How Many Dividing Pairs?

For how many integer values of i i , 1 i 1000 1 \leq i \leq 1000 , does there exist an integer j j , 1 j 1000 1 \leq j \leq 1000 , such that i i is a divisor of 2 j 1 2^j - 1 ?


The answer is 500.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

7 solutions

Jubayer Nirjhor
Dec 2, 2013

Note that if i i is a divisor of 2 j 1 2^j - 1 , we have: 2 j 1 0 ( m o d i ) 2^j - 1 \equiv 0 \pmod{i}

Another thing to note: 2 j 1 1 1 ( m o d 2 ) 2^j-1 \equiv -1 \equiv 1 \pmod{2}

Hence, 2 j 1 2^j-1 is odd which implies i i to be odd...

When I saw an expression of the form a n 1 0 ( m o d N ) ~a^n-1\equiv 0 \pmod{N} , two things went on my mind: Fermat's Little Theorem and Euler's Theorem ..

Fermat's Little Theorem : If a a is an integer and p p is a prime, then...

a p 1 1 0 ( m o d p ) a^{p-1}-1 \equiv 0 \pmod{p}

We can use a = 2 a=2 and p = i p=i ... But this gives only the values of i i when i i is a prime... So we need a more generalized version...

Euler's Theorem : If n n and a a are positive integers and they are coprime, then...

a φ ( n ) 1 0 ( m o d n ) a^{\varphi(n)}-1\equiv 0 \pmod{n}

Now we have something useful... We can set a = 2 a=2 and n = i n=i and get...

2 φ ( i ) 1 ( m o d i ) 2^{\varphi(i)}-1\equiv \pmod{i}

And hence, we can set j = φ ( i ) j=\varphi(i) (Note that i i is odd and 2 2 is even and also a prime, and so they are obviously coprime)... By the definition of Euler's totient function and according to the question and our assumption, we have...

1 φ ( i ) = j i 1000 1\le \varphi(i)=j\le i\le 1000

So we're done... Don't forget that i i is odd... Hence for all such odd i i , there exists a j j , such that...

2 j 1 0 ( m o d i ) 2^j-1\equiv 0 \pmod{i}

All that's left is counting the number of odd integers in the interval [ 1 , 1000 ] [1,1000] which is simply...

1000 2 = 500 \dfrac{1000}{2}=\fbox{500}

Note : This solution could be shortened further to only 4-5 lines... But I tried to illustrate every step I took such that the beginners like me can easily understand... :) Thanks!

WTF!!! I thought that I wrote at least one solution without any typo... I wanna diiiiiiiiiiiiiiieeeeeeeeeeee...............

EDIIIIIIIIIIIIIIIIIIIIIT : 2 φ ( i ) 1 0 ( m o d i ) \Large{2^{\varphi(i)}-1\equiv 0 \pmod{i}}

Jubayer Nirjhor - 7 years, 6 months ago

Log in to reply

Calm down! :P

Sreejato Bhattacharya - 7 years, 6 months ago

Log in to reply

I wrote only 22 22 solutions so far... And all of them includes at least one typo... What the hell is the Nobel Prize Committee doing???

Jubayer Nirjhor - 7 years, 6 months ago

Is Fermat's Little Theorem given by you is true for all integer a a ?

sujoy roy - 7 years, 6 months ago

Log in to reply

Yes, but there's a condition (I should have mentioned)... Only if... p a \large{p \nmid a}

Jubayer Nirjhor - 7 years, 6 months ago
Daniel Chiu
Dec 1, 2013

First of all, note that i i must be odd, since otherwise we get that an even number divides an odd number. We claim that all odd values of i i work, and thus the answer is 500 \boxed{500} .

Consider the set { 2 1 , 2 2 , 2 3 , , 2 1000 } \{2^1,2^2,2^3,\cdots,2^{1000}\} . By the Pigeonhole principle, there must be two elements that have the same residue modulo i i , since i 999 i\le 999 and has at most 999 residue classes. Let these two elements be 2 a 2^a and 2 b 2^b , with a > b a>b . Then, since 2 and i i are coprime, 2 a 2 b ( m o d i ) 2 a b 1 ( m o d i ) 2^a\equiv 2^b\pmod i\implies 2^{a-b}\equiv 1\pmod i Since a 1000 a\le 1000 and b 1 b\ge 1 , a b 999 a-b\le 999 , and so j = a b j=a-b must satisfy i 2 j 1 i|2^j-1 and 1 j 1000 1\le j\le 1000 . We conclude that a j j exists for all 500 \boxed{500} odd i i .

Motivation: First, I eliminated even numbers, and then I suspected that all odd numbers, or maybe all but a couple, like 999, worked. The proof followed relatively quickly when I thought about the different possible remainders.

+1 for not depending on Euler's theorem.

Erick Wong - 7 years, 6 months ago

Great work!!

Dhruv Bhasin - 7 years, 6 months ago
Michael Tang
Dec 1, 2013

It is clear that i i must be odd; otherwise it is impossible for 2 j 1 , 2^{j}-1, an always odd number, to be a multiple of i . i. So, assume that i i is odd. We can rewrite the given equation using modular arithmetic to get 2 j 1 ( m o d i ) . 2^j \equiv 1 \pmod i. By Euler's Theorem, 2 ϕ ( i ) 1 ( m o d i ) , 2^{\phi(i)} \equiv 1 \pmod i, where ϕ ( i ) \phi(i) is the Euler function; then since ϕ ( i ) i \phi(i) \le i and i 1000 i \le 1000 for all i , i, we can choose j = ϕ ( i ) j = \phi(i) for all i , i, implying that all odd i i work. Thus, the possible i i are precisely the odd numbers in the range 1 , 2 , , 1000 , 1, 2, \ldots, 1000, of which there are 500 . \boxed{500}.

Anqi Li
Dec 1, 2013

Clearly, there are a lot of options for i i and j j so we really want to limit our options. When we see the word divisor , it is quite motivated to think about parity , in other words, odd and even factors. Also, we see 2 j 1 2^j - 1 which is a power , and we can change the problem to one regarding ( m o d i ) \pmod i , so it would be motivated to consider (the generalised form of Fermat's Little Theorem) Euler's theorem which, by the way, says that: For a a relatively prime to m m , we have

a ϕ ( m ) 1 ( m o d m ) a^\phi(m) \equiv 1 \pmod m .


So since i i is a divisor, it would be a neat idea to do some analysis of i i first. Since 2 j 1 2^j - 1 is odd, i i must be odd too (and not even).

So turning to our second idea, for odd i i , let j = ϕ ( i ) j = \phi(i) to get by Euler's theorem, 2 ϕ ( i ) 1 0 ( m o d i ) 2^\phi(i) - 1 \equiv 0 \pmod i . Here's the catch, notice that since i < 1000 i < 1000 , by definition, ϕ ( i ) < 1000 \phi(i) < 1000 , so for every odd i i we can find a matching j j .

Lastly, just remember that there are 1000 2 = 500 \frac{1000}{2} = \fbox{500} odd i i that is less than 1000 1000 .

  • Please correct my solution if there are any shortcomings in the solution.

2^j - 1 is always an odd number. By Fermat's Little Theorem, 2^j is congruent to 1 (modulo p). We can say that j = p - 1. 2^(p-1) is congruent to 1 modulo p. For any odd p, this condition satisfies. How about the composite odd numbers? By the theorem again, the gcd of 2 and any odd p is always 1. Hence, there 500 odd numbers that satisfies this condition.

how can u say that there are 500 odd nos. that satisfies this condition??

Vighnesh Raut - 7 years, 5 months ago
Cody Johnson
Dec 2, 2013

If i i is even, then n 2 j 1 n\nmid2^j-1 since 2 2 j 1 2\nmid2^j-1 . Otherwise, i i is odd and so gcd ( 2 , i ) = 1 \gcd(2,i)=1 , hence i 2 ϕ ( n ) 1 i\mid2^{\phi(n)}-1 . Since ϕ ( n ) n 1 \phi(n)\le n-1 , it holds for all odd i i , which are 500 500 i i 's in total on this interval.

I apologize. I put n n in place of i i in the second line (3 times).

Cody Johnson - 7 years, 6 months ago
Adit Mohan
Dec 4, 2013

by Fermat's little theorem a^{p-1} -1 is divisible by p if p and a are coprime. All odd numbers are coprime to 2,hence all odd numbers are divisible by 2^{j} -1

p p must be a prime... So Fermat's little theorem doesn't work here, I guess...

Jubayer Nirjhor - 7 years, 6 months ago

Log in to reply

no, a and p must be coprime .

Adit Mohan - 7 years, 6 months ago

Log in to reply

Well, 15 15 and 4 4 are coprime...

1 5 4 1 1 = 1 5 3 1 = 3375 1 = 3374 2 0 ( m o d 4 ) 15^{4-1} -1 = 15^3 - 1 = 3375 - 1 = 3374 \equiv 2 ≢ 0 \pmod{4}

You can read more here in Wikipedia and here in AoPS ... :)

Jubayer Nirjhor - 7 years, 6 months ago

Log in to reply

@Jubayer Nirjhor fair point . i guess i was lucky no such cases turned up in the question

Adit Mohan - 7 years, 6 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...