AlgebraDifficulty 9.1Prove it2022 China Team Selection Test · China · 2022
Find all functions f:R→R such that, for any real numbers x,y, the following two multisets are equal {f(xf(y)+1),f(yf(x)−1)}={xf(f(y))−1,yf(f(x))+1}.
*Remark: {a,b}={c,d} are equal as multisets if a=c and b=d, or a=d and b=c.*
This one wants a proof. Work it on paper, read the official solution, then mark
yourself honestly — the ladder only means something if the record is true.
Official solution
Solution: All functions satisfying (*) are f(x)=x and f(x)=−x. It is easy to verify that these two functions satisfy (*). In what follows, all sets are multisets. Take x=y=0 in (*) gives {f(1),f(−1)}={1,−1}. First consider the case when f(1)=1. We want to show that f(x)=x for any x∈R. *Step 1:* Prove that f(n)=n for n∈Z. We first show that f(0)=0. Take x=0 in (*), we have {f(1),f(yf(0)−1)}={−1,yf(f(0))+1}. If f(0)=0, take y=f(0)2; then the LHS is {1,1}, which cannot contain the −1 on the RHS. So f(0)=0. Next, we prove inductively f(n)=n for n∈N. The case n=1 is known. Assume that f(m)=m holds for m≤n. Take x=1 and y=n in (*), we deduce {f(n+1),f(n−1)}={n−1,n+1} From this, we deduce that f(n+1)=n+1, completing the inductive proof. Using a similar argument but taking x=1 and y=−n proves that f(−n)=−n for n∈N. This completes Step 1. *Step 2:* Prove that f is a bijection. f is surjective because when setting y=1 in (*), the RHS contains xf(f(1))−1=x−1, which can take an arbitrary real number. By the form of its LHS, all real numbers are in the image of f. Next, we prove f(y0)=0 implies y0=0. Suppose not, namely y0=0. Then taking y=y0 in (*) gives {f(1),f(y0f(x)−1)}={−1,y0f(f(x))+1}. We must have 1=f(1)=y0f(f(x))+1. So f(f(x))=0 holds for all x. This contradicts with f(1)=1. We next prove the injectivity of f. If y1=y2 satisfy f(y1)=f(y2)=0. Taking respectively y=y1 and y=y2 in (*) gives {f(xf(y1)+1),f(y1f(x)−1)}={xf(f(y1))−1,y1f(f(x))+1}, {f(xf(y2)+1),f(y2f(x)−1)}={xf(f(y2))−1,y2f(f(x))+1}.
Note that the first elements on both sides of the two equations are the same, respectively. If some x0∈R satisfies f(x0f(y1)+1)=x0f(f(y1))−1, then y1f(f(x0))+1=f(xf(y1)+1)=f(xf(y2)+1)=y2f(f(x0))+1 From this, we get y1f(f(x0))=y2f(f(x0)), thus f(f(x0))=0, which implies x0=0. So if x=0, then f(xf(y1)+1)=xf(f(y1))−1. From this, we deduce that f(x)=ax+b for some a,b∈R and any x=1 is a linear function. But f(n)=n for n∈Z implies that a=1 and b=0. So f(x)=x is injective. Step 3: Prove that for any n∈Z and y∈R, we have f(f(f(y)n))=yn(1) It is enough to consider the case when n=0 and y=0. Taking x=f(y)n in (∗) gives {n+1,f(yf(f(y)n)−1)}={nf(y)f(f(y))−1,yf(f(f(y)n))+1}.(2) Similarly, substituting y for x and f(y)n for y in (∗) gives {n−1,f(yf(f(y)n)+1)}={nf(y)f(f(y))+1,yf(f(f(y)n))−1}.(3) If there exists some y0∈R∖{0} such that n=y0f(f(f(y0)n)), then combining (2) and (3) gives n+1=nf(y0)f(f(y0))−1,n−1=nf(y0)f(f(y0))+1. Taking the differences of the two equations gives 2=−2, which is a contradiction. Step 4: For α∈Q and y∈R, prove that f(αy)=αf(y). We may assume that α=0 and y=0. Substituting f(f(y)m) with m∈Z∖{0} for y in (1) gives fff(f(f(y)m))n=f(f(y)m)n. Plugging (1) into the denominator of the LHS gives f(f(mny))=f(f(y)m)n. In the above equality, substituting f(y)t with t∈Z∖{0} for y gives ynt/m=f(f(mnf(y)t))=f(f(f(y)t)m)n.
This is equivalent to f(f(f(y)t)m)=tmy. Taking f on both sides gives tmf(y)=f(f(f(f(y)t)m))=f(tmy). This completes Step 4. Step 5: Prove that f(y+a)=f(y)+a for all y∈R and a∈Q. Taking x=r1∈Q∖{0} in (*) gives {f(r1f(y)+1),f(r1y−1)}={r1f(f(y))−1,r1y+1}. Using Step 4 and multiplying both sides by r gives {f(f(y)+r),f(y−r)}={f(f(y))−r,y+r}.(4) We want to prove that f(y−r)=f(f(y))−r. Suppose that there exist some y0∈R and r0∈R∖{0} such that f(y0−r0)=y0+r0. Taking y=y0−r0 and r=−2r0 in (4) gives {f(f(y0−r0)−2r0),f(y0−r0+2r0)}={f(f(y0−r0))+2r0,y0−3r0}. This is the same as {y0+r0,f(y0+r0)}={f(y0+r0)+2r0,y0−3r0}. Since y0+r0=y0−3r0, we must have y0+r0=f(y0+r0)+2r0, i.e. f(y0+r0)=y0−r0. But then the two sides of the equality becomes {y0+r0,y0−r0}={y0+r0,y0−3r0}. They are obviously not equal, giving a contradiction. So f(y−r)=f(f(y))−r must hold for all y∈R and r∈Q∖{0}. Taking two different values of r and taking the difference gives f(y+a)=f(y)+a for all y∈R and a∈Q. Step 6: Prove that f(y)=y for y∈R. Combining Step 5 and (4) gives {f(f(y))+r,f(y)−r}={f(f(y))−r,y+r}. Taking the sum of the two elements on both sides of the equality, we deduce that f(f(y))+f(y)=f(f(y))+y, i.e. f(y)=y. Now we consider the case when f(1)=−1. We aim to prove that f(x)=−x (x∈R). Step 1: Prove that f(n)=−n (n∈Z).
Taking y=1 and respectively x=1,−1 in (*) gives {f(0),f(−2)}={0,2},{f(2),f(0)}={−2,0}. From this, we deduce that f(0)=0, f(2)=−2, f(−2)=2. Next, we inductively prove that f(n)=−n. When ∣n∣=1,2, this is already proved. Suppose that we have proved this when ∣n∣≤n0 (n0≥2). Taking y=1 and x=n0,−n0 in (*) gives {f(−n0+1),f(f(n0)−1)}{f(n0+1),f(f(−n0)−1)}={n0−1,f(f(n0))+1},={−n0−1,f(f(−n0))+1} We deduce from this that f(n0+1)=−n0−1 and f(−n0−1)=n0+1. This completes the induction in Step 1. Step 2: Prove that f(x)=−x holds for all x∈R. First, as in the f(1)=1 case, we can prove that f is surjective. For a nonzero integer n and z∈R{0}, taking x=f(z)n, y=z, and taking x=z, y=f(z)n in (*), we obtain {−n−1,f(zf(f(z)n)−1)}{−n+1,f(zf(f(z)n)+1)}={nf(z)f(f(z))−1,zf(f(f(z)n))+1},={nf(z)f(f(z))+1,zf(f(f(z)n))−1}.(5) If for some z=z0∈R{0} we have f(f(z0))=f(z0), then by (5), we deduce that −n−1=zf(f(f(z)n))+1,−n+1=zf(f(f(z)n))−1. Taking the difference of the two equalities gives 2=−2. This is a contradiction! So f(f(z0))=−f(z0). But f is surjective. So f(x)=−x holds for all x∈R{0}. Moreover, f(0)=0 is known. This completes the proof of f(x)=−x when f(1)=−1. Combining the above two cases, the needed functions are f(x)=x (x∈R) and f(x)=−x (x∈R).
Source: MathNet,
licensed CC-BY-4.0.
Statement and solution reproduced as published; topic, difficulty and ordering added
by this site.