Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated


Source. The first paragraph of p. 53 of P. Erdős, Combinatorial problems in geometry, Math. Chronicle 12 (1983), 35--54, the transcript of an invited address at the 17th New Zealand Mathematics Colloquium (Dunedin, 17--19 May 1982), as named on the source card. The lecture numbers none of its statements; the pages are the journal's own.

Statement

Problem and conjecture (p. 53). How many distinct distances do nn points in the plane determine? Erdős thinks the answer is n/log⁡nn/\sqrt{\log n}, which the lattice points attain: by a theorem of Landau the number of distinct values u2+v2\sqrt{u^2+v^2} in the relevant range is of order n/log⁡nn/\sqrt{\log n} (the print writes "the number of distinct integers of the form u2+v2\sqrt{u^2+v^2} is n/log⁡nn/\sqrt{\log n}"). He offers a prize for the problem.

Reported lower bounds (p. 53). Erdős originally proved that nn points determine n\sqrt n distinct distances; Leo Moser improved this to n2/3n^{2/3}, and Fan Chung very recently to n5/7n^{5/7}, with a proof Erdős calls very tricky. The print states the bounds without constants.

Read depth. Claims checked: the passage was read clause by clause on the page images of the print. A second reader checked the statement, hypotheses, label and page against the print.

Proof pointer

The paper proves none of these; it reports them.

Dependencies

None.

Bears on

  • Problem 89: Erdős's conjecture here is that the right order is n/log⁡nn/\sqrt{\log n}, which includes the problem's lower bound ≫n/log⁡n\gg n/\sqrt{\log n}. The lecture records the lower bounds known in 1982 and proves nothing toward the problem.