Wiki
Wiki

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

Updated


Statement

Setting (pp. 487--488). LkL_k (k≥2k\ge2) is the lattice of integer points (x1,…,xk)(x_1,\ldots,x_k). A point is visible when its coordinates have no common divisor greater than 11, and nonvisible otherwise; the origin counts as nonvisible. Visible points are drawn as circles and nonvisible points as crosses. A pattern PkP_k assigns to each of the wkw^k points with 1≤xλ≤w1\le x_\lambda\le w a circle, a cross, or neither. PkP_k is realized in LkL_k when some (u1,…,uk)∈Lk(u_1,\ldots,u_k)\in L_k makes (u1+x1,…,uk+xk)(u_1+x_1,\ldots,u_k+x_k) visible for every circle (x1,…,xk)(x_1,\ldots,x_k) of PkP_k and nonvisible for every cross.

Definition (p. 490). For a positive integer mm, a complete square modulo mm is a set of m2m^2 points of L2L_2 forming a complete system of residues modulo mm: every (x,y)(x,y) with 0≤x<m0\le x<m, 0≤y<m0\le y<m is congruent modulo mm, coordinate by coordinate, to exactly one point of the set.

Theorem 1 (p. 490, quoted). "A given pattern P2P_2 can be realized in L2L_2 if and only if the set CC of circles in P2P_2 fails to contain a complete square modulo pp for every prime pp."

So the condition concerns the circles alone; the crosses never obstruct a realization (p. 489). A pattern of four circles on a 2×22\times2 block cannot be realized, since one of its points has both coordinates even (p. 489).

Corollary 1 (p. 492, quoted). "Every pattern P2P_2 consisting only of crosses can be realized." The condition of Theorem 1 holds vacuously.

The paper also remarks (p. 492) that the construction shows a realizable pattern occurs in L2L_2 infinitely often.

Proof pointer

Pp. 490--492. Necessity: if CC contains a complete square modulo pp, then for every translate some circle lands on a point with both coordinates divisible by pp. Sufficiency, by the Chinese Remainder Theorem in three steps: for each prime p≤wp\le w, choose (u,v)(u,v) modulo pp so that no translated circle is ≡(0,0)\equiv(0,0), using a residue class that CC misses (congruences (5)); give each cross (i,j)(i,j) its own prime Q(i,j)>wQ(i,j)>w and put (u,v)≡(−i,−j)(u,v)\equiv(-i,-j) modulo it (congruences (6)); then fix u>0u>0 and require v≡0v\equiv0 modulo every prime q>wq>w other than the Q(i,j)Q(i,j) dividing one of u+1,…,u+wu+1,\ldots,u+w (congruences (7)), so that no such qq divides v+yv+y for 1≤y≤w1\le y\le w.

Read depth

Claims checked: the definitions, Theorem 1, Corollary 1 and the remarks on pp. 489 and 492 were read clause by clause on the page images of the print, and the proof on pp. 490--492 was followed. Nothing here is independently reviewed.

Dependencies

None in the corpus. The proof uses only the Chinese Remainder Theorem.

Source. Fritz Herzog and B. M. Stewart, Patterns of visible and nonvisible lattice points, Amer. Math. Monthly 78 (1971), no. 5, 487--496, doi:10.2307/2317753; the edition read is named on the source card.

Bears on

None of the problem pages directly. Theorem 1 is the criterion from which Corollary 2 follows; that page states the paper's relation to Problem 1212.