A Fresh Power Diophantine Equation

a n b n + 2 = ( a + b ) n 1 a^n-b^{n+2}=(a+b)^{n-1}

a a and b b are relatively prime positive integers and n ( > 1 ) n\, (>1) is an integer. Find all solutions to the equation above and enter your answer as ( a + b + n ) . \sum (a+b+n).


This is a generalization of the problem from 1997 Russian Olympiad:

For prime numbers p p and q , q, solve p 3 q 5 = ( p + q ) 2 . p^3-q^5=(p+q)^2.


The answer is 18.

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.

2 solutions

Kazem Sepehrinia
Apr 20, 2017

It is clear that a > b 1 a>b \ge 1 . Looking mod b b gives b a n 1 ( a 1 ) b| a^{n-1} (a-1) and since gcd ( a , b ) = 1 \gcd(a, b)=1 we get b a 1 b|a-1 and a = b c + 1 a=bc+1 for natural number c c . Looking mod a a gives a b 3 + 1 a|b^3+1 . Thus a b 3 + 1 a = b 3 + 1 ( b c + 1 ) = b ( b 2 c ) a|b^3+1-a=b^3+1-(bc+1)=b(b^2-c) and there is a non-negative integer d d such that b 2 c = d a = d ( b c + 1 ) b^2-c=da=d(bc+1) or b 2 c d b ( c + d ) = 0 b^2-cdb-(c+d)=0 Which is a quadratic in b b . Its discriminant must be a perfect square, but note that for c > 1 c>1 and d > 1 d>1 ( c d ) 2 Δ b = c 2 d 2 + 4 ( c + d ) ( c d + 2 ) 2 (cd)^2 \le \Delta_b=c^2d^2+4(c+d) \le (cd+2)^2 Thus we must have c 2 d 2 + 4 ( c + d ) = ( c d + 1 ) 2 c^2d^2+4(c+d)=(cd+1)^2 , which leads to c = 4 d 1 2 d 4 c=\frac{4d-1}{2d-4} Which is not an integer. This is a contradiction and so c = 1 c=1 or d 1 d \le 1 . Therefore we have three cases to check: c = 1 c=1 , d = 0 d=0 and d = 1 d=1 .

For c = 1 c=1 we get a = b + 1 a=b+1 and equation becomes ( b + 1 ) n = b n + 2 + ( 2 b + 1 ) n 1 ( 1 ) (b+1)^n=b^{n+2}+(2b+1)^{n-1} \qquad (1) From this equation observe that ( b + 1 ) n > b n + 2 (b+1)^n > b^{n+2} and ( b + 1 ) n > ( 2 b + 1 ) n 1 (b+1)^n > (2b+1)^{n-1} . Multiply these two inequalities to get ( b 2 + 2 b + 1 ) n > ( 2 b 2 + b ) n 1 b 3 (b^2+2b+1)^n > (2b^2+b)^{n-1} b^3 But for b 3 b \ge 3 we have ( b 2 + b + 1 ) n = ( b 2 + b + 1 ) n 1 ( b 2 + b + 1 ) < ( 2 b 2 + b ) n 1 b 3 (b^2+b+1)^n=(b^2+b+1)^{n-1} (b^2+b+1)<(2b^2+b)^{n-1} b^3 which is a contradiction and b 2 b \le 2 . For b = 1 b=1 from equation ( 1 ) (1) we get 2 n = 1 + 3 n 1 2^n=1+3^{n-1} and one can prove by induction that 1 + 3 n 1 > 2 n 1+3^{n-1}>2^n for n > 2 n>2 . n = 2 n=2 satisfies the equation and leads to a = 2 a=2 . We get a solution from here: ( a , b , n ) = ( 2 , 1 , 2 ) (a, b, n)=(2, 1, 2) . For b = 2 b=2 from equation ( 1 ) (1) we get 3 n = 2 n + 2 + 5 n 1 3^n=2^{n+2}+5^{n-1} and one can prove by induction that 3 n < 2 n + 2 + 5 n 1 3^n<2^{n+2}+5^{n-1} for n > 3 n>3 and for n 3 n \le 3 equality does not hold.

For d = 0 d=0 we have a = b 3 + 1 a=b^3+1 and original equation becomes ( b 3 + 1 ) n b n + 2 = ( b 3 + b + 1 ) n 1 ( 2 ) (b^3+1)^n-b^{n+2}=(b^3+b+1)^{n-1} \qquad (2) Suppose that p p is prime divisor of b 3 + b + 1 b^3+b+1 . Looking mod p p gives ( b ) n p b n + 2 ‎(-b)^n ‎‎\stackrel{‎p‎}{\equiv} b^{n+2} or ( 1 ) n p b 2 ‎(-1)^n ‎‎\stackrel{‎p‎}{\equiv} b^2 . For odd n n it follows that b 2 + 1 p 0 ‎b^2+1 ‎‎\stackrel{‎p‎}{\equiv} 0 , which is impossible considering that gcd ( b 2 + 1 , b 3 + b + 1 ) = 1 \gcd(b^2+1, b^3+b+1)=1 . Thus n n is even and p b 2 1 p| b^2-1 and p b 3 + b + 1 b ( b 2 1 ) = 2 b + 1 p b ( 2 b + 1 ) 2 ( b 2 1 ) = b + 2 p 2 ( b + 2 ) ( 2 b + 1 ) = 3 p|b^3+b+1-b(b^2-1)=2b+1 \\ p|b(2b+1)-2(b^2-1)=b+2 \\ p|2(b+2)-(2b+1)=3 So p = 3 p=3 is the only prime divisor of b 3 + b + 1 b^3+b+1 and b 3 + b + 1 = 3 k b^3+b+1=3^k for a natural number k k . Suppose that k > 1 k>1 . RHS of b 3 + b + 1 = 3 k b^3+b+1=3^k is divisible by 9 9 . Let us look when b 3 + b + 1 b^3+b+1 is divisible by 9 9 . \begin{array}{|*{10}{c|}} \hline b \pmod 9 & 0 & 1 & 2 & 3 & 4 & 5 & 6 & 7 & 8 \\ \hline b^3 \pmod 9 & 0 & 1 & 8 & 0 & 1 & 8 & 0 & 1 & 8 \\ \hline b^3+b+1 \pmod 9 & 1 & 3 & 2 & 4 & 6 & 5 & 7 & 0 & 8 \\ \hline \end{array} Thus we must have b 7 ( m o d 9 ) b \equiv 7 \pmod{9} . Now from equation ( 2 ) (2) looking mod 9 9 gives 2 n 7 n + 2 9 0 2^n-7^{n+2} ‎‎\stackrel{‎9‎}{\equiv} 0 , which is impossible. Thus k = 1 k=1 and b = 1 b=1 and 2 n 1 = 3 n 1 2^n-1=3^{n-1} , which gives the same solution of case c = 1 c=1 .

Final case is d = 1 d=1 , which gives a = b 2 b + 1 a=b^2-b+1 and equation becomes ( b 2 b + 1 ) n b n + 2 = ( b 2 + 1 ) n 1 ( 3 ) (b^2-b+1)^n-b^{n+2}=(b^2+1)^{n-1} \qquad (3) Suppose that p p is prime divisor of b 1 b-1 . Looking mod p p gives 0 p 2 n 1 ‎0 ‎‎\stackrel{‎p‎}{\equiv} 2^{n-1} and tells us that 2 2 is the only prime divisor of b 1 b-1 and b = 2 k + 1 b=2^k+1 for a natural number k k . Once again looking mod b 2 + 1 b^2+1 in equation ( 3 ) (3) gives ( b ) n b n + 2 ( m o d b 2 + 1 ) (-b)^n \equiv b^{n+2} \pmod{b^2+1} . For even n n it follows that b 2 1 0 ( m o d b 2 + 1 ) ‎b^2-1 \equiv 0 \pmod{b^2+1} , which is impossible and thus n n is odd. Finally look mod 2 k 2^k in equation ( 3 ) (3) ; It follows that 0 2 n 1 ( m o d 2 k ) 0 \equiv 2^{n-1} \pmod{2^k} , thus n 1 k n-1 \ge k . Now rearrange equation 3 3 in the following form ( b 2 b + 1 ) n b n = b n ( b 2 1 ) + ( b 2 + 1 ) n 1 ( 4 ) (b^2-b+1)^n-b^n=b^n(b^2-1)+(b^2+1)^{n-1} \qquad (4) Suppose k > 1 k>1 . Let us find power of 2 2 in the LHS of this equation using LTE lemma. Note that since n n is odd v 2 ( ( b 2 b + 1 ) n b n ) = v 2 ( b 2 b + 1 b ) + v 2 ( n ) + v 2 ( b 2 + 1 ) 1 = 2 k ( 5 ) ‎v_2 \left(‎‎(b^2‎-b‎+1)^n-b^n \right)‎=‎v_2(b^2-b+1-b)+v_2(n)+‎v_2(b^2+1)-1‎‎=2k \qquad (5) For finding power of 2 2 in the RHS of equation ( 5 ) (5) write v 2 ( b n ( b 2 1 ) + ( b 2 + 1 ) n 1 ) = min { v 2 ( b n ( b 2 1 ) ) , v 2 ( ( b 2 + 1 ) n 1 ) } = min { v 2 ( 2 k + 1 ( 2 k 1 + 1 ) ( 2 k + 1 ) n ) , v 2 ( 2 n 1 ( 2 2 k 1 + 2 k + 1 ) n 1 ) } = min { k + 1 , n 1 } = 1 2 ( k + n k n + 2 ) \displaystyle \begin{array}{c}‎ ‎v_2 \left(b^n(b^2-1)+(‎b^2‎+1)^{n-1} \right) &=‎\min \left\{ ‎‎v_2\left(b^n(b^2-1) \right), ‎‎v_2\left((‎b^2‎+1)^{n-1} \right)\right\} \\ &=\min \left\{‎‎v_2 \big( 2^{k+1} (2^{k-1}+1)‎(2^k+1)^n \big), ‎‎v_2\left(2^{n-1}(2^{2k-1}+2^k+1)^{n-1} \right)\right\} \\ &=\min \left\{k+1, n-1\right\} \\ &=\frac{1}{2}(k+n-|k-n+2|) \end{array} So from relation ( 5 ) (5) 2 k = 1 2 ( k + n k n + 2 ) ( 6 ) 2k=\frac{1}{2}(k+n-|k-n+2|) \qquad (6) But we proved n 1 k n-1 \ge k ; If k n = 1 k-n=-1 relation ( 6 ) (6) gives 2 k = k 2k=k , which is impossible and for k n 2 k-n \le -2 relation ( 6 ) (6) turns to 2 k = 1 2 ( k + n k n + 2 ) = 1 2 ( k + n + k n + 2 ) = k + 1 2k=\frac{1}{2}(k+n-|k-n+2|)=\frac{1}{2}(k+n+k-n+2)=‎k+1 Which gives k = 1 k=1 , but this is a contradiction, because we supposed k > 1 k>1 . So k = 1 k=1 and b = 3 b=3 . Putting these values in equation ( 3 ) (3) gives 7 n = 3 n + 2 + 1 0 n 1 7^n=3^{n+2}+10^{n-1} . One can prove using induction that 7 n < 3 n + 2 + 1 0 n 1 7^n<3^{n+2}+10^{n-1} for n > 6 n>6 . Trying values n 6 n \le 6 gives n = 3 n=3 as an answer. It leads to second solution: ( a , b , n ) = ( 7 , 3 , 3 ) (a, b, n)=(7, 3, 3) .

When b>1, it's quite easy to show a=7. Since b^(a-1) = 1 (mod a), then multiplying the formula by b^(a-n-3) and expressing it in base a gives b^(a-1) + b^(a-4) =0 (mod a).

This implies b^(a-4) = -1 (mod a) so b^(2a-8) = 1 (mod a). Which is true only if a-7 divides a.

Malcolm Rich - 4 years, 1 month ago

Log in to reply

How do you get b a 1 1 ( m o d a ) b^{a-1} \equiv 1 \pmod{a} ?

Kazem Sepehrinia - 4 years, 1 month ago

Log in to reply

It's one of Format's theorems.

Malcolm Rich - 4 years, 1 month ago

Log in to reply

@Malcolm Rich Notice that none of a a and b b are prime. Therefore b ϕ ( a ) 1 ( m o d a ) b^{\phi(a)} \equiv 1 \pmod{a} will not reduce to b a 1 1 ( m o d a ) b^{a-1} \equiv 1 \pmod{a} !

Kazem Sepehrinia - 4 years, 1 month ago
Yathish Dhavala
May 2, 2017

The solution is very lengthy. Here's the solution to the '97 Russian Olympiad problem: p 3 q 5 = ( p + q ) 2 p^{3} - q^{5} = (p + q)^{2}

First suppose neither p p nor q q equals 3. If they are congruent modulo 3, the left side is divisible by 3 while the right is not; if they are not congruent modulo 3, the right side is divisible by 3 while the left side is not, so there are no such solutions. If p = 3 p = 3 , we have q 5 < 27 q^{5} < 27 , which is impossible. Thus q = 3 q= 3 and p 3 243 = ( p + 3 ) 2 p^{3} - 243 = (p+ 3)^{2} , whose only integer root is p = 7 p = 7 .

For the generalised problem a n b n + 2 = ( a + b ) n 1 a^{n} - b^{n+2} = (a + b)^{n-1} , we can observe another trivial solution for n = 2 n = 2 which is ( 2 , 1 , 2 ) (2,1,2) . If we can prove that there exists no solution for n 4 n \geq 4 , then we have our answer here as ( 7 + 3 + 3 ) + ( 2 + 1 + 2 ) = 18 (7+3+3)+(2+1+2) = 18

But the problem tells us that a and b must be relatively prime. Is it proper to say that 2 is "relatively prime" to 1 -- or for that matter, any integer k>1 is "relatively prime" to 1?

Alex Silverman - 4 years, 1 month ago

Log in to reply

Yes, it's proper to say that. Indeed, any integer k>1 is relatively prime to 1. Because that's the way co-primes are defined: the two integers should have no positive common factors other than 1. The numbers 2 and 1 'properly' satisfy this definition.

Yathish Dhavala - 4 years, 1 month ago

It's lengthy, but not ugly. It's a beautiful solution, I promise! :)) I'd love to see simpler ones.

Kazem Sepehrinia - 4 years, 1 month ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...