Author |
Topic: A non-zero exponent puzzle (Read 682 times) |
|
K Sengupta
Senior Riddler
Gender:
Posts: 371
|
|
A non-zero exponent puzzle
« on: Jan 21st, 2007, 8:20am » |
Quote Modify
|
Determine analytically all possible non-zero integer solutions to this equation: p^3 = q(q^8 - 9*q^5+ 81q- 81)
|
|
IP Logged |
|
|
|
Eigenray
wu::riddles Moderator Uberpuzzler
Gender:
Posts: 1948
|
|
Re: A non-zero exponent puzzle
« Reply #1 on: Jan 21st, 2007, 9:44pm » |
Quote Modify
|
I know there are only finitely many solutions. There are four cases to consider, and the four relevent equations are x3-y3=1,3, whose solutions are evident; and x3-2y3=1,3, whose only solutions seem to be x=1,-1,-5, but in any case are finite in number by Thue's theorem. So presumably (p,q)=(-2,1) is the only non-zero solution.
|
|
IP Logged |
|
|
|
Eigenray
wu::riddles Moderator Uberpuzzler
Gender:
Posts: 1948
|
|
Re: A non-zero exponent puzzle
« Reply #2 on: Jan 22nd, 2007, 9:23am » |
Quote Modify
|
Okay, I have an elementary proof: hidden: | p3 = q(q2-3)(q2+3)(q4-9q+9) =: q*A*B*C. 1) First suppose q is not divisible by 3. Then none of A,B,C are divisible by 3. Moreover, q is relatively prime to A,B,C, and so q=r3 and ABC must both be cubes. Note C = AB - 9(q-2). Since A = (q-2)(q+2)+1, it follows gcd(A,C)=1. Also, B = (q-2)(q+2)+7, so (B,C) is either 1 or 7. But q=r3 is either 0,1, or 6 mod 7, so B is not divisible by 7. Hence (B,C)=1. Now, AB is relatively prime to C, so AB=s3 is a cube, and 9 = q4-AB = (r4)3-s3, and the only way to write 9 as a difference of cubes is 8 - (-1) or 1 - (-8). So r4=2 (impossible), or r4=1, which gives r=+/-1 and q=+/-1. q=1 gives p3=-8, okay, but q=-1 gives p3=152, no good. So we get one solution: q=1 (and p=-2). 2) Now suppose q=3t. We have p3 = 35t(3t2-1)(3t2+1)(9t4-3t+1) =: 35t*X*Y*Z. Now X,Y,Z are relatively prime to 3 and t, so 35t and XYZ are both cubes. Write t=3r3. Since Y-X=2, we have (X,Y)=1 (if t is even) or 2 (if t is odd), and as before we find (X,Z)=1. Case I: (X,Y)=1. Then X is relatively prime to YZ, so X=s3 is a cube, and 1 = 3t2-X = (3r2)3-s3, and the only cubes differing by 1 are 0 and +/-1. So 3r2 = 1 (impossible) or 0 (giving q=0). Case II: (X,Y)=2. Then t is odd, so X=2 mod 4, so X/2 is odd, hence relatively prime to Y. Then X/2 is relatively prime to 2YZ, so X/2 is a cube, hence 0,1, or 8 mod 9. But 3|t, so X=8 mod 9, a contradiction. | It works out pretty nicely after all (I originally had 8 cases to consider).
|
« Last Edit: Jan 22nd, 2007, 9:23am by Eigenray » |
IP Logged |
|
|
|
K Sengupta
Senior Riddler
Gender:
Posts: 371
|
|
Re: A non-zero exponent puzzle
« Reply #3 on: Jan 23rd, 2007, 6:59am » |
Quote Modify
|
On Jan 22nd, 2007, 9:23am; Eigenray wrote: Quote:(p,q)=(-2,1) is the only non-zero solution |
| True. However, the problem under reference can also be solved in the following manner. p^3 = q^9 - 9*q^6 + 81*q^2 -81q Or, p^3 = (q^3 -3)^3 - (3q-3)^3 By Fermat's Last Theorem, for non-zero integers a, b and c; a^3 cannot equal b^3+c^3. So, relaxing the non-zero restriction on a,b and c, we observe that: (a,b,c) = (0, k,-k); (m,0,m); (n,n,0) gives all possible integer solutions for fixed k,m,n....(B) Now, from (B) we observe that: (a,b,c) =(0,k,-k) would imply that p=0. This is a contradiction. (a,b,c) =(m,0,m) would imply that q^3 = 3, so that q = cuberoot(3), which is not an integer, and thus constitutes a violation of the tenets comprising the problem under reference. (a,b,c) = (n,n,0) gives 3x=3, so that q=1, giving : p= 1-3 = -2. Thus, (q,p) = (1,-2) is the only possible solution to the given problem.
|
|
IP Logged |
|
|
|
|