For a positive integer k≥2 define Tk={(x,y)∣x,y=0,1,…,k−1} to be a collection of k2 lattice points on the cartesian coordinate plane. Let d1(k)>d2(k)>⋯ be the decreasing sequence of the distinct distances between any two points in Tk. Suppose Si(k) be the number of distances equal to di(k). Prove that for any three positive integers m>n>i we have Si(m)=Si(n).
This one wants a proof. Work it on paper, then read the official solution and mark
yourself. Be honest about it: the record is only any use to you if it is.
Official solution
1. Define the problem and notation: For a positive integer k≥2, define Tk={(x,y)∣x,y=0,1,…,k−1} to be a collection of k2 lattice points on the Cartesian coordinate plane. Let d1(k)>d2(k)>⋯ be the decreasing sequence of the distinct distances between any two points in Tk. Suppose Si(k) is the number of distances equal to di(k). We need to prove that for any three positive integers m>n>i, we have Si(m)=Si(n).
2. Introduce the induction hypothesis: Set Di(n)=di(n+1), and associate with Di(n) a pair (xn,i,yn,i)∈{0,1,…,n}2 with n≥xn,i≥yn,i≥0 and xn,i2+yn,i2=Di(n)2. We will prove by induction on n that for 1≤i≤n, the pairs (xi,yi) are unique and equal, in order: (n,n),(n,n−1),(n,n−2),(n−1,n−1),(n,n−3),(n−1,n−2),…
3. Base case: When n=1, the result is obvious because the only possible pairs are (1,1) and (1,0).
4. Induction step: Suppose the hypothesis holds for n=N−1. We need to show it holds for n=N. Notice that: (xN,i,yN,i)=(xN−1,i+1,yN−1,i+1) for 1≤i≤N−1. This is because when a2+b2>c2+d2 and a+b≥c+d, (a+1)2+(b+1)2=a2+b2+2(a+b)+1>c2+d2+2(c+d)+1=(c+1)2+(d+1)2 and we know yi>0 and the sum xN−1,i+yN−1,i is non-increasing in i.
5. Bijection and uniqueness: There is a bijection from Di(N−1) to Di(N) for 1≤i≤N, so the pairs (xN,i,yN,i) are unique and the hypothesis holds for 1≤i≤N−1.
6. Final pair: It remains to show that (xN,N,yN,N) obeys the hypothesis as well. Let s=xN,N−1+yN,N−1. If xN,N−1−yN,N−1≤1, we have exhausted all possible values of xN,i+yN,i greater than or equal to s, and so xN,N+yN,N≤s−1. It follows that xN,N=n and yN,N=s−n−1 by the convexity of f(t)=t2.
7. Explicit computation: Otherwise, using the hypothesis, we can explicitly compute: xN,N−1=N−(k−1) yN+1,N=N+1+(k−1)−⌊4N+2⌋ where k=N−1−⌊4⌊4N+2⌋2⌋
8. Convexity argument: We claim xN,N=xN,N−1−1 and yN,N+yN,N−1+1. Again by the convexity of f(t)=t2, it is sufficient to show: (N−k)2+(N+1+k−⌊4N+2⌋)2>N2+(N−⌊4N+2⌋)2 2k2−(2⌊4N+2⌋−2)k+(2N−2⌊4N+2⌋+1)>0 k=N−1−⌊4⌊4N+2⌋2⌋<42⌊4N+2⌋−2−(2⌊4N+2⌋−2)2−8(2N−2⌊4N+2⌋+1)
9. Simplification: Now set g=⌊4N+2⌋. This can be rewritten: N−1−⌊4g2⌋<2g−1−(g+1)2−(4N+2) (g+1)2−(4N+2)<2⌊4g2⌋+g+1−2N 2(g+1)2−(4N+2)<4⌊4g2⌋+2g+4−(4N+2)
10. AM-GM inequality: By AM-GM, it's sufficient to show: 1+[(g+1)2−(4N+2)]<4⌊4g2⌋+2g+4−(4N+2) g2<4⌊4g2⌋+2 which follows from the fact that squares are either 0 or 1 modulo 4.
Thus, we have shown that Si(m)=Si(n) for all m>n>i.
■
Source: NuminaMath-1.5,
licensed Apache-2.0.
Statement and solution reproduced as published; topic, difficulty and ordering added
by this site.