Let Z be the set of all integers. Determine all functions f:Z→Z such that f(f(x)+f(y))+f(x)f(y)=f(x+y)f(x−y) holds for all x,y∈Z
Solution
Clearly f(x)=0 is a solution to the original equation. So suppose there exists t such that f(t)=0. Substituting (0,0) into the original equation gives f(2f(0))=0 Substituting (2f(0),0) gives f(f(0))=0 Substituting (f(0),f(0)) gives f(0)=0 Substituting (x,0) gives f(f(x))=f(x)2(1) Substituting (x,x) gives f(2f(x))=−f(x)2(2) For convenience, let g:Z→Z satisfy, for any integer n, g(n)=⎩⎨⎧0,1,−1,when n≡0(mod5);when n≡1,4(mod5);when n≡2,3(mod5). Next we first prove a lemma. Lemma. If f(s)=0, then f(mf(s))=g(m)f(s)2∀m∈N Proof: We use mathematical induction. For m=1,2, this follows from (1) and (2) respectively. Suppose it holds for all m<k; then for m=k (with k≥3) (a) k=5q. Substituting ((5q−1)f(s),f(s)) into the original equation and combining with the induction hypothesis, we obtain f(2f(s)2)+f(s)4=−f(5qf(s))f(s)2(3) Also, by (1) we know f(s)2=f(f(s)), so by (1) and (2) we know f(2f(s)2)=f(2f(f(s)))=−f(f(s))2=−f(s)4 Substituting into (3) shows f(5qf(s))f(s)2=0; and since f(s)=0, we have f(5qf(s))=0=g(5q)f(s)
(b) k=5q+1. Substituting ((5q−1)f(s),2f(s)) into the original equation and combining with the induction hypothesis gives f((5q+1)f(s))=f(s)2=g(5q+1)f(s)2 (c) k=5q+2. Substituting ((5q−1)f(s),3f(s)) into the original equation and combining with the induction hypothesis gives f((5q+2)f(s))=−f(s)2=g(5q+2)f(s)2 (d) k=5q+3. Substituting ((5q+2)f(s),f(s)) into the original equation and combining with the induction hypothesis gives f((5q+3)f(s))=−f(s)2=g(5q+3)f(s)2 (e) k=5q+4. Substituting ((5q+3)f(s),f(s)) into the original equation and combining with the induction hypothesis gives f((5q+4)f(s))=f(s)2=g(5q+4)f(s)2 In all cases, f(kf(s))=g(k)f(s)2, so by mathematical induction the lemma is proved. Returning to the original problem, by (1) and (2) there exists p such that f(p)=∣f(t)∣f(t) (p being either f(t) or 2f(t)). By (1) and the lemma we know f(t)4=f(p)2=f(f(p))=f(∣f(t)∣f(t))=g(∣f(t)∣)f(t)2 So f(t)2=g(∣f(t)∣) (since f(t)=0) But ∣g(∣f(t)∣)∣≤1 and f(t)=0, so f(t)2=1. By (1) and (2) we get f(f(t))=1,f(2f(t))=−1. Next we prove that for all n∈Z, f(n)=g(n). If n=0, then f(n)=0=g(n) If n>0, substituting s with f(t) and m with n in the lemma gives f(n)=f(nf(f(t)))=g(n)f(f(t))2=g(n) If n<0, substituting s with 2f(t) and m with −n in the lemma gives f(n)=f(−nf(2f(t)))=g(−n)f(2f(t))2=g(−n)=g(n) In summary, f(n)=g(n). Verification by substitution: it is easy to see that g(n)2≡n2(mod5) So the left-hand side of the original equation≡(x2+y2)2+(xy)2≡(x2+y2)2−(2xy)2≡(x2−y2)2≡the right-hand side (mod 5 Also, since ∣g(n)∣≤1, we have ∣left-hand side−right-hand side∣≤3. Hence the left-hand side equals the right-hand side, and the verification is complete. Therefore there are two solutions for f: f(x)=0∀x∈Z f(x)=⎩⎨⎧0,1,−1,when x≡0(mod5);when x≡1,4(mod5);when x≡2,3(mod5).∀x∈Z.
Want a route through all this instead of an archive? The track
puts 2,000 problems in a working order, from AMC 10 level to the IMO shortlist.
Source: MathNet,
licensed CC-BY-4.0.
Statement translated into English from zh; metadata (topic, difficulty) added by this project.