Fermat Might Like This One...

How many positive integers n > 1 n>1 evenly divide a 13 a a^{13}-a for all positive integers a ? a?


The answer is 31.

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.

5 solutions

Paola Ramírez
Jan 26, 2015

We have that p 2 p^2 ( p p is a prime), don't divide n n because p 2 p 13 p p^2 \nmid p^{13}-p

Then n n can be factoring in primes. As n a 13 a n|a^{13}-a for all a > 1 a>1 , n n has to divide at 2 13 2 = 2 × 3 2 × 5 × 7 × 13 2^{13}-2=2\times3^2\times5\times7\times13 .

Apply Fermat's Little Theorem such that a 13 a m o d p a^{13}\equiv a \mod p for ( p = 2 , 3 , 5 , 7 , 13 ) a 13 a m o d ( 2 × 3 × 5 × 7 × 13 ) (p=2,3,5,7,13) \therefore a^{13} \equiv a \mod (2\times3\times5\times7\times13) for all a a .

Then, n n that we search are × 3 × 5 × 7 × 13 \times3\times5\times7\times13 divisors except 1 1

Apply Tao and have 2 5 1 = 31 2^5-1=\boxed{31} that satisfy this.

I entered 32 and spent about 5 minutes thinking where I went wrong :').

Shourya Pandey - 5 years, 3 months ago

Log in to reply

@Shourya Pandey , Same here

Pratyush Dash - 1 year, 9 months ago

Shouldnt the number of divisors be (1+1) (2+1) (1+1) (1+1) (1+1)-1= 47 when a =2

And why cant we take a=3 ? we get 2^4 3 5 7 13 73, and the number of factors would be (4+1) (1+1) (1+1) (1+1) (1+1) (1+1) -1 ?

Murali Kancharla - 6 years, 4 months ago

Log in to reply

That's exactly what I did. But 3^2 can't be a factor. 3 is a factor.

Trevor Arashiro - 6 years, 4 months ago

Nice problem and solution. I guess when you use Fermat's Little Theorem you are using the generalization of the theorem, i.e., if p p is prime and m , n m,n are positive integers such that m n m o d ( p 1 ) m \equiv n \mod{(p - 1)} , then for every integer a a we have that a m a n m o d p a^{m} \equiv a^{n} \mod{p} .

Brian Charlesworth - 6 years, 4 months ago

Unless I have misunderstood the question, surely n {n} = a 13 a^{13} - a {a} is a solution (as well as n {n} = a {a} that is valid for an infinite number of natural numbers (except), such that there would be infinitely many solutions

Curtis Clement - 6 years, 4 months ago

Log in to reply

We're looking for integers n n that divide a 13 a a^{13} - a for all positive integers a > 1. a \gt 1. So essentially we're looking for the number of positive factors (other than 1 1 ) common to ( 2 13 2 ) , ( 3 13 3 ) , ( 4 13 4 ) , (2^{13} - 2), (3^{13} - 3), (4^{13} - 4), etc..

As Paola shows in her solution, these end up being all the factors of 2730 2730 other than 1 1 , of which there are 31 31 of them.

For the similar problem involving a 17 a a^{17} - a we would end up getting all the factors of 2 3 5 17 = 510 2*3*5*17 = 510 other than 1 1 , of which there are 15. 15.

Brian Charlesworth - 6 years, 4 months ago

Awwww, I thought it was 47, didn't realize that 3 2 3^2 couldn't be a factor. So I did 3 × 2 4 1 3\times2^4-1 rather than 2 5 1 2^5-1

Very nice problem however.

Trevor Arashiro - 6 years, 4 months ago

What about a=2? we get only 23 numbers that divide the expression evenly.. => with even quotient. Where in your logic have you mentioned about "Dividing Evenly"??

Ananya Aaniya - 5 years ago

Your solution is not justified....That's bcuz you have not shown that there can't exist any other prime other than the above mentioned......Not a proper problem.....Coz it has ambiguous wording..

Anubhav Mahapatra - 3 years, 8 months ago

The solution is great! But there is a need to show that 2, 3, 5, 7, and 13 are the only highest prime-power factors of a 13 a a^{13}-a , for all a > 1 a > 1 . You have shown that p 2 p 13 p p^2\nmid p^{13}-p , but it is quite possible that q 2 p 13 p q^2\mid p^{13}-p for p q p\neq q . For that, we can just show that 2 13 2 = 2 3 2 5 7 13 2^{13}-2 = 2\cdot 3^2\cdot5\cdot7\cdot13 and 3 13 3 = 2 4 3 5 7 13 73 3^{13}-3 = 2^4\cdot 3\cdot5\cdot 7\cdot 13\cdot 73 .

William Chau - 6 years, 4 months ago
Datu Oen
Nov 21, 2016

Alternative solution(?)

Let us consider a positive integer A = p x q y A = p^x q^y , where p p and q q are primes. Then the number of divisors of A is ( x + 1 ) ( y + 1 ) (x+1)(y+1) .
For example, we have 500. Note that 500 = 2 2 5 3 500 = 2^2 5^3 . So 100 has a total of ( 2 + 1 ) ( 3 + 1 ) = 12 (2+1)*(3+1) = 12 factors. These are 1 , 2 , 4 , 5 , 10 , 20 , 25 , 50 , 100 , 125 , 250 , 500 {1, 2, 4, 5, 10, 20, 25, 50, 100, 125, 250, 500} .
This set of number, however, includes 1 as a factor.

Now, a 13 a a^{13} - a can be factored as: a 13 a = a ( a 12 1 ) a^{13} - a = a ( a^{12} - 1)

Further factoring of ( a 12 1 ) (a^{12} - 1) leads to

( a 12 1 ) = ( ( a 4 ) 3 1 ) = ( a 4 1 ) ( a 8 + a 4 + 1 ) = ( a 1 ) ( a + 1 ) ( a 2 + 1 ) ( a 8 + a 4 + 1 ) (a^{12} - 1) = ((a^{4})^{3} -1 ) = (a^4 - 1 ) (a^8 + a^4 + 1) = (a-1)(a+1)(a^2+1) (a^8 + a^4 + 1)

Thus, the factors of a 13 a a^{13} - a are: a , ( a 1 ) , ( a + 1 ) , ( a 2 + 1 ) , ( a 8 + a 4 + 1 ) a, (a-1), (a+1), (a^2+1), (a^8 + a^4 + 1)

This leads to a total of ( 1 + 1 ) ( 1 + 1 ) ( 1 + 1 ) ( 1 + 1 ) ( 1 + 1 ) = 2 5 = 32 (1+1)(1+1)(1+1)(1+1)(1+1) = 2^5 = 32 factors, which includes 1. since we need only to consider factors n > 1 n > 1 , then we only have 32 1 = 32-1 = 31 factors.

Good ! Well explained

Subhadeep Paul - 3 years, 11 months ago

For odd a's all the factors you wrote for (a^12-1) are even, making some of them not prime. What do I miss here?

Laszlo Kocsis - 3 years, 8 months ago

The factor (a^8+a^4+1) can be further factored into (a^4+1-a^2) (a^2+1-a) (a^2+1+a). So please check the solution again or if I'm wrong pls clarify me the solution.

Ashwini Kumar - 3 years, 3 months ago

Great!!Helped a lot.

Sparrsh Nagdda - 2 years, 12 months ago
Nithish Kumar
Feb 12, 2015

In particular n divides 2¹³-2 and 3¹³-3. Decomposition of these 2 numbers in prime factors yields 2¹³-2 = 2 3² 5 7 13 and 3¹³-3 = 2^4 3 5 7 13 73

So n is a divisor of 2 3 5 7 13 (the smaller of the 2 exponents for each prime number is kept for the common divisor).

Now the problem is to determine if all of these 5 prime numbers always divide a¹³-a for all a > 1. a¹³-a = a(a¹²-1) = a[a^6-1][a^6+1] = a[(a³)²-1][(a²)³+1] = a[a³-1][a³+1][a²+1][(a²)²-(a²)+1] = a[a-1][a²+a+1][a+1][a²-a+1][a²+1][(a²)²-(a²)+1] = [a-1]a[a+1][a²-a+1][a²+1][a²+a+1][(a²)²-(a²)+1] (rearranged)

Among a[a+1], there is always a multiple of 2 Among [a-1]a[a+1], there is always a multiple of 3

Modulo 5, if a ≡ -2, -1, 0, 1, 2 respectively, then [a²+1], [a+1], a, [a-1], [a²+1] ≡ 0 respectively. So 5 is a common prime factor

Modulo 7, if a ≡ -3, -2, -1, 0, 1, 2, 3 respectively, then [a²+a+1], [a²-a+1], [a+1], a, [a-1], [a²+a+1], [a²-a+1] ≡ 0 respectively. So 7 is a common prime factor

Modulo 13, if a ≡ -6, -5, -4, -3, -2, -1, 0, 1, 2, 3, 4, 5, 6 respectively, then [(a²)²-(a²)+1], [a²+1], [a²+a+1], [a²-a+1], [(a²)²-(a²)+1], [a+1], a, [a-1], [(a²)²-(a²)+1], [a²+a+1], [a²-a+1], [a²+1], [(a²)²-(a²)+1] ≡ 0 respectively. So 13 is a common prime factor.

Therefore all divisors of 2 3 5 7 13 (except 1) are the required divisors. Nb of such divisors = 2^5 - 1 = 31.

First, to ease things out, we should factor that polynomial a 13 a = a ( a 12 1 ) = a ( a 6 + 1 ) ( a 6 1 ) = a ( a 6 + 1 ) ( a 3 + 1 ) ( a 3 1 ) a^{13} - a = a(a^{12} - 1) = a(a^6 + 1)(a^6 - 1) = a(a^6 + 1)(a^3 + 1)(a^3 - 1)

Then, notice that p 2 ∤ p 13 p p^2 \not| p^{13} - p , so no n n is a prime power greater than 1 1 .

We start looking at primes up to 13 13 for values of n n , using the factorization and Fermat's Little Theorem (and the corollary that p a p 1 1 p | a^{p - 1} - 1 ) :

2 a ( a 12 1 ) 2 | a(a^{12} - 1)

3 ( a 3 1 ) ( a ) ( a 3 + 1 ) 3 | (a^3 - 1)(a)(a^3 + 1) by Fermat and " n n divides one amongst n n consecutive numbers"

5 a ( a 12 1 ) 5 | a(a^{12} - 1) notice that a 12 = ( a 3 ) 4 a^{12} = (a^3)^{4}

7 a ( a 6 1 ) 7 | a(a^6 - 1)

11 11 does not

13 a 13 a 13 | a^{13} - a

And as we have 5 5 prime factors (to the first power), using combinatorics we get 2 5 1 = 31 2^5 - 1 = 31 divisors

Sean Tremba
Aug 2, 2018

a^13-a = a(a^12-1). By Fermat’s little theorem it is therefore divisible by 13. Substitute b = a^2. You get a^12-1 = b^6-1. This shows (again using Fermat’s little theorem) that this factor is divisible by 7. In similar fashion (using c=a^3, d=a^6) it can be shown that a^12-1 is divisible by 3 and 5.

The full expression a^13-a is also divisible by 2 for all a. This is because a^13 and a either are both even or both odd. In either case the difference is even.

It remains to be shown that 2,3,5,7 and 13 are the only prime factors for all values of a and that there is no higher power of any that divides the original expression for all a. To do so it is sufficient to find a countrrexample.

For the first part, let a=2. 2^13-2=8190. The prime factorization is 2 *3^2 * 5 * 7 * 13. This shows that ONLY these five primes are prime factors of a^13-a for all a.

This also shows that for all a, the only factor that could possibly appear as a power greater than the first for all a is 3. Next consider 3^13-3. Dividing by 3 gives 3^12-1, which is not divisible by 3. Therefore 3^13-3 is divisible by 3 but not 3^2.

This shows that a^13-a is divisible only by 2 3 5 7 13 and any of its factors. Any factor of this value may be expressed as the product of these five primes raised to an exponent of either 1 or 0. Five factors with 2 choices of exponents gives 2^5 = 32 possible values. The case where all exponents are zero yields 1, though, so there are 31 integers >1 that divide a^13-a for all integer values of a.

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...