Wiki
Wiki

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

Updated


Source. Section 1, p. 347, of P. Erdős, F. Herzog and G. Piranian, Polynomials whose zeros lie on the unit circle, Duke Math. J. 22 (1955), 347--351, DOI 10.1215/S0012-7094-55-02237-7, the edition named on the source card.

Statement

Setting (p. 347). CC is the unit circle and a polynomial (1) is P(z)=∏j=1n(1−z/ωj)P(z)=\prod_{j=1}^{n}(1-z/\omega_j) with every ωj\omega_j on CC. Cohen's theorem, as the paper states it, gives for every such PP a path from the origin to CC on which "the inequality ∣P∣<1\lvert P\rvert<1 holds everywhere except at z=0z=0" (p. 347).

The question (p. 347, quoted). "Does there exist a universal constant LL such that for every polynomial (1) the inequality ∣P∣<1\lvert P\rvert<1 holds on a path which connects the origin to CC and has length at most LL?"

Answer reported (p. 347). The paper says the question was recently answered in the negative by G. R. MacLane, citing his paper On a conjecture of Erdős, Herzog, and Piranian, Michigan Math. J. 2 (1953--1954), 147--148 (reference [2]). The paper raises the question in connection with Theorem 1.

Read depth. Claims checked: the paragraph of p. 347 and reference [2] on p. 351 were read clause by clause on the page images of the print. MacLane's paper was not read here.

Proof pointer

The paper proves nothing about the question; it reports MacLane's answer and cites it.

Dependencies

Cohen, Modulus of an analytic function, Amer. Math. Monthly 59 (1952), 704--705 (reference [1]), for the path from the origin to CC; MacLane (reference [2]) for the answer.

Bears on

  • Problem 1215: the problem asks this question, with CC for the paper's LL and the polynomials normalized by P(0)=1P(0)=1 with all roots on the unit circle. The paper poses it and reports MacLane's negative answer without proving it.