Maths Olympiad Prep

Track / Stage 7 / 48 of 300 #2088 of 2604

Problem 2088

National Olympiad second round; IMO P1/P4
Algebra Difficulty 7.1 Prove it Iranian Mathematical Olympiad · Iran

For nonnegative integers mm and nn, the sequence a(m,n)a(m, n) of real numbers is defined as follows: a(0,0)a(0, 0) is equal to 22, and for each natural number nn, a(0,n)=1a(0, n) = 1 and a(n,0)=2a(n, 0) = 2. Also for m,n∈Nm, n \in \mathbb{N}:
a(m,n)=a(m−1,n)+a(m,n−1) a(m, n) = a(m - 1, n) + a(m, n - 1)
Prove that for each natural number kk, all roots of the polynomial Pk(x)=∑i=0ka(i,2k+1−2i)xiP_k(x) = \sum_{i=0}^{k} a(i, 2k + 1 - 2i)x^i are real numbers.

This one wants a proof. Work it on paper, then read the official solution and mark yourself. Be honest about it: the record is only any use to you if it is.

Next problem →

Official solution

Let Qk(x)=∑i=0ka(i,2k−2i)xiQ_k(x) = \sum_{i=0}^{k} a(i, 2k - 2i)x^i. According to the recurrence relation for a(m,n)a(m, n), we get
Pk(x)=xPk−1(x)+Qk(x)Qk(x)=xQk−1(x)+Pk−1(x). \begin{align*} P_k(x) &= xP_{k-1}(x) + Q_k(x) \\ Q_k(x) &= xQ_{k-1}(x) + P_{k-1}(x). \end{align*}
So Qk(x)=Pk(x)−xPk−1(x)Q_k(x) = P_k(x) - xP_{k-1}(x) and therefore, Pk(x)−xPk−1(x)=x(Pk−1(x)−xPk−1(x))+Pk−1(x)P_k(x) - xP_{k-1}(x) = x(P_{k-1}(x) - xP_{k-1}(x)) + P_{k-1}(x). Finally, we get
Pk(x)=(2x+1)Pk−1(x)−x2Pk−1(x). P_k(x) = (2x + 1)P_{k-1}(x) - x^2 P_{k-1}(x).
Hence we get a recurrence relation for Pk+1(x)P_{k+1}(x) where k≥2k \ge 2, P1(x)=3x+1P_1(x) = 3x + 1 and P2(x)=5x2+5x+1P_2(x) = 5x^2 + 5x + 1.

Claim. For each positive integer k≥2k \ge 2 all of the roots of Pk(x)P_k(x) and Pk−1(x)P_{k-1}(x) are real and distinct. Furthermore, if a1<a2<⋯<ak−1a_1 < a_2 < \cdots < a_{k-1} and b1<b2<⋯<bkb_1 < b_2 < \cdots < b_k are roots of Pk−1P_{k-1} and PkP_k, respectively, we have
b1<a1<b2<a2<⋯<ak−1<bk. b_1 < a_1 < b_2 < a_2 < \cdots < a_{k-1} < b_k.
Proof. We proceed by induction. For the base case k=2k = 2, −13-\frac{1}{3} is the only root of P1P_1 and P2(−13)<0P_2(-\frac{1}{3}) < 0 so −13-\frac{1}{3} lies between the two roots of P2(x)P_2(x).
Suppose that Pk−1(x)P_{k-1}(x) and Pk(x)P_k(x) satisfy the induction hypothesis. We know Pk+1(x)=(2x+1)Pk(x)−x2Pk−1(x)P_{k+1}(x) = (2x+1)P_k(x) - x^2P_{k-1}(x). Now, we consider the signs of PkP_k and Pk−1P_{k-1} on different real numbers. First suppose that kk is even.
x−∞b1a1b2a2⋯bk−1ak−1bk+∞Pk+0−0+⋯0−0+x2Pk−1−−0+0⋯−0++ \begin{array}{c|cccccccccc} x & -\infty & b_1 & a_1 & b_2 & a_2 & \cdots & b_{k-1} & a_{k-1} & b_k & +\infty \\ \hline P_k & + & 0 & - & 0 & + & \cdots & 0 & - & 0 & + \\ x^2 P_{k-1} & - & - & 0 & + & 0 & \cdots & - & 0 & + & + \end{array}
Since Pk+1(x)=(2x+1)Pk(x)−x2Pk−1(x)P_{k+1}(x) = (2x+1)P_k(x) - x^2P_{k-1}(x), for the sign of Pk+1P_{k+1} in respective bib_i's we have
x−∞b1a1b2a2⋯bk−1ak−1bk+∞Pk+1−+−−−⋯+−−+ \begin{array}{c|cccccccccc} x & -\infty & b_1 & a_1 & b_2 & a_2 & \cdots & b_{k-1} & a_{k-1} & b_k & +\infty \\ \hline P_{k+1} & - & + & - & - & - & \cdots & + & - & - & + \end{array}
According to the change of signs of Pk+1(x)P_{k+1}(x) and by the Mean Value Theorem, for each 1≤i≤k−11 \le i \le k-1, Pk+1P_{k+1} has a root between bib_i and bi+1b_{i+1}. Also, it has one root less than b1b_1 and one root greater than bkb_k, which completes the proof for even values of kk. The case where kk is an odd number is similar, the only difference being the signs of b1b_1 and −∞-\infty.

Note that
(i) 00 is not a root of any of the PkP_k's. Because Pk(0)=a(0,2k+1)=1P_k(0) = a(0, 2k+1) = 1, Pk−1(x)P_{k-1}(x) and x2Pk−1(x)x^2P_{k-1}(x) have the same sign.
(ii) The coefficient of xkx^k in Pk(x)P_k(x) equals a(2k,1)>0a(2k, 1) > 0, so Pk(+∞)>0P_k(+\infty) > 0 and the sign of Pk(−∞)P_k(-\infty) is related to the parity of kk. □\square

Source: MathNet, licensed CC-BY-4.0. Statement reproduced verbatim; metadata (topic, difficulty, ordering) added by this project.