Olympiad Maths Prep

Library / /29 of 33

Algebra Difficulty 7.5 National olympiad, round 2 Prove it Belarus

a) Determine all functions f:ZZf: \mathbb{Z} \to \mathbb{Z} such that
f(xf(y))=f(f(x))f(y)1 f(x - f(y)) = f(f(x)) - f(y) - 1
for all integers xx and yy.

b) The same question if
f(xf(y))=f(f(x))f(y)2 f(x - f(y)) = f(f(x)) - f(y) - 2
for all integers xx and yy.

Solution

a) f(x)=1f(x) = -1 or f(x)=x+1f(x) = x + 1.

(Alternative solution by I. Voronovich)
Setting y=f(x)y = f(x) in the given equation
f(xf(y))=f(f(x))f(y)1,(1) f(x - f(y)) = f(f(x)) - f(y) - 1, \qquad (1)
we obtain f(xf(f(x)))=1f(x - f(f(x))) = -1, i.e., there exists an integer λ\lambda such that f(λ)=1f(\lambda) = -1. Set y=λy = \lambda in (1), then
f(f(x))=f(x+1).(2) f(f(x)) = f(x + 1). \qquad (2)
Substituting f(x)+f(y)f(x) + f(y) for xx in (1), we obtain
f(f(x))=f(f(x)+f(y))f(y)1. f(f(x)) = f(f(x) + f(y)) - f(y) - 1.
Due to symmetry
f(f(y))=f(f(y)+f(x))f(x)1. f(f(y)) = f(f(y) + f(x)) - f(x) - 1.
Therefore f(f(x))f(x)=f(f(y))f(y)f(f(x)) - f(x) = f(f(y)) - f(y), so f(f(x))f(x)=cf(f(x)) - f(x) = c or, in view of (2),
f(x+1)f(x)=c,cZ(3) f(x+1) - f(x) = c, \quad c \in \mathbb{Z} \qquad (3)

If c=0c = 0, then f(x+1)=f(x)f(x+1) = f(x) for all integer xx, so f(x)f(x) is a constant function, thus from (1) it follows that c=1c = -1, i.e., f(x)=1f(x) = -1 for all integer xx.

If c>0c > 0 or c<0c < 0, then we conclude from (3) that f(x)f(x) is increasing or decreasing, respectively. Anyway, ff is injective, so (2) gives f(x)=x+1f(x) = x + 1.

Both the functions obviously satisfy the initial equation (1).

b) f(x)=2f(x) = -2 or f(x)=x+2f(x) = x + 2.

(Solution by A. Asanau.) First, in the same way as in a) we can easily establish for any f:ZZf: \mathbb{Z} \to \mathbb{Z} satisfying the given equation
f(xf(y))=f(f(x))f(y)2(1) f(x - f(y)) = f(f(x)) - f(y) - 2 \quad (1)
the following equalities:
f(xf(f(x)))=2,(2) f(x - f(f(x))) = -2, \quad (2)
f(f(x))=f(x+2),(3) f(f(x)) = f(x + 2), \quad (3)
f(x+2)f(x)=c,(4) f(x + 2) - f(x) = c, \quad (4)
where cc is a constant. In particular, (2) shows that 2E(f)-2 \in E(f), the range of ff. Now we have two cases.

I. c=0c = 0. Let f(0)=af(0) = a, f(1)=bf(1) = b. Then from (4) we have f(x+2)=f(x)f(x+2) = f(x), so f(x)=af(x) = a for all even xx and f(x)=bf(x) = b for all odd xx. Note that one of aa, bb equals 2-2 in view of (2). We claim that a=b=2a = b = -2. Suppose that aba \neq b. Then
f(x)=f(y)    xy(mod2). f(x) = f(y) \iff x \equiv y \pmod{2}.
Therefore (3) implies f(x)x+2x(mod2)f(x) \equiv x + 2 \equiv x \pmod{2}. In particular, f(0)0(mod2)f(0) \equiv 0 \pmod{2}, whence f(0)=a=2f(0) = a = -2.
Now we set x=0x = 0, y=1y = 1 in (1); then we get f(b)=f(2)b2=b4f(-b) = f(-2) - b - 2 = -b - 4. But f(b)f(-b) equals either bb or 2-2. In both cases we have b=2b = -2, contrary to aba \neq b.
So a=b=2a = b = -2, and f(x)=2f(x) = -2 for all xx. It is easy to see that this function satisfies the initial equation.

II. c0c \neq 0. Let again f(0)=af(0) = a, f(1)=bf(1) = b. Then from (4) we have
f(2n)=cn+a,f(2n+1)=cn+b.(5) f(2n) = cn + a, \quad f(2n + 1) = cn + b. \quad (5)
In particular, (5) implies that, on the one hand, f(x)f(x) \to \infty as xx \to \infty, and, on the other hand, limxf(x)x=c2\lim_{x \to \infty} \frac{f(x)}{x} = \frac{c}{2}. Then from (3) we have
f(f(x))f(x)=f(x+2)f(x)1 as x, \frac{f(f(x))}{f(x)} = \frac{f(x+2)}{f(x)} \to 1 \text{ as } x \to \infty,
but f(f(x))f(x)c2\frac{f(f(x))}{f(x)} \to \frac{c}{2}. Hence, c/2=1c/2 = 1, i.e. c=2c = 2. Then (5) becomes
f(x)=x+afor even x,f(x)=x+dfor odd x(where d=b1).(6) \begin{aligned} f(x) &= x + a & \text{for even } x, \\ f(x) &= x + d & \text{for odd } x \quad (\text{where } d = b - 1). \end{aligned} \quad (6)
1) Now, if ff is an injective function, then (3) immediately implies f(x)=x+2f(x) = x + 2, and this function satisfies (1).

2) Suppose that ff is not injective. But from (6) it easily follows that ff is injective on the set of even numbers and also injective on the set of odd numbers. Thus we see that some f(2n)f(2n) and f(2m+1)f(2m+1) are equal. From (6) it follows that 2n2m1=da2n - 2m - 1 = d - a, i.e., aa and dd are of different parity. Then (6) implies that all values of ff are of the same parity. Since 2E(f)-2 \in E(f), all the values of f(x)f(x) are even. Then from (3) we have f(x+2)=f(f(x))=f(x)+af(x+2) = f(f(x)) = f(x) + a. Thus (4) gives a=c=2a = c = 2. In particular, dd is odd. So, (6) becomes
f(x)=x+2for even x,f(x)=x+dfor odd x. \begin{aligned} f(x) &= x + 2 & \text{for even } x, \\ f(x) &= x + d & \text{for odd } x. \end{aligned}
One can verify that for any odd dd this function is a solution of (1).

Looking for a route rather than 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 and solution reproduced as published; topic and difficulty added by this site.