200 Followers Problem!

r 2 s + 200 = s 2 r + 20 0 2 \frac{r^2}{s}+200=\frac{s^2}{r}+200^2

In the equation above, r r and s s are both rational numbers and their difference is an integer.

What is the number of solutions to the equation?


The answer is 7.

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
Jun 21, 2017

Let A = 20 0 2 200 A=200^2-200 . Thus r 3 s 3 = A r s r^3-s^3=Ars . Write r s = n r-s=n , where n Z n \in \mathbb{Z} . Then ( s + n ) 3 s 3 = A ( s + n ) s (s+n)^3-s^3=A(s+n)s , or ( A 3 n ) s 2 + ( A n 3 n 2 ) s n 3 = 0 (A-3n)s^2+(An-3n^2)s-n^3=0 According to quadratic formula, we get s = n 2 ± n 2 ( A + n ) ( A 3 n ) A 3 n s=-\frac{n}{2}\pm \frac{|n|}{2} \frac{\sqrt{(A+n)(A-3n)}}{A-3n} Therefore, ( A + n ) ( A 3 n ) = k 2 (A+n)(A-3n)=k^2 , where k k is a non-negative integer. It follows that A n A / 3 -A \le n \le \lfloor A/3 \rfloor .

Note that A A is divisible by 200 200 . In the equation ( A + n ) ( A 3 n ) = k 2 (A+n)(A-3n)=k^2 look mod 5 5 , then 3 n 2 k 2 ( m o d 5 ) -3n^2 \equiv k^2 \pmod{5} . Hence n k 0 ( m o d 5 ) n \equiv k \equiv 0 \pmod{5} , since a perfect square mod 5 5 is congruent to 0 , 1 , 4 0, 1, 4 only. If you divide both sides of the equation by 25 25 and look mod 5 5 again, then it comes out that n k 0 ( m o d 25 ) n \equiv k \equiv 0 \pmod{25} .

In the equation ( A + n ) ( A 3 n ) = k 2 (A+n)(A-3n)=k^2 look mod 8 8 , then 3 n 2 k 2 ( m o d 8 ) -3n^2 \equiv k^2 \pmod{8} . Hence n n and k k are both even, since a perfect square mod 8 8 is congruent to 0 , 1 , 4 0, 1, 4 only. If you divide both sides of the equation by 4 4 and look mod 8 8 again, then a similar argument gives n k 0 ( m o d 4 ) n \equiv k \equiv 0 \pmod{4} . Now write n = 4 n n=4n' and k = 4 k k=4k' and divide both sides of the equation by 16 16 to get ( A / 4 + n ) ( A / 4 3 n ) = k 2 (A/4+n')(A/4-3n')=k'^2 . Look mod 8 8 , then 4 + 4 n + 5 n 2 k 2 ( m o d 8 ) 4+4n'+5n'^2 \equiv k'^2 \pmod{8} . If n n' is odd, then LHS of this congruence is 5 5 mod 8 8 , which is contradiction. Thus n n' and k k' are both even and finally n k 0 ( m o d 8 ) n \equiv k \equiv 0 \pmod{8} .

Thus in total we have that n k 0 ( m o d 200 ) n \equiv k \equiv 0 \pmod{200} . Write n = 200 N n=200 N and k = 200 K k=200 K to get ( 199 + N ) ( 199 3 N ) = K 2 (199+N)(199-3N)=K^2 For N = 199 \color{#D61F06}N=-199 and N = 0 \color{#D61F06}N=0 the LHS becomes a square. Then, for N ≢ 0 ( m o d 199 ) N \not\equiv 0 \pmod {199} we have that d = gcd ( 199 + N , 199 3 N ) 3 ( 199 + N ) + 199 3 n = 4 × 199 d=\gcd(199+N, 199-3N) \mid 3(199+N)+199-3n=4\times 199 Hence d = 1 , 2 , 4 d=1, 2, 4 .

If d = 1 d=1 , then 199 + N = a 2 199+N=a^2 and 199 3 N = b 2 199-3N=b^2 , where a a and b b are co-prime odd positive integers. Hence 3 a 2 + b 2 = 796 3a^2+b^2=796 . Its easy to see that a 15 a \le 15 . Checking a = 1 , 3 , 5 , . . . , 15 a=1, 3, 5, ..., 15 , reveals that a = 13 , 15 a=13, 15 make 796 3 a 2 796-3a^2 a perfect square. Thus we get two pairs: ( a , b ) = ( 13 , 17 ) , ( 15 , 11 ) \color{#3D99F6}(a, b)=(13, 17), (15, 11) that result in N = 30 \color{#D61F06}N=-30 and N = 26 \color{#D61F06}N=26 , respectively.

If d = 2 d=2 , then 199 + N = 2 a 2 199+N=2a^2 and 199 3 N = 2 b 2 199-3N=2b^2 , where a a and b b are co-prime positive integers. It follows that 3 a 2 + b 2 = 398 3a^2+b^2=398 . Its easy to see that a 11 a \le 11 . Check that none of a 11 a \le 11 make 398 3 a 2 398-3a^2 a perfect square.

If d = 4 d=4 , then 199 + N = 4 a 2 199+N=4a^2 and 199 3 N = 4 b 2 199-3N=4b^2 , where a a and b b are co-prime positive integers. It follows that 3 a 2 + b 2 = 199 3a^2+b^2=199 . Its easy to see that a 7 a \le 7 . Checking a 7 a \le 7 , reveals that a = 1 a=1 makes 199 3 a 2 199-3a^2 a perfect square. Thus we get one pair: ( a , b ) = ( 1 , 14 ) \color{#3D99F6}(a, b)=(1, 14) that results in N = 198 \color{#D61F06}N=-198 .

Finally we get 5 5 solutions for n = 200 N n=200N , namely n = 39800 , 39000 , 6000 , 0 , 5200 n=-39800, -39000, -6000, 0, 5200 n = 39800 n=-39800 gives only s = 19900 s=19900 . n = 0 n=0 gives A = 0 A=0 , which is a contradiction! Other values of n n give two different solutions for s s . Thus the total number of solutions for the original equation is 7 \boxed{7} .


Notice that if ( r , s ) (r, s) is a solution, then ( s , r ) (-s, -r) is a solution too. Thus I write four of the solutions to the original equation. ( r , s ) = ( s + n , s ) = ( 19900 , 19900 ) , ( 126750 7 , 146250 7 ) , ( 12000 17 , 90000 17 ) , ( 67600 11 , 10400 11 ) (r, s)=(s+n, s)=\left(-19900, 19900 \right), \left(-\frac{126750}{7}, \frac{146250}{7} \right), \left(-\frac{12000}{17}, \frac{90000}{17} \right), \left( \frac{67600}{11}, \frac{10400}{11} \right)

Mark Hennings
Jun 20, 2017

The equation becomes r 3 s 3 = N ( N 1 ) r s r^3 - s^3 \; =\; N(N-1)rs where N = 200 N=200 , and we want r = s + n r = s+n for some integer n n . Thus 3 s 2 n + 3 s n 2 + n 3 = N ( N 1 ) ( s 2 + s n ) [ N ( N 1 ) 3 n ] s 2 + n [ N ( N 1 ) 3 n ] s n 3 = 0 \begin{aligned} 3s^2n + 3sn^2 + n^3 & = N(N-1)(s^2 + sn) \\ \big[N(N-1) - 3n\big]s^2 + n\big[N(N-1) - 3n\big]s - n^3 & = 0 \end{aligned} and so s = 1 2 n ± n 2 [ N ( N 1 ) 3 n ] ( N ( N 1 ) 3 n ) ( N ( N 1 ) + n ) s \; = \; -\tfrac12n \pm \frac{n}{2\big[N(N-1) - 3n\big]}\sqrt{\big(N(N-1)-3n\big)\big(N(N-1) + n\big)} Thus N ( N 1 ) n < 1 3 N ( N 1 ) -N(N-1) \le n < \tfrac13N(N-1) , and ( N ( N 1 ) 3 n ) ( N ( N 1 ) + n ) \big(N(N-1) - 3n\big)\big(N(N-1) + n\big) must be a perfect square.

Simple calculations with N = 200 N=200 shows that there are only 5 5 values of n n for which this is true: 39800 , 39000 , 6000 , 0 , 5200 -39800, -39000, -6000, 0, 5200 .

When n = 39800 n=-39800 , the only value of s s is 19900 19900 . When n = 0 n=0 , the only value of s s is 0 0 , and this is not a permitted solution of the original equation. The other three values of n n each yield two values of s s , and so there are a total of 7 \boxed{7} solutions.

Could u explain me the calculations u did to get the 5 values for n?

Maninder Dhanauta - 3 years, 11 months ago

Log in to reply

Set N= 200 into (N(N-1) - 3n) ( N(N-1) + n) = m^2 for some positive integer m m .

Expand and simplify, then apply Simon's favorite factoring trick .

Pi Han Goh - 3 years, 11 months ago

Log in to reply

@Pi Han Goh - I get m^2 + 3n^2 +79600n =1584040000 but how do i apply that trick to this type of equation

Maninder Dhanauta - 3 years, 11 months ago

Log in to reply

@Maninder Dhanauta Sorry, let me start over: expand my original equation gives

-3n^2 - 79600n + 1584040000 = m^2

since we are looking for integer roots, then the quadratic discriminant of the equation above (in n) is a perfect square:

79600^2 - 4(-3)(1584040000 - m^2) = c^2, for some integer c

Now can you solve it from here?

Pi Han Goh - 3 years, 11 months ago

Log in to reply

@Pi Han Goh You've got a Pell equation. That opens a whole new bag of difficulties!

Kazem Sepehrinia - 3 years, 11 months ago

Log in to reply

@Kazem Sepehrinia No, we don't get Pell's equation.

We are looking to solve ( N ( N 1 ) + n ) ( N ( N 1 ) 3 n ) = q 2 \big(N(N-1) + n\big)\big(N(N-1) - 3n\big) \; = \; q^2 for integers n , q n,q , where N = 200 N=200 . Expanding and collecting terms, this becomes [ 3 n + N ( N 1 ) ] 2 + 3 q 2 = 4 N 2 ( N 1 ) 2 \big[3n + N(N-1)\big]^2 + 3q^2 \; = \; 4N^2(N-1)^2 From this it is clear that n n and q q must be of the same parity, so we can write 3 n + N ( N 1 ) = 2 p q 3n + N(N-1) = 2p-q and rewrite the condition in the form ( 2 p q ) 2 + 3 q 2 = 4 N 2 ( N 1 ) 2 p 2 p q + q 2 = N 2 ( N 1 ) 2 ( p + q ω ) ( p + q ω 2 ) = N 2 ( N 1 ) 2 \begin{aligned} (2p-q)^2 + 3q^2 & = 4N^2(N-1)^2 \\ p^2 - pq + q^2 & = N^2(N-1)^2 \\ (p + q\omega)(p + q\omega^2) & = N^2(N-1)^2 \end{aligned} working in the Euclidean domain Z [ ω ] \mathbb{Z}[\omega] , where ω \omega is the principal cube root of unity. Since 2 2 and 5 5 are both irreducible in this Euclidean domain, we deduce that N = 200 N = 200 divides both p p and q q , and hence p = N P , q = N Q p = NP, q = NQ where ( P + Q ω ) ( P + Q ω 2 ) = 19 9 2 = ( 15 + 2 ω ) 2 ( 15 + 2 ω 2 ) 2 (P + Q\omega)(P + Q\omega^2) \; = \; 199^2 \; = \; (15 + 2\omega)^2(15 + 2\omega^2)^2 where the RHS is now a factorisation into irreducibles in Z [ ω ] \mathbb{Z}[\omega] ,

It now follows that P + Q ω P + Q\omega must be equal to one of ( 15 + 2 ω ) 2 u (15 + 2\omega)^2u , ( 15 + 2 ω 2 ) 2 u (15 + 2\omega^2)^2u or 199 u 199u , where u u is one of the six units in Z [ ω ] \mathbb{Z}[\omega] . Running through these 18 18 cases, only a selection of them end up giving an integer value for n n . Wtthout giving the details, we obtain 39000 , 39800 , 6000 , 0 , 5200 -39000, -39800, -6000, 0, 5200 as the only possible values of n n .

The way @Pi Han Goh was going with this, he would obtain an equation with c 2 + 12 m 2 c^2 + 12m^2 equal to a particular integer. This is basically the same as the one I have analysed, with an extra factor of 2 2 to worry about.

This approach gives us a strategy for dealing with the general case of other values of N N . We have to factorise N ( N 1 ) N(N-1) into irreducibles in Z [ ω ] \mathbb{Z}[\omega] ...

Mark Hennings - 3 years, 11 months ago

Log in to reply

@Mark Hennings On second viewing, I discovered that my work (without using Mark's Euclidean domain thingy technique, because I haven't mastered it yet!!) still requires plenty of trial and error. But I did try to minimize the amount of trials needed. Here's my failed attempt:

From what I've written, we find to find non-negative integer solutions to c 2 + 12 m 2 = 25344640000 c^2 + 12m^2 = 25344640000 .

Consider modulo 4, we can see that c 2 m o d 4 = 0 c^2 \bmod 4= 0 . Thus, let c = 2 c 1 c = 2c_1 and m = m 1 m =m_1 (for consistency sake). Then the equation simplifies to c 1 2 + 3 m 1 2 = 6336160000 c_1 ^2 + 3m_1 ^2 =6336160000 .

Consider modulo 5. The quadratic residues of 5 are 0,1 and 4. Since RHS is divisible by 5, then so does LHS. Thus c 1 c_1 and m 1 m_1 must be divisible by 5. Let c 1 = 5 c 2 , m 1 = 5 m 2 c_1 = 5c_2, m_1 = 5m_2 . The equation simplifies to c 2 2 + 3 m 2 2 = 253446400 c_2^2 + 3 m_2 ^2 = 253446400 .

Repeat the process again: Consider modulo 5. Since RHS is divisible by 5, then so does LHS. Thus c 2 c_2 and m 2 m_2 must be divisible by 5. Let c 2 = 5 c 3 , m 2 = 5 m 3 c_2 = 5c_3, m_2 = 5m_3 . The equation simplifies to c 3 2 + 3 m 3 2 = 10137856 = 318 4 2 c_3^2 + 3 m_3 ^2 = 10137856 = 3184^2 .

Not sure how to finish this off (elegantly)... I was thinking of sum of 2 (or 4) squares theorem, but I'm pretty sure that will fail.

Pi Han Goh - 3 years, 11 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...