Wiki
Wiki

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

Updated


Source. Theorem 9, printed p. 346, physical p. 6 of the published paper. The displayed statement is planar: its ambient space is R2\mathbb R^2.

Let d>0d>0. Let T1,T2,T3T_1,T_2,T_3 be triangles such that T1T_1 has a side of length dd, T2T_2 has a side of length 3d\sqrt3d, and T3T_3 has a side of length 2d2d. Every two-coloring of R2\mathbb R^2 contains a monochromatic triangle congruent to at least one of T1,T2,T3T_1,T_2,T_3.

Equilateral forcing

Put

u=d(1,0),v=d(1/2,3/2).(1)u=d(1,0),\qquad v=d(1/2,\sqrt3/2). \tag{1}

By the positive part of Theorem 5, there are same-colored points at distance dd. Apply an isometry and call 0,u0,u red, exchanging the color names if necessary. Suppose there is no monochromatic equilateral triangle with side length dd, 3d\sqrt3d, or 2d2d.

The points vv and u−vu-v complete the two unit-scale equilateral triangles on 0,u0,u, so both are blue. The three points

v,u−v,2u(2)v,\qquad u-v,\qquad 2u \tag{2}

form an equilateral triangle of side 3d\sqrt3d; hence 2u2u is red. The triangle 0,2u,2v0,2u,2v has side 2d2d, so 2v2v is blue.

Now u,2u,u+vu,2u,u+v form an equilateral triangle of side dd, so u+vu+v cannot be red. But v,2v,u+vv,2v,u+v is another such triangle, so u+vu+v cannot be blue. This contradiction proves that a monochromatic equilateral triangle exists at one of the three scales.

Passing to the prescribed triangles

The coordinate construction in Theorem 8 has the following planar consequence: from a monochromatic equilateral triangle whose side equals one side of a prescribed triangle TT, its six congruent planar copies force a monochromatic copy of TT. Apply that gadget to T1T_1, T2T_2, or T3T_3 according to which scale was forced above. This proves the theorem.

Used by. Corollary 10.

Bears on. #173: the proof's first step shows that no two-coloring of the plane misses the equilateral triangles of all three sides dd, 3d\sqrt3d and 2d2d. The problem's page, through Theorem 1 of the 1975 sequel, reduces the question to whether a coloring can miss equilateral triangles of two different sides; this theorem excludes only missing all three sides in these ratios at once. It forces a single prescribed triangle only when one triangle has all three sides, the case of Corollary 10.