Conjecture 5. Let q be a (large) prime power, put n = q 2 + q + 1, let P be the set of n points of a projective plane of order q and let L1 , L2 , .
Statement
Conjecture 5. Let q be a (large) prime power, put n = q 2 + q + 1, let P be the set of n points of a projective plane of order q and let L1 , L2 , . . . , Ln be the sets of points of its lines. Let R be a random subset of P obtained by picking each point of P randomly and independently to lie in R with probability 1/2. Then with high probability the smallest cardinality of a subset B of R that intersects all the subsets L1 \cap R, L2 \cap R, . . . , Ln \cap R satisfies |B|/q > f (q) for some function f (q) tending to infinity as q tends to infinity. In fact, this may even be true with f (q) = Ω(log q). 2. Remarks • It is easy to see that the estimate in Theorem 3 is tight up \sqrt{ } to constant factors, for every partial design in which all blocks are of sizes Θ( n). • Problem 1 for block designs (and not for partial designs) was also asked by Erdős in [5]. This remains open although we suspect that the answer here is negative as well. We suggest the following conjecture which, if true, would establish this negative answer. This conjecture remains open, although related results have been proved in [2], [3] using the container method. The parameters in these papers are very different and it seems that a proof here, if true, would require additional ideas. • Call a sequence n \geq x1 \geq x2 \geq . . . \geq xm \geq 2 line-compatible for n if there is a set P of n points in the Euclidean plane R2 so that for the family L1 , L2 , . . . , Lm of all lines in R2 determined by the points of P , |Li \cap P | = xi for 1 \leq i \leq m. Note that every line-compatible sequence for n is also block- compatible for n, but the converse is not true. Erdős conjectured in [5] (see also [4], problem 733) that the number of sequences which are line compatible 1/2 for n is only 2O(n ) . This upper bound was proved by Szemerédi and Trotter in [7]. Note that in view of Theorem 4 this is much smaller than the number of block-compatible sequences for n. Indeed, there are far more block designs on n points than designs that can be described by the lines determined by a set of points in the plane. This is demonstrated by the following result.
Record
Comments
Nothing has been published against this problem yet, and nothing has been said about it.
Sign in with an institutional address to take part in the discussion. Reading every thread stays open to everyone.
Sign inSolve with an agent
Open the statement in a chat, with the problem and the ground rules already written into the prompt.