‚nice‘ graphs – e.g. for cliques or cycles – we have (such graphs are called 1-balanced) For these graphs,. is also the threshold for the property ‚There are more than n copies of F in G n, p ‘ Intuition: For p [ p 0, the copies of F overlap in vertices, and coloring G n, p becomes difficult.

For arbitrary F and r we thus have Theorem Let F be a clique or a cycle of arbitrary size. Then the threshold for the online vertex-coloring game with respect to F and with r R 1 available colors is r = 1  Small Subgraphs r    exponent tends to exponent for offline case Main result revisited

Lower bound (r = 2 ) Let p(n) / p 0 (F, 2, n) be given. We need to show: There is a strategy which allows Painter to color all vertices of G n, p a.a.s.

Lower bound (r = 2 ) We consider the greedy strategy: color all vertices red if feasible, blue otherwise.  after the losing move, G n, p contains a blue copy of F, every vertex of which would close a red copy of F. For F = K 4, e.g. or

Lower bound (r = 2 )  Painter is safe if G n, p contains no such ‚dangerous‘ graphs. Lemma Among all dangerous graphs, F * is the sparsest one, i.e., m(F *) % m(D) for all dangerous graphs D. F*F* D

Lower bound (r = 2 ) Corollary Let F be a clique or a cycle of arbitrary size. Playing greedily, Painter a.a.s. wins the online vertex- coloring game w.r.t. F and with two available colors if F *

Lower bound (r = 3 ) Corollary Let F be a clique or a cycle of arbitrary size. Playing greedily, Painter a.a.s. wins the online vertex- coloring game w.r.t. F and with three available colors if F 3*F 3* F *

Lower bound Corollary Let F be a clique or a cycle of arbitrary size. Playing greedily, Painter a.a.s. wins the online vertex- coloring game w.r.t. F and with r R 2 available colors if …

The general case In general, it is smarter to greedily avoid a suitably chosen subgraph H of F instead of F itself.  general threshold function for game with r colors is where Maximization over r possibly different subgraphs H i  F, corresponding to a „smart greedy“ strategy. F H

A surprising example F = H 1 ] H 2 H1H1 H2H2 (lower bound only)

Upper bound Let p(n) [ p 0 (F, r, n) be given. We need to show: The probability that Painter can color all vertices of G n, p tends to 0 as n  , regardless of her strategy. Proof strategy: two-round exposure & induction on r First round n/ 2 vertices, Painter may see them all at once use known ‚offline‘ results Second round remaining n/ 2 vertices Due to coloring of first round, for many vertices one color is excluded  induction.

Upper bound V1V1 V2V2 F ° 1)Suppose Painter‘s offline-coloring of V 1 creates many (w.l.o.g.) red copies of F ° 2)Depending on the edges between V 1 and V 2, these copies induce a set Base(R) 4 V 2 of vertices that cannot be colored red. 3)Edges between vertices of Base(R) are independent of 1) and 2)  Base(R) induces a binomial random graph Base(R) F  need to show: Base(R) is large enough for induction hypothesis to be applicable.

There are a.a.s. many monochromatic copies of F‘° in V 1 provided that work (Janson, Chernoff,...)  These induce enough vertices in (w.l.o.g.) Base(R) such that the induction hypothesis is applicable to the binomial random graph induced by Base(R). Upper bound

Main result (full) Theorem (Marciniszyn, S., 2006+) Let F be a graph for which at least one F ° satisfies Then the threshold for the online vertex-coloring game w.r.t. F and with r R 1 colors is F°F°

Side remark: Trees Greedy strategy gives lower bound of f for any tree T on v T vertices and any number r of colors. Theorem (Mütze, S., 2008+): For any fixed tree T and any number r of colors, the precise threshold can be found by finite computation. For, the threshold for the online vertex- coloring game with respect to and with two available colors is But: the threshold for the online vertex-coloring game with respect to and with two available colors is at least

Back to online edge colorings Threshold is given by appearance of F *, yields threshold formula similarly to vertex case. Lower bound: Much harder to deal with overlapping outer copies! Works for arbitrary number of colors. Upper bound: Two-round exposure as in vertex case But: unclear how to setup an inductive argument to deal with r R 3 colors. F_F_ F °F ° F *

Online edge colorings Theorem (Marciniszyn, S., Steger, 2005+) Let F be a graph that is not a tree, for which at least one F_ satisfies Then the threshold for the online edge-coloring game w.r.t. F and with two colors is F_F_

Open problems More colors (edge case) Simplest open case: F = K 3, r = 3 General non-trees is not the truth in general! Is there an explicit general threshold formula? Trees Is it just combinatorial chaos, or is there a hidden pattern ?

Outlook: balanced online games The greedy strategy produces very unbalanced colorings.  consider new ‚balanced‘ game: in every step, r vertices/edges appear at once, and Painter has to assign each of the r available colors to exactly one of these vertices. The case of edge colorings and r = 2 was previously studied by Marciniszyn, Mitsche, Stojakovi ć (2005), who proved e.g. a threshold of n 6/5 for the triangle (the threshold in the unbalanced game is n 4/3 ).

Balanced online vertex-coloring games Theorem (Prakash, S., 2007+) Let F be a clique or a cycle of arbitrary size. Then the threshold for the online balanced vertex- coloring game with respect to F and with r R 1 available colors is again r = 1 corresponds to Small Subgraphs theorem, and for r  , the exponent tends to exponent for offline case. Key insight to go to arbitrary number of colors: in every step, finding a valid coloring corresponds to finding a perfect matching in some bipartite graph.  use Hall‘s Theorem

Two recent results… Theorem (Thomas, S., 2008+) [similar result for balanced edge-coloring game] We can prove the upper bound for arbitrary r R 1 because no induction is needed. Theorem (Krivelevich, S., Steger, 2008+) The corresponding offline edge-coloring problem has the same threshold as the normal Ramsey problem [except …].

Thank you! Questions?