Olympiad Maths Prep

Track / Stage 7 / 93 of 300 #1493 of 2000

Problem 1493

National olympiad second round; IMO P1/P4
Number theory Difficulty 7.1 Find the answer

Let an integer k>1k > 1 be given. For each integer n>1n > 1, we put

f(n)=kn(11p1)(11p2)(11pr)f(n) = k \cdot n \cdot \left(1-\frac{1}{p_1}\right) \cdot \left(1-\frac{1}{p_2}\right) \cdots \left(1-\frac{1}{p_r}\right)

where p1,p2,,prp_1, p_2, \ldots, p_r are all distinct prime divisors of nn. Find all values kk for which the sequence {xm}\{x_m\} defined by x0=ax_0 = a and xm+1=f(xm),m=0,1,2,3,x_{m+1} = f(x_m), m = 0, 1, 2, 3, \ldots is bounded for all integers a>1a > 1.

Official solution

To solve the problem, we need to analyze the behavior of the sequence {xm}\{x_m\} defined by x0=ax_0 = a and xm+1=f(xm)x_{m+1} = f(x_m) for m=0,1,2,3,m = 0, 1, 2, 3, \ldots, where f(n)=kn(11p1)(11p2)(11pr)f(n) = k \cdot n \cdot \left(1-\frac{1}{p_1}\right) \cdot \left(1-\frac{1}{p_2}\right) \cdots \left(1-\frac{1}{p_r}\right) and p1,p2,,prp_1, p_2, \ldots, p_r are all distinct prime divisors of nn.

1. **Express f(n)f(n) in terms of Euler's Totient Function:**
f(n)=kn(11p1)(11p2)(11pr)=kϕ(n) f(n) = k \cdot n \cdot \left(1-\frac{1}{p_1}\right) \cdot \left(1-\frac{1}{p_2}\right) \cdots \left(1-\frac{1}{p_r}\right) = k \cdot \phi(n)
where ϕ(n)\phi(n) is Euler's Totient Function, which counts the number of integers up to nn that are coprime with nn.

2. **Behavior of ϕ(n)\phi(n):**
Euler's Totient Function ϕ(n)\phi(n) is always less than nn for n>1n > 1. Specifically, ϕ(n)=ni=1r(11pi)\phi(n) = n \prod_{i=1}^r \left(1 - \frac{1}{p_i}\right), where p1,p2,,prp_1, p_2, \ldots, p_r are the distinct prime divisors of nn.

3. Boundedness of the sequence:
We need to determine for which values of kk the sequence {xm}\{x_m\} is bounded for all integers a>1a > 1.

4. **Case k=2k = 2:**
f(n)=2ϕ(n) f(n) = 2 \cdot \phi(n)
For sufficiently large ii, all prime divisors of xix_i are less than or equal to the largest prime divisor of kk, which is 2. Thus, xix_i will eventually be a power of 2, and ϕ(2m)=2m(112)=2m1\phi(2^m) = 2^m \left(1 - \frac{1}{2}\right) = 2^{m-1}. Therefore, f(2m)=22m1=2mf(2^m) = 2 \cdot 2^{m-1} = 2^m, and the sequence becomes constant and hence bounded.

5. **Case k=3k = 3:**
f(n)=3ϕ(n) f(n) = 3 \cdot \phi(n)
For sufficiently large ii, all prime divisors of xix_i are less than or equal to the largest prime divisor of kk, which is 3. Thus, xix_i will eventually be of the form 2mi3li2^{m_i}3^{l_i}. For such xix_i, ϕ(2mi3li)=2mi13li1(21)(31)=2mi13li12=2mi3li1\phi(2^{m_i}3^{l_i}) = 2^{m_i-1}3^{l_i-1}(2-1)(3-1) = 2^{m_i-1}3^{l_i-1}2 = 2^{m_i}3^{l_i-1}. Therefore, f(2mi3li)=32mi3li1=2mi3lif(2^{m_i}3^{l_i}) = 3 \cdot 2^{m_i}3^{l_i-1} = 2^{m_i}3^{l_i}, and the sequence becomes constant and hence bounded.

6. **Case k>3k > 3:**
For k>3k > 3, consider the product of the terms:
kpp0(11p) k \prod_{p \le p_0} \left(1 - \frac{1}{p}\right)
where p0p_0 is the largest prime divisor of kk. If k>3k > 3, then:
kpp0(11p)>1 k \prod_{p \le p_0} \left(1 - \frac{1}{p}\right) > 1
This implies that f(n)=kϕ(n)f(n) = k \cdot \phi(n) will eventually increase xix_i rather than decrease it, making the sequence unbounded.

Conclusion:
The sequence {xm}\{x_m\} is bounded for all integers a>1a > 1 if and only if k=2k = 2 or k=3k = 3.

The final answer is k=2 \boxed{ k = 2 } or k=3k = 3.

Source: NuminaMath-1.5, licensed Apache-2.0. Statement and solution reproduced as published; topic, difficulty and ordering added by this site.