Showing posts with label Italian TST. Show all posts
Showing posts with label Italian TST. Show all posts

Friday, December 2, 2011

Italian IMO Team Selection Test 1990

Problem:
Let $P(x)=x^3-x+1$. Prove that, for every natural number $n > 1$, the numbers $n,P(n),P(P(n)),\ldots$ taken pairwise are relatively primes.

Solution:
Let $P_k(n)=\underbrace{P(P(\ldots P(n))))}_{k \ \textrm{times}}$. We prove by induction on $k \in \mathbb{N}^*$ that $P_k(n) \equiv 1 \pmod{P_i(n)}$ for every $0 \leq i \leq k-1$, where we set $P_0(n)=n$. In this way it's clear that taken two arbitrary numbers from the sequence, they will be relatively primes. If $k=1$ it is trivial, since $P(n)-n(n^2-1)=1$. Suppose that the statement is true for some $k$. Then, $$P_{k+1}(n)=P^3_{k}(n)-P_{k}(n)+1 \equiv 1 \pmod{P_{k}(n)}$$ and by induction hypothesis, $P_k(n) \equiv 1 \pmod{P_i(n)}$ for every $0 \leq i \leq k-1$, so $P_{k+1}(n) \equiv 1 \pmod{P_i(n)}$ for every $0 \leq i \leq k$ and the desired conclusion follows by the Principle of Mathematical Induction.

Thursday, December 1, 2011

Italian IMO Team Selection Test 1988

Problem:
Decompose the natural number $n$ into a sum of natural numbers so that the product of the summands is maximum.

Solution:
We go further and give explicitly the maximum product. Let $n=a_1+a_2+\ldots+a_m$ and $P=a_1a_2\cdots a_m$, where $a_i \in \mathbb{N}^*$ for every $1 \leq i \leq m$ . If $n=1,2$, clearly $P_{\max}=1$. Assume $n > 2$. To maximize $P$, any number $a_i$ can't be $1$ since $a_i + 1 > a_i \cdot 1$ for every $1 \leq i \leq m$ and we can replace the two summands $a_i$ and $1$ with $a_i + 1$. Likewise, if $a_i > 4$, we can replace $a_i$ with $(a_i-2)+2$ since $2(a_i-2) > a_i$. At last, we observe that it's indifferent to have $4$ or $2+2$, so $P$ can be maximized with $2$'s and $3$'s. Since $3^2 > 2^3$, we can have at most two summands equal to $2$ since we can replace $2+2+2$ with $3+3$. To be more precise, we have $$P_{max} = \left\{ \begin{array}{lll} 3^k & \textrm{if } n=3k \\ 2^2 \cdot 3^{k-1} & \textrm{if } n=3k+1 \\ 2\cdot 3^k & \textrm{if } n=3k+2 \end{array} \right.$$ where $k \in \mathbb{N}^*$.

Wednesday, November 30, 2011

Italian IMO Team Selection Test 1994 - Problem 2

Problem:
Find all prime numbers $p$ for which $\dfrac{2^{p-1}-1}{p}$ is a perfect square.

Solution:
Since $2^{p-1}-1$ is odd for all primes $p$, it's easy to see that $p \neq 2$. Moreover, by Fermat's Little Theorem we have $2^{p-1} \equiv 1 \pmod{p}$ for all primes $p \neq 2$, so $\dfrac{2^{p-1}-1}{p}$ is an integer and we want this integer to be a perfect square. Then, $$pn^2 = 2^{p-1} - 1, \quad n \in \mathbb{N}.$$ Since $p > 2$, $p-1$ is even and so we can write $$pn^2=(2^{\frac{p-1}{2}}-1)(2^{\frac{p-1}{2}}+1).$$ Both factor on the right hand side are odd and are relatively primes since their difference is $2$. This means that $p$ divides exactly one between the two factors and the other is a perfect square. If $p$ divides the first factor, we have $$2^{\frac{p-1}{2}}-1=pa^2, \qquad 2^{\frac{p-1}{2}}+1=b^2$$ where $a,b \in \mathbb{N}^*$. From the second equation we find $2^{\frac{p-1}{2}}=(b-1)(b+1)$ and these two factors are both powers of $2$ whose difference is $2$, so $b=3$ and $p=7$. If $p$ divides the second factor, we have $$2^{\frac{p-1}{2}}-1=a^2, \qquad 2^{\frac{p-1}{2}}+1=pb^2$$ and from the first equation,  if $p>3$ then $2^{\frac{p-1}{2}}-1=a^2 \equiv 3 \pmod{4}$, contradiction. So, it must be $p=3$ and for such value, $\dfrac{2^{3-1}-1}{3}=1$ which is a perfect square. In conclusion, the only prime numbers $p$ which satisfy the given condition are $p=3$ and $p=7$.