Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Source. Published p. 265, Theorem 1.15, and pp. 279–281 (PDF).
Statement. Given , there is such that for a compatible matrix with integer entries , and partition families , , density product at least implies
Compatibility means exactly the stated row and column marginals. All pairs are ordered. Constants are uniform over the feasible matrix sizes and marginals.
Proof. If one matrix dimension is one, its corresponding partition family, being nonempty, contains its unique possible partition. Every member of the other family then has pattern , so choosing proves (1). For , induct on . The base is Theorem 1.14: recording the first cell of each partition identifies its four atoms with the four entries of .
Transpose if necessary so that . Write , , and let be the fiber of with first cell , an ordered -partition of . Put
Since , Lemma 4.1, or its direct fiber count, gives a family of at least sets , each with .
Collapse rows of into one row, obtaining the matrix . Its entries remain at least . The pair of families consisting of the two-cell partitions for , and , has density product at least . Apply the induction hypothesis for , with a small output tolerance . It gives at least
This use is valid because . For a fixed , there are at most compatible full-family partitions . Removing those with fewer than such members of loses at most half the count in (2). There are therefore at least remaining .
Fix one. For each ordered partition of with , let be the family of residual partitions of obtained from members of with . There are choices of and at most residual partitions in each fiber. The count at this is at least , so at least choices of satisfy
Let be with its first row removed. Its entries are at least . Apply the induction hypothesis to on the actual -set, with output tolerance . Their density product is at least . To justify the parameter order, first take the input tolerance supplied for this residual induction, then choose small in comparison with it and . Since , choose still smaller; for large the last density product exceeds the required threshold on coordinates. Finally decrease to meet the earlier coarse induction (2). This proves at least residual pairs.
Joining the residual row partition to , and joining each residual column to its specified , reconstructs one original pair with pattern . Conversely that original pair determines and both residual partitions, so no pair is counted twice. The total is at least
The factorial identity is exact. Choosing and then large proves (1).
For fixed only finitely many shapes occur, since . The same applies to all residual and collapsed shapes. The uniform Theorem 1.14 and the proportional bound therefore allow a common positive tolerance throughout this finite induction. For the finitely many excluded small , reduce so that the density hypothesis forces both partition families to be full. This completes the proof with the stated uniformity.
Precision. In the source's intermediate set-family notation, a single displayed column records a set together with its implicit complement. Here both cells, the residual column sizes, and all normalizing factors are retained explicitly; no one-column pattern count is substituted for a two-cell partition count.
Dependencies. theorem_1_14, lemma_4_1, definitions.