Maths Olympiad Prep

Library / /18 of 31

Algebra Difficulty 6.3 National Olympiad Prove it Czech Republic

Determine all functions f:RRf: \mathbb{R} \to \mathbb{R} such that, for all real numbers xx and yy,
f(x2+xy)=f(x)f(y)+yf(x)+xf(x+y). f(x^2 + xy) = f(x)f(y) + yf(x) + xf(x + y).

Solutions — 2

Solution 1

Setting x=0x = 0 yields f(0)=f(0)f(y)+yf(0)f(0) = f(0)f(y) + yf(0). If f(0)0f(0) \neq 0, we obtain 1=f(y)+y1 = f(y) + y or equivalently f(y)=1yf(y) = 1 - y for all yRy \in \mathbb{R}. Inserting this in the original equation yields
1x2xy=(1x)(1y)+y(1x)+x(1xy) 1 - x^2 - xy = (1-x)(1-y) + y(1-x) + x(1-x-y)
for x,yRx, y \in \mathbb{R}, which is true.
Therefore, we are left with f(0)=0f(0) = 0.
With the substitution x+y=zx + y = z, the functional equation is equivalent to
f(xz)=f(x)f(zx)+(zx)f(x)+xf(z).(1) f(xz) = f(x)f(z-x) + (z-x)f(x) + xf(z). \quad (1)
Exchanging xx and zz yields
f(zx)=f(z)f(xz)+(xz)f(z)+zf(x).(2) f(zx) = f(z)f(x-z) + (x-z)f(z) + zf(x). \quad (2)
Combining (1) and (2) yields
f(x)f(zx)+(zx)f(x)+xf(z)=f(z)f(xz)+(xz)f(z)+zf(x) f(x)f(z-x) + (z-x)f(x) + xf(z) = f(z)f(x-z) + (x-z)f(z) + zf(x)
or equivalently
f(x)(f(zx)x)=f(z)(f(xz)z).(3) f(x)(f(z-x) - x) = f(z)(f(x-z) - z). \quad (3)
Setting z=0z = 0 yields
f(x)(f(x)x)=f(0)f(x)=0, f(x)(f(-x) - x) = f(0)f(x) = 0,
so for each xRx \in \mathbb{R} we either have f(x)=0f(x) = 0 or f(x)=xf(-x) = x.
Assume that there is an x0x \neq 0 with f(x)=0f(x) = 0. For z0z \neq 0, we have f(xz){0,zx}f(x-z) \in \{0, z-x\}, so f(xz)zf(x-z) \neq z. Thus (3) implies f(z)=0f(z) = 0 for all z0z \neq 0. It is clear that the constant function f=0f = 0 is a solution.
Otherwise, we have f(x)0f(x) \neq 0 for all x0x \neq 0 and therefore f(x)=xf(x) = -x for all xx. This is also a solution.
We conclude that there are three solutions ff, namely f=0f = 0, f(x)=xf(x) = -x and f(x)=1xf(x) = 1 - x.

Solution 2

Setting x=0x = 0 gives f(0)=f(0)f(y)+yf(0)f(0) = f(0)f(y) + yf(0). Since f(y)=1yf(y) = 1 - y gives a solution, we remain with the case f(0)=0f(0) = 0. Setting x=1x = 1 gives
f(1+y)=f(1)f(y)+yf(1)+f(1+y),i. e.0=f(1)(f(y)+y). f(1+y) = f(1)f(y) + yf(1) + f(1+y), \quad \text{i. e.} \quad 0 = f(1)(f(y) + y).
Since f(y)=yf(y) = -y is a solution, we remain with the case f(1)=0f(1) = 0.

Now suppose f(0)=f(1)=0f(0) = f(1) = 0. Setting y=xy = -x gives 0=f(x)f(x)xf(x)0 = f(x)f(-x) - xf(x), that is
f(x)=0f(x)=x.(4) f(x) = 0 \quad \lor \quad f(-x) = x. \tag{4}
Setting y=1xy = 1 - x gives f(x)=f(x)f(1x)+(1x)f(x)f(x) = f(x)f(1-x) + (1-x)f(x), that is
f(x)=0f(1x)=x.(5) f(x) = 0 \quad \lor \quad f(1-x) = x. \tag{5}
If f(x)0f(x) \neq 0 for some x0x \neq 0, we have f(x)=x0f(-x) = x \neq 0 and f(1x)=x0f(1-x) = x \neq 0 and applying (4) and (5) to x-x and 1x1-x instead of xx we get f(x)=x=1xf(x) = -x = 1-x, a contradiction. Hence f(x)=0f(x) = 0 for all xRx \in \mathbb{R}.

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 reproduced verbatim; metadata (topic, difficulty) added by this project.