EGMO 2012

Let ( a 1 , b 1 ) (a_1,b_1) , ( a 2 , b 2 ) (a_2,b_2) , . . . ... , ( a n , b n ) (a_n,b_n) be all the pairs of prime numbers that satisfy a a + 1 + b + 1 b = 2 n n + 2 \frac{a}{a+1}+\frac{b+1}{b}=\frac{2n}{n+2} for some positive integer n n . Then find the sum of all possible values of b a b-a .


The answer is 10.

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.

3 solutions

Kazem Sepehrinia
Jun 3, 2017

a a + 1 + b + 1 b = 2 n n + 2 1 1 a + 1 + 1 + 1 b = 2 4 n + 2 \frac{a}{a+1}+\frac{b+1}{b}=\frac{2n}{n+2} \ \ \Longrightarrow \ \ 1-\frac{1}{a+1}+1+\frac{1}{b}=2-\frac{4}{n+2} Or 1 a + 1 1 b = 4 n + 2 b a 1 b ( a + 1 ) = 4 n + 2 \frac{1}{a+1}-\frac{1}{b}=\frac{4}{n+2} \ \ \Longrightarrow \ \ \frac{b-a-1}{b(a+1)}=\frac{4}{n+2} And finally ( n + 2 ) ( b a 1 ) = 4 b ( a + 1 ) (n+2)(b-a-1)=4b(a+1) Notice that since b > a + 1 b > a+1 and a , a, b b are prime numbers, thus, gcd ( b , a + 1 ) = 1 \gcd(b, a+1)=1 . It follows that b ( a + 1 ) b-(a+1) is co-prime to both b b and a + 1 a+1 . Hence b a 1 4 b-a-1 \mid 4 , b a 1 = 1 , 2 , 4 b-a-1=1, 2, 4 and b a = 2 , 3 , 5 b-a=2, 3, 5 . Notice that n = 4 b a 1 b ( a + 1 ) 2 n=\frac{4}{b-a-1} b(a+1)-2 is a positive integer.

mind blowing solution @Kazem Sepehrinia I actually found all the solutions of the equation.

First I divided the question in two parts where

Case I : a = 2 a=2 which gives the solution as ( a , b , n ) = ( 2 , 5 , 28 ) , ( 2 , 7 , 19 ) (a,b,n)=(2,5,28),(2,7,19)

Case II : a , b a,b are odd primes. ( a , b , n ) = ( p , p + 2 , 2 ( 2 p 2 + 6 p + 3 ) ) (a,b,n)=(p,p+2,2(2p^{2}+6p+3)) where p , p + 2 p,p+2 both are prime numbers.

For finding the solution of the case II, I wrote b = a + m b=a+m where m m is an even natural number. (since a , b a,b are odd primes.)

Then I solved for n n in terms of a , m a,m .

n = 2 ( 2 a 2 + 2 a m + 2 a + m + 1 ) m 1 n=\frac{2(2a^{2}+2am+2a+m+1)}{m-1} = 4 ( a + 1 ) 2 m 1 + ( 4 a + 2 ) =\frac{4(a+1)^{2}}{m-1}+(4a+2)

(Now see that ( 4 ( a + 1 ) 2 ) = e v e n , ( m 1 ) = o d d (4(a+1)^{2})=even,(m-1)=odd .

So unless m = 2 m=2

4 ( a + 1 ) 2 m 1 = n o n i n t e g e r \frac{4(a+1)^{2}}{m-1}=non-integer and hence forth n = n o n i n t e g e r n=non-integer ;which is false .

Therfore m = 2 m=2 and we are done .

For case I after putting a = 2 a=2 we get b = 3 ( n + 2 ) n 10 b=\frac{3(n+2)}{n-10} .

b = 3 + 36 n 10 b=3+\frac{36}{n-10} since b b is an integer we get n = ( 11 , 12 , 14 , 16 , 13 , 19 , 22 , 28 , 46 ) n=(11,12,14,16,13,19,22,28,46) .

But b b is prime . therefore n = 19 , 28 n=19,28

Shivam Jadhav - 4 years ago

Why does b > a + 1 b > a + 1 imply that gcd ( b , a + 1 ) = 1 \gcd(b, a + 1 ) = 1 ? Can't we have b = 10 , a = 4 b = 10, a = 4 which gives us gcd ( 10 , 5 ) = 5 \gcd (10, 5 ) = 5 ?

Edit: I forgot that they had to be prime numbers. It would be helpful to add a short explanation.

Chung Kevin - 4 years ago

Log in to reply

Thanks. I edited the solution to address this point.

Kazem Sepehrinia - 4 years ago
Shivam Jadhav
Jun 3, 2017

First I divided the question in two parts where

Case I : a = 2 a=2 which gives the solution as ( a , b , n ) = ( 2 , 5 , 28 ) , ( 2 , 7 , 19 ) (a,b,n)=(2,5,28),(2,7,19)

Case II : a , b a,b are odd primes. ( a , b , n ) = ( p , p + 2 , 2 ( 2 p 2 + 6 p + 3 ) ) (a,b,n)=(p,p+2,2(2p^{2}+6p+3)) where p , p + 2 p,p+2 both are prime numbers.

For finding the solution of the case II, I wrote b = a + m b=a+m where m m is an even natural number. (since a , b a,b are odd primes and b > a + 1 b>a+1 .)

Then I solved for n n in terms of a , m a,m .

n = 2 ( 2 a 2 + 2 a m + 2 a + m + 1 ) m 1 n=\frac{2(2a^{2}+2am+2a+m+1)}{m-1} = 4 ( a + 1 ) 2 m 1 + ( 4 a + 2 ) =\frac{4(a+1)^{2}}{m-1}+(4a+2)

(Now see that ( 4 ( a + 1 ) 2 ) = e v e n , ( m 1 ) = o d d (4(a+1)^{2})=even,(m-1)=odd .

So unless m = 2 m=2

4 ( a + 1 ) 2 m 1 = n o n i n t e g e r \frac{4(a+1)^{2}}{m-1}=non-integer and hence forth n = n o n i n t e g e r n=non-integer ;which is false .

Therfore m = 2 m=2 and we are done .

For case I after putting a = 2 a=2 we get b = 3 ( n + 2 ) n 10 = 3 + 36 n 10 b=\frac{3(n+2)}{n-10}=3+\frac{36}{n-10} .

Since b b is an integer we get n = ( 11 , 12 , 14 , 16 , 13 , 19 , 22 , 28 , 46 ) n=(11,12,14,16,13,19,22,28,46) .

But b b is prime .

Therefore n = 19 , 28 n=19,28

4 ( a + 1 ) 2 4(a+1)^2 being even and m 1 m-1 odd does not imply that m = 2 m=2 .

Kazem Sepehrinia - 4 years ago

Log in to reply

you are right this means that these are not all the solutions. But can we find other solutions. Any idea.

Shivam Jadhav - 4 years ago

Log in to reply

Regarding my solution, all of the solutions are ( a , b , n ) = ( p , p + 2 , 4 ( p + 1 ) ( p + 2 ) ) , (a, b, n)=(p, p+2, 4(p+1)(p+2)), where p p and p + 2 p+2 are prime ( a , b , n ) = ( p , p + 3 , 2 ( p + 1 ) ( p + 3 ) ) , (a, b, n)=(p, p+3, 2(p+1)(p+3)), where p p and p + 3 p+3 are prime ( a , b , n ) = ( p , p + 5 , ( p + 1 ) ( p + 5 ) ) , (a, b, n)=(p, p+5, (p+1)(p+5)), where p p and p + 5 p+5 are prime.

Kazem Sepehrinia - 4 years ago

Log in to reply

@Kazem Sepehrinia so i need to use your technique to prove that no other solutions exist

Shivam Jadhav - 4 years ago

@Kazem Sepehrinia try my new question very nice you will like it .

Shivam Jadhav - 4 years ago
Kelvin Hong
Jun 5, 2017

We can rewriting equation to

b a 1 b ( a + 1 ) = 4 n + 2 \frac{b-a-1}{b(a+1)}=\frac{4}{n+2}

So it leads to two conditions

Case I:

{ k ( b a 1 ) = 4 ——-(1) k b ( a + 1 ) = n + 2 ——-(2) \begin{cases} k(b-a-1)=4 \text{-------(1)}\\ kb(a+1)=n+2 \text{-------(2)}\\ \end{cases}

For some positive integer k k , because R H S RHS of ( 2 ) (2) must be positive.

So k = 1 , 2 , 4 k=1,2,4 ,

When k = 1 k=1 , b a = 5 b-a=5 , solving this we get a only solution which is ( 2 , 7 , 19 ) (2,7,19) proves b a = 5 b-a=5 is valid.

When k = 2 k=2 , b a = 3 b-a=3 , solving this we can easily figure out a solution ( 2 , 5 , 28 ) (2,5,28) proves b a = 3 b-a=3 is valid.

When k = 4 k=4 , b a = 2 b-a=2 , solving this we can easily figure out a solution ( 3 , 5 , 78 ) (3,5,78) proves b a = 2 b-a=2 is valid.

Case II:

{ ( b a 1 ) = 4 k ——-(3) b ( a + 1 ) = k ( n + 2 ) ——-(4) \begin{cases} (b-a-1)=4k \text{-------(3)}\\ b(a+1)=k(n+2) \text{-------(4)}\\ \end{cases}

For some positive integer k k .

When k = 1 k=1 , b a = 5 b-a=5 , don't need to prove this.

When k = 2 k=2 , b a = 9 b-a=9 , we get a = 2 , b = 11 a=2 , b=11 but b ( a + 1 ) = 33 2 ( n + 1 ) b(a+1)=33 \neq 2(n+1) so b a = 9 b-a=9 invalid.

When k = 4 k=4 , b a = 17 b-a=17 , we get a = 2 , b = 19 a=2 , b=19 but b ( a + 1 ) = 19 × 3 4 ( n + 1 ) b(a+1)=19 \times 3 \neq 4(n+1) so b a = 17 b-a=17 invalid.

So sum of the values of ( b a ) (b-a) be 5 + 3 + 2 = 10 5+3+2 = \boxed {10}

Feel free to tell me if I have wrong.

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...