Wiki
Wiki

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

Updated


Statement

Setting (p. 4, Chapter I). DD is a natural number that is not a square and AA is a nonzero squarefree integer dividing 2D2D. M(D,A)\mathfrak M(D,A) is the set of all pairs of integers X,YX,Y with X2−DY2=AX^2-DY^2=A. The fundamental pair u,vu,v of M(D,A)\mathfrak M(D,A) is the pair of natural numbers with u2−Dv2=Au^2-Dv^2=A and vv smallest. N(D,A)\mathfrak N(D,A) is the set of pairs x,yx,y in M(D,A)\mathfrak M(D,A) with y≠0y\ne0 and every prime factor of yy dividing DD.

Satz 1 (p. 15). If A≠1A\ne1 and A≠−DA\ne-D, then N(D,A)\mathfrak N(D,A) is either empty, or consists of the four pairs (±u,±v)(\pm u,\pm v), or consists of the eight pairs

(±u, ±v),(±u3+3uv2D∣A∣, ±3u2v+v3D∣A∣),(\pm u,\ \pm v),\qquad \left(\pm\frac{u^3+3uv^2D}{|A|},\ \pm\frac{3u^2v+v^3D}{|A|}\right),

where u,vu,v is the fundamental pair of M(D,A)\mathfrak M(D,A) and the signs are taken in all combinations.

The excluded cases (p. 5). For A=1A=1 the paper quotes Størmer's theorem (Videnskabsselskabets Skrifter, 1897): N(D,1)\mathfrak N(D,1) is either empty or consists of x=±ux=\pm u, y=±vy=\pm v, where u,vu,v is the fundamental pair of u2−Dv2=1u^2-Dv^2=1. For A=−DA=-D it shows that N(D,−D)\mathfrak N(D,-D) consists only of x=0x=0, y=±1y=\pm1. It adds that N(D,D)\mathfrak N(D,D) is empty for D≠2D\ne2 and consists only of x=±2x=\pm2, y=±1y=\pm1 for D=2D=2.

Singular pairs (§ 11, p. 16). The paper calls the pair D,AD,A singular when N(D,A)\mathfrak N(D,A) is the eight-pair set of Satz 1, and regular otherwise. From the formulas of § 11 it derives in § 12 (pp. 17–19; the three statements below are on pp. 18–19):

  • every pair D,−1D,-1 is regular, so for every non-square natural DD the set N(D,−1)\mathfrak N(D,-1) is empty or consists of the four pairs (±u,±v)(\pm u,\pm v); the paper identifies this sharpening of Satz 1 with Størmer's second theorem mentioned on p. 5;
  • every pair D,+2D,+2 is regular;
  • there are exactly three singular pairs D,−2D,-2, namely D=3,6,123D=3,6,123, with N(3,−2)={(±1,±1),(±5,±3)}\mathfrak N(3,-2)=\{(\pm1,\pm1),(\pm5,\pm3)\}, N(6,−2)={(±2,±1),(±22,±9)}\mathfrak N(6,-2)=\{(\pm2,\pm1),(\pm22,\pm9)\} and N(123,−2)={(±11,±1),(±2695,±243)}\mathfrak N(123,-2)=\{(\pm11,\pm1),(\pm2695,\pm243)\}; for every other DD the set N(D,−2)\mathfrak N(D,-2) is empty or consists of (±u,±v)(\pm u,\pm v) (under the standing assumption A≠−DA\ne-D of § 2, which excludes D=2D=2).

A table of singular pairs for A0=±1,±2A_0=\pm1,\pm2 is printed on p. 20.

Proof pointer

§§ 2–10, pp. 5–15. A result of D. Schepel (Nieuw Archief voor Wiskunde, 1935), quoted on pp. 6–7, describes M(D,A)\mathfrak M(D,A) for A≠1A\ne1, A≠−DA\ne-D through the fundamental pair: the solutions are ±Xm,±Ym\pm X_m,\pm Y_m, where Xm+YmD=(u+vD)2m+1/∣A∣mX_m+Y_m\sqrt D=(u+v\sqrt D)^{2m+1}/|A|^m. So the question becomes which odd n=2m+1n=2m+1 give a yny_n with only prime factors dividing DD; the set of such nn is written n(D,A)\mathfrak n(D,A) (p. 8). Since yνy_\nu divides yny_n when ν\nu divides nn, the set is closed under odd divisors. Writing D=D0D1D=D_0D_1, A=A0D1A=A_0D_1, u=u0D1u=u_0D_1 with D1=(D,A)D_1=(D,A) (§ 5, pp. 9–10), a binomial expansion of yny_n and congruences modulo D1D_1, D0D_0 and p2p^2 show that every element of n(D,A)\mathfrak n(D,A) is a power of 33 (§§ 6–7, pp. 10–13) and that 99 is not an element (§§ 8–9, pp. 13–14). Hence n(D,A)\mathfrak n(D,A) is empty, {1}\{1\} or {1,3}\{1,3\} (§ 10, pp. 14–15), the three cases of Satz 1.

Read depth

Claims checked: the setting, Satz 1, the excluded cases and the consequences of § 12 were read clause by clause on the page images of the print; the three singular pairs D,−2D,-2 were checked against x2−Dy2=−2x^2-Dy^2=-2 here. The proof was followed but not checked step by step. Nothing here is independently reviewed.

Dependencies

External inputs named by the paper: Schepel's description of M(D,A)\mathfrak M(D,A) and Størmer's theorem for A=1A=1.

Source. K. Mahler, Über den grössten Primteiler spezieller Polynome zweiten Grades, Archiv for Mathematik og Naturvidenskab 41 (1935), no. 6, pp. 3–26; the edition read is named on the source card.

Bears on

None directly. Satz 1, with Størmer's theorem for A=1A=1, is the step that makes the set M(z)M(z) of Satz 3 computable, and Satz 3 bears on Problem 368 and Problem 649.