Maths Olympiad Prep

Library / /105 of 155

Algebra Difficulty 6.5 National olympiad Prove it Saudi Arabia

Let P(x)P(x) be a monic polynomial of degree 100100 with 100100 distinct noninteger real roots. Suppose that each of polynomials P(2x24x)P\left(2x^{2}-4x\right) and P(4x2x2)P\left(4x-2x^{2}\right) has exactly 130130 distinct real roots. Prove that there exist non constant polynomials A(x),B(x)A(x), B(x) such that A(x)B(x)=P(x)A(x)B(x)=P(x) and A(x)=B(x)A(x)=B(x) has no root in (1;1)(-1 ; 1).

Solution

Denote P(x)=(xa1)(xa2)(xa100)P(x) = (x - a_1)(x - a_2) \cdots (x - a_{100}) with a1,a2,,a100a_1, a_2, \ldots, a_{100} are real roots of polynomial P(x)P(x). So
P(2x24x)=02x24xai=0 for 1i100. P\left(2x^{2}-4x\right) = 0 \Leftrightarrow 2x^{2}-4x-a_i = 0 \text{ for } 1 \leq i \leq 100.
This equation cannot have 11 root since Δ=16+8ai0\Delta = 16 + 8a_i \neq 0. So each equation can have 00 or 22 roots. Note that P(2x24x)P\left(2x^{2}-4x\right) has 130130 roots so there are 1302=65\frac{130}{2} = 65 equations have 22 roots, which mean there are 6565 numbers ai>2a_i > -2 and 3535 numbers ai<2a_i < -2.

By the same way, we have
P(4x2x2)=02x24x+ai=0 P\left(4x-2x^{2}\right) = 0 \Leftrightarrow 2x^{2}-4x+a_i = 0
and Δ=168ai\Delta = 16 - 8a_i for any 1i1001 \leq i \leq 100. And there are 6565 numbers ai<2a_i < 2 and 3535 numbers ai>2a_i > 2.

By applying the principle of inclusion and exclusion, there are
65+65100=30 numbers ai(2;2). 65 + 65 - 100 = 30 \text{ numbers } a_i \in (-2 ; 2).

Suppose that a1<a2<<a35<2<a36<<a64<2<a65<<a100a_1 < a_2 < \ldots < a_{35} < -2 < a_{36} < \ldots < a_{64} < 2 < a_{65} < \ldots < a_{100}; and denote M,N,KM, N, K as the subsets of these numbers with the indices 135,3664,651001 \rightarrow 35, 36 \rightarrow 64, 65 \rightarrow 100.

Take A1(x)=mM(xm)A_1(x) = \prod_{m \in M}(x - m), B(x)=nN(xn)B(x) = \prod_{n \in N}(x - n), A2(x)=kK(xk)A_2(x) = \prod_{k \in K}(x - k) then we will prove that
A1(x0)A2(x0)>B(x0) |A_1(x_0) \cdot A_2(x_0)| > |B(x_0)|
for any number x0(1;1)x_0 \in (-1 ; 1).

Note that x0(1;1)\forall x_0 \in (-1 ; 1) then x0m>1,m<2|x_0 - m| > 1, \forall m < -2 and x0k>1,k>2|x_0 - k| > 1, \forall k > 2. We can suppose that x0>0x_0 > 0 and for any nNn \in N, we have two cases:

1. If n>0n > 0 then n(0;2)n \in (0 ; 2) and x0n<x02<x0k|x_0 - n| < |x_0 - 2| < |x_0 - k| with any k>2k > 2.
2. If n<0n < 0 then n(2;0)n \in (-2 ; 0) and x0n<x0(2)<x0m|x_0 - n| < |x_0 - (-2)| < |x_0 - m| with any m<2m < -2.

So in all cases of nn respect to the factor x0nx_0 - n in B(x0)B(x_0), we can choose factor from A1(x0)A_1(x_0) or A2(x0)A_2(x_0) with absolute value greater than it, and since M,K>N|M|, |K| > |N|, we always can do that. Hence A1(x0)A2(x0)>B(x0),x0(1;1)|A_1(x_0) \cdot A_2(x_0)| > |B(x_0)|, \forall x_0 \in (-1 ; 1) which mean this equation has no solution.

Therefore, we can choose A(x)=A1(x)A2(x)A(x) = A_1(x)A_2(x) and B(x)B(x) to satisfy the given condition. \square

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 and solution reproduced as published; topic and difficulty added by this site.