Unique Prime Numbers

Find the sum of all primes a < 1000 , a<1000, such that a k + 1 = b m + 1 a^k+1=b^{m+1} for some positive integers k , b , m k,b,m .


The answer is 170.

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.

4 solutions

Zi Song Yeoh
May 20, 2014

a k + 1 = b m + 1 a k = b m + 1 1 a k = ( b 1 ) ( b m + b m 1 + . . . + 1 ) a^{k} + 1 = b^{m + 1} \Rightarrow a^{k} = b^{m + 1} - 1 \Rightarrow a^{k} = (b - 1)(b^{m} + b^{m - 1} + ... + 1) .
b 1 < b + 1 b m + b m 1 + . . . b + 1 b - 1 < b + 1 \le b^{m} + b^{m - 1} + ... b + 1 , since m 1 m \ge 1 . Note that b 1 b - 1 and b m + b m 1 + . . . b + 1 b^{m} + b^{m - 1} + ... b + 1 are both powers of a a , since a a is a prime. Therefore, b 1 b m + b m 1 + . . . b + 1 b - 1 \mid b^{m} + b^{m - 1} + ... b + 1 .

But 0 b m + b m 1 + . . . b + 1 ( b m 1 ) + ( b m 1 1 ) + . . . + ( b 1 ) + m + 1 ( m o d b 1 ) 0 \equiv b^{m} + b^{m - 1} + ... b + 1 \equiv (b^{m} - 1) + (b^{m - 1} - 1) + ... + (b - 1) + m + 1 \pmod{b - 1} . Since b k 1 = ( b 1 ) ( b k 1 + b k 2 + . . . b + 1 ) b^{k} - 1 = (b - 1)(b^{k - 1} + b^{k - 2} + ... b + 1) , so b 1 b k 1 b - 1 \mid b^{k} - 1 , for k N k \in \mathbb{N} .

( b m 1 ) + ( b m 1 1 ) + . . . + ( b 1 ) + m + 1 m + 1 ( m o d b 1 ) (b^{m} - 1) + (b^{m - 1} - 1) + ... + (b - 1) + m + 1 \equiv m + 1 \pmod{b - 1} , so b 1 m + 1 \boxed{b - 1 \mid m + 1} .

a) If m + 1 m + 1 is a prime,

I) If m = 1 m = 1 ,

a k = b 2 1 = ( b + 1 ) ( b 1 ) a^{k} = b^{2} - 1 = (b + 1)(b - 1) . Since ( b + 1 ) ( b 1 ) = 2 (b + 1) - (b - 1) = 2 , and both are powers of a a , so 2 = a j a k j = a k j ( a 2 j k 1 ) 2 = a^{j} - a^{k - j} = a^{k - j}(a^{2j - k} - 1) , where j k j \le k and b + 1 = a j b + 1 = a^{j} . Since both factors are positive integers, one of them is 1 1 and the other is 2 2 .

If a k j = 1 , a 2 j k 1 = 2 a^{k - j} = 1, a^{2j - k} - 1 = 2 , then k = j k = j , and consequently a k = 3 , a = 3 , k = 1 , m = 1 , b = 2 a^{k} = 3, a = 3, k = 1, m = 1, b = 2 . If a k j = 2 , a 2 j k 1 = 1 a^{k - j} = 2, a^{2j - k} - 1 = 1 , then a = 2 , k j = 1 a = 2, k - j = 1 , so k = j + 1 k = j + 1 . 2 j 1 1 = 1 2 j 1 = 2 j = 2 k = 3 2^{j - 1} - 1 = 1 \Rightarrow 2^{j - 1} = 2 \Rightarrow j = 2 \Rightarrow k = 3 , so a = 2 , k = 3 , m = 1 , b = 3 a = 2, k = 3, m = 1, b = 3 .

So, this gives two solutions for a a , namely a = 2 , 3 a = 2, 3 .

II) If m > 1 m > 1 ,

From b 1 m + 1 b - 1 \mid m + 1 , and that m + 1 m + 1 is a prime, we conclude that b = 2 b = 2 or b = m + 2 b = m + 2 .

i) b = 2 b = 2 ,

Then 2 m + 1 1 = a k 2^{m + 1} - 1 = a^{k} .

Suppose k > 1 k > 1 , then 2 m + 1 1 2^{m + 1} - 1 is a perfect prime power, and m + 1 m + 1 is a prime. But from the fact that the p-th Mersenne number is never a perfect prime power if p is a prime , we have a contradiction.

So, k = 1 k = 1 , and a = 2 m + 1 1 a = 2^{m + 1} - 1 , which is a Mersenne prime. We can check that if a a is a Mersenne prime less than 1000 1000 , m = l o g 2 ( a + 1 ) 1 m = log_{2}(a + 1) - 1 , b = 2, k = 1) works. (Note that n N n \in \mathbb{N} .)

ii) b = m + 2 b = m + 2

Then ( m + 2 ) m + 1 1 = ( m + 1 ) [ ( m + 2 ) m + ( m + 2 ) m 1 + . . . + 1 ] = a k (m + 2)^{m + 1} - 1 = (m + 1)[(m + 2)^{m} + (m + 2)^{m - 1} + ... + 1] = a^{k} .

Since m + 1 m + 1 is a prime and a power of a a , it follows that a = m + 1 a = m + 1 ,

( m + 2 ) m + 1 1 = ( m + 1 ) k (m + 2)^{m + 1} - 1 = (m + 1)^{k}

Consider the equation mod m + 2 m + 2 ,

( m + 2 ) m + 1 1 1 ( m o d m + 2 ) (m + 2)^{m + 1} - 1 \equiv -1 \pmod{m + 2} and ( m + 1 ) k ( 1 ) k ( m o d m + 2 ) (m + 1)^{k} \equiv (-1)^{k} \pmod{m + 2} , so ( 1 ) k 1 ( m o d m + 2 ) (-1)^{k} \equiv -1 \pmod{m + 2} . Note that 1 -1 is not congruent to 1 ( m o d m + 2 ) 1 \pmod{m + 2} , so k k is odd.

But ( m + 2 ) m + 1 = ( m + 1 ) k + 1 = ( m + 2 ) [ ( m + 1 ) k 1 ( m + 1 ) k 2 + . . . ( m + 1 ) + 1 ] (m + 2)^{m + 1} = (m + 1)^{k} + 1 = (m + 2)[(m + 1)^{k - 1} - (m + 1)^{k - 2} + ... - (m + 1) + 1] , since k is odd. So ( m + 1 ) k 1 ( m + 1 ) k 2 + . . . ( m + 1 ) + 1 (m + 1)^{k - 1} - (m + 1)^{k - 2} + ... - (m + 1) + 1 is divisible by m + 2 m + 2 .

But 0 ( m + 1 ) k 1 ( m + 1 ) k 2 + . . . ( m + 1 ) + 1 ( 1 ) k 1 ( 1 ) k 2 ( 1 ) + 1 ( m o d m + 2 ) 0 \equiv (m + 1)^{k - 1} - (m + 1)^{k - 2} + ... - (m + 1) + 1 \equiv (-1)^{k - 1} - (-1)^{k - 2} - (-1) + 1 \pmod{m + 2} But the last expression is congruent to k k mod m + 2 m + 2 , since k k is odd, so m + 2 k m + 2 \mid k , which means k k is even. (Since m + 2 m + 2 is even.), a contradiction!

b) If m + 1 m + 1 is composite,

Then m + 1 = p q m + 1 = pq , for some prime p p and some positive integer q q greater than 1 1 . Therefore, b m + 1 = ( b q ) p b^{m + 1} = (b^{q})^{p} is a perfect p-th power. But we have shown that there are no solutions when p p is a prime and k 1 , 3 k \ne 1, 3 .

When k = 3 k = 3 , then the only solution is a = 2 , k = 3 , p = 2 , b q = 3 a = 2, k = 3, p = 2, b^{q} = 3 , giving q = 1 q = 1 , a contradiction. When k = 1 k = 1 , then the only solution for b q b^{q} is 2 2 , which again gives q = 1 q = 1 , a contradiction.

Therefore, the only solutions for a a are 2 2 and the Mersenne primes less than 1000 1000 , giving 2 , 3 , 7 , 31 , 127 \boxed{2, 3, 7, 31, 127} and thus the answer is 2 + 3 + 7 + 31 + 127 = 170 2 + 3 + 7 + 31 + 127 = \boxed{170} .

This solution is long, but gives a complete description. The writeup could be tidied up slightly, by approaching the cases in a slightly different order. The approach I used, it to first deal with the case that b = 2 b = 2 , and then with the case that b > 2 b>2 and m + 1 m+1 is prime.

Ahmad remarks that "If k > 1 k > 1 , Mihăilescu's theorem states that the only possible solution for the equation a k + 1 = b m + 1 a^k + 1 = b^{m+1} is ( a , k , b , m ) = ( 2 , 3 , 3 , 1 ) (a,k,b,m) = (2,3,3,1) ". This is a much stronger statement than needed, and also requires a lot more machinery and high power tools to develop. It was conjectured in 1844 and only proved in 2002.

Calvin Lin Staff - 7 years ago
Ahmad Zaky
May 20, 2014

If k > 1 k > 1 , Mihăilescu's theorem states that the only possible solution for the equation a k + 1 = b m + 1 a^k + 1 = b^{m+1} is ( a , k , b , m ) = ( 2 , 3 , 3 , 1 ) (a,k,b,m) = (2,3,3,1) .

If k = 1 k = 1 , then we can rewrite the equation as a = b m + 1 1 = ( b 1 ) ( b m + b m 1 + + 1 ) a = b^{m+1}-1 = (b-1)(b^m+b^{m-1}+\dots+1) . Since a a is a prime number, we must have b 1 = 1 b-1 = 1 or b m + b m 1 + + 1 = 1 b^m+b^{m-1}+\dots+1 = 1 . The latter is impossible, so b b must be equal to 2 2 . Checking whether a = 2 m + 1 1 a = 2^{m+1}-1 is prime or not for m m from 1 1 to 8 8 (the largest m m for which 2 m + 1 < 1000 2^{m+1} < 1000 is 8 8 ), we get that the possible values for a a are 3 , 7 , 31 , 127 3, 7, 31, 127 .

So, the sum of all possible values of a a is 2 + 3 + 7 + 31 + 127 = 170 2 + 3 + 7 + 31 + 127 = 170 .

Calvin Lin Staff
May 13, 2014

We will distinguish two cases.

Case 1. b=2. Then a a is odd. Since m + 1 2 m+1\geq 2 , by looking at the equation a k + 1 = 2 m + 1 a^k+1=2^{m+1} modulo 4 4 we conclude that a 3 ( m o d 4 ) a\equiv 3 \pmod 4 and k k is odd. Since k k is odd, we get a k + 1 = ( a + 1 ) ( a k 1 . . . + 1 ) a^k+1=(a+1)(a^{k-1}-...+1) . So a + 1 a+1 is a power of 2 2 , say a + 1 = 2 s . a+1=2^s. This implies that a k + 1 = 1 + ( 2 s 1 ) k = 1 + ( 1 ) + k 2 s k ( k 1 ) 2 2 2 s + . . . k 2 s ( m o d 2 2 s ) a^k+1=1+ (2^s-1)^k= 1+(-1)+k2^s-\frac{k(k-1)}{2}2^{2s}+...\equiv k2^s \pmod {2^{2s}}

Because k k is odd, this implies that m + 1 = s m+1=s , which forces k k to be 1 1 . So the only possibility in this case is that a a is a prime of the form 2 s 1 2^s-1 . Because a < 1000 a<1000 , s 9 s\leq 9 . Checking powers of 2 2 , we get the following values for a a : 3 , 7 , 31 , 127 3,7,31,127 .

Case 2. b>2. By replacing b b with a suitable power, we will assume that m + 1 m+1 is prime. Rewrite the equation as a k = ( b 1 ) ( b m + . . . + b + 1 ) a^k=(b-1)(b^m+...+b+1) .

Note that 1 < b 1 < b m + . . . + 1 1<b-1<b^m+...+1 , so both b 1 b-1 and b m + . . . + 1 b^m+...+1 are divisible by the prime a a . Since b 1 0 ( m o d a ) , b-1\equiv 0 \pmod a, we have b 1 ( m o d a ) , b\equiv 1 \pmod a, so b m + . . . + 1 ( m + 1 ) ( m o d a ) . b^m+...+1\equiv (m+1) \pmod a. As we assumed that m + 1 m+1 is prime, we have m + 1 = a . m+1=a. So our equation becomes a k + 1 = b a a^k+1=b^a Moreover, b 1 b-1 is a power of a a , so b = a s + 1 b=a^s+1 for some s 1 s\geq 1 . Therefore b a 1 = ( 1 + a s ) a 1 = a a s + a ( a 1 ) 2 a 2 s + . . . b^a-1=(1+a^s)^a-1=a\cdot a^s + \frac{a(a-1)}{2}\cdot a^{2s}+... If s 2 , s\geq 2, this is congruent to a s + 1 a^{s+1} modulo a 2 s a^{2s} and is greater than a s + 1 a^{s+1} . So s = 1 s=1 , thus b = a + 1 b=a+1 . If a 3 , a\geq 3, then ( a + 1 ) a 1 = a 2 + a ( a 1 ) 2 a 2 + . . . (a+1)^a-1=a^2+\frac{a(a-1)}{2}a^2+... is congruent a 2 ( 1 + a ( a 1 ) 2 ) a^2(1+\frac{a(a-1)}{2}) modulo a 3 a^3 and is greater than that. So it cannot be a power of a a . Therefore, a = 2 a=2 , b = 3 b=3 . So we get 2 3 + 1 = 3 2 2^3+1=3^2 , which is a solution.

Putting this all together, we get the answer as 2 + 3 + 7 + 31 + 127 = 170. 2+3+7+31+127=170.

i found out the solution by substituting values for a,b,k and m.

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...