Wiki
Wiki

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

Updated


Source. P. Erdős, Some unsolved problems, Magyar Tud. Akad. Mat. Kutató Int. Közl. 6 (1961), 221--254, printed pp. 238--239 / PDF pp. 18--19 of the archive's scan. The source was checked visually. The historical statement and the imported formulation are distinguished below.

What the 1961 page says

Following the five-variable indefinite-form discussion, printed p. 239 reads: "It is not known if for every irrational α\alpha and ϵ>0\epsilon>0 the inequalities

∣(x2+y2)α−z2∣<ϵand∣x2+y2−z2α∣<ϵ\left|(x^2+y^2)\alpha-z^2\right|<\epsilon \qquad\text{and}\qquad \left|x^2+y^2-z^2\alpha\right|<\epsilon

are solvable in integers. The case α=2\alpha=\sqrt2 is also undecided." The first sentence prints neither α>0\alpha>0 nor positivity of every coordinate, and it does not explicitly exclude (0,0,0)(0,0,0). Taken literally without a nonzero condition, that triple satisfies both inequalities. The preceding indefinite-form context motivates a nontrivial positive-parameter reading, but it does not insert missing words into the printed sentence.

The imported Problem 496 instead retains only the second inequality, quantifies over all irrational real α\alpha, and requires all three coordinates to be positive integers. Its exact statement is preserved on the problem page. These are traceable formulations, not a silent replacement of one by the other.

The imported formulation fails at every negative irrational α\alpha: for positive integers, ∣x2+y2−αz2∣=x2+y2+∣α∣z2≥2+∣α∣|x^2+y^2-\alpha z^2|=x^2+y^2+|\alpha|z^2\ge2+|\alpha|. Boris Alexeev's Lean development Erdos496.lean proves this negation at α=−2\alpha=-\sqrt2, ϵ=1\epsilon=1.

Positive-parameter formulation

For each irrational α>0\alpha>0, the all-positive-coordinate conclusion for the second inequality follows from the distinct published Margulis Banach Center Theorem 1 and the complete coordinate transfer. Its source theorem is used as an explicitly stated external input; its homogeneous-dynamics proof is not reconstructed here.

Bears on. Problem 496.