Maths Olympiad Prep

Library / /72 of 105

Number theory Difficulty 5.2 AIME, harder Prove it United States

Problem:

Determine all triples (x,y,n)(x, y, n) of integers such that x2+2y2=2nx^{2} + 2 y^{2} = 2^{n}.

Solution

Solution:

It is easy to check that (±2r,0,2r)\left( \pm 2^{r}, 0, 2r \right) and (0,±2r,2r+1)\left(0, \pm 2^{r}, 2r+1 \right) satisfy this equation for any nonnegative integer rr. We will show that these are all the solutions by an infinite descent method.

So suppose we have some solution (x0,y0,n0)\left(x_{0}, y_{0}, n_{0}\right). If x0x_{0} is odd, then 2n02^{n_{0}} is odd, which forces n0=0n_{0}=0 and then x02+2y02=1x_{0}^{2} + 2 y_{0}^{2} = 1, so y0=0y_{0} = 0 (or else 2y02>12 y_{0}^{2} > 1 already) and then x0=±1x_{0} = \pm 1.

On the other hand, if x0x_{0} is even, we can let x0=2x0x_{0} = 2 x_{0}' and then 4x02+2y02=2n0y02+2x02=2n014 x_{0}'^{2} + 2 y_{0}^{2} = 2^{n_{0}} \Rightarrow y_{0}^{2} + 2 x_{0}'^{2} = 2^{n_{0}-1}, so (x1,y1,n1)=(y0,x0/2,n01)\left(x_{1}, y_{1}, n_{1}\right) = \left(y_{0}, x_{0}/2, n_{0}-1\right) is another solution to our equation, where n0n_{0} has been replaced by n01n_{0}-1. Now if x1x_{1} is even, we can repeat this construction to get another new solution (x2,y2,n2)\left(x_{2}, y_{2}, n_{2}\right) with n01n_{0}-1 replaced by n02n_{0}-2, and so on. These integers nn cannot go on decreasing forever, since there does not exist an integral solution where n<0n < 0. Thus, eventually our process terminates, which means we get to a solution (xk,yk,nk)\left(x_{k}, y_{k}, n_{k}\right) with xkx_{k} odd. By the above, this is possible only if xk=±1,yk=0,nk=0x_{k} = \pm 1, y_{k} = 0, n_{k} = 0.

On the other hand, the above construction can be performed in reverse: we have xi=2yi+1,yi=xi+1,ni=ni+1+1x_{i} = 2 y_{i+1}, y_{i} = x_{i+1}, n_{i} = n_{i+1} + 1 for each value of i0i \geq 0. Now we claim that (xi,yi,ni)=(±2(ki)/2,0,ki)\left(x_{i}, y_{i}, n_{i}\right) = \left( \pm 2^{(k-i)/2}, 0, k-i \right) when kik-i is even, and (0,±2(ki1)/2,ki)\left(0, \pm 2^{(k-i-1)/2}, k-i \right) when kik-i is odd. The proof is by downward induction: the base case i=ki = k is certainly true; given that the statement holds for some i>0i > 0, it is simple algebra to check that it holds for i1i-1 by applying our reverse construction. Thus, the claim is true for each i0i \geq 0. In particular, (x0,y0,n0)=(±2k/2,0,k)\left(x_{0}, y_{0}, n_{0}\right) = \left( \pm 2^{k/2}, 0, k \right) or (0,±2(k1)/2,k)\left(0, \pm 2^{(k-1)/2}, k \right), which fits the form above. So, every solution is of this form.

Want a route through all this instead of an archive? The track puts 2,000 problems in a working order, from AMC 10 level to the IMO shortlist.

Source: MathNet, licensed CC-BY-4.0. Statement reproduced verbatim; metadata (topic, difficulty) added by this project.