Maths Olympiad Prep

Library / /517 of 520

Number theory Difficulty 7.9 National olympiad, round 2 Prove it

Theorem 3 The indeterminate equation (6) has solutions satisfying the condition (x,y)=1(x, y)=1 if and only if n=1,2,n1n=1,2, n_{1} or 2n12 n_{1}, where
n1=p1a1prar,pj1(mod4),1jrn_{1}=p_{1}^{a_{1}} \cdots p_{r}^{a_{r}}, \quad p_{j} \equiv 1(\bmod 4), 1 \leqslant j \leqslant r

Solution

To prove the necessity, we need to show that 4n4 \nmid n and nn has no prime factors of the form 4k+34k+3. In this case, we must have (x,y)=(x,n)=(y,n)=1(x, y) = (x, n) = (y, n) = 1. If 4n4 \mid n, then we have
x2+y20(mod4)x^{2} + y^{2} \equiv 0 \pmod{4}

This is impossible when (x,2)=(y,2)=1(x, 2) = (y, 2) = 1. Therefore, we must have 4n4 \nmid n. If there exists a prime p3(mod4)p \equiv 3 \pmod{4} such that pnp \mid n, then we have (p,x)=(p,y)=1(p, x) = (p, y) = 1 and
x2+y20(modp)x^{2} + y^{2} \equiv 0 \pmod{p}

Following the proof of the necessity in Theorem 2, we can deduce that p1(mod4)p \equiv 1 \pmod{4}, which is a contradiction.

For sufficiency, 1=12+021 = 1^{2} + 0^{2} and 2=12+122 = 1^{2} + 1^{2}, so the conclusion is obviously true. We will use induction to prove that when n=pαn = p^{\alpha} and p1(mod4)p \equiv 1 \pmod{4},
x2+y2=pα,(x,y)=1x^{2} + y^{2} = p^{\alpha}, \quad (x, y) = 1

has a solution. When α=1\alpha = 1, this is Theorem 1, so there is a solution. Assume the conclusion holds for α=k\alpha = k. We will prove that the conclusion also holds for α=k+1\alpha = k + 1. Suppose
x12+y12=p,(x1,y1)=1xk2+yk2=pk,(xk,yk)=1.\begin{array}{ll} x_{1}^{2} + y_{1}^{2} = p, & (x_{1}, y_{1}) = 1 \\ x_{k}^{2} + y_{k}^{2} = p^{k}, & (x_{k}, y_{k}) = 1 . \end{array}

From this and equation (5), we have
(x1xk+y1yk)2+(x1yky1xk)2=pk+1(x1xky1yk)2+(x1yk+y1xk)2=pk+1\begin{array}{l} (x_{1} x_{k} + y_{1} y_{k})^{2} + (x_{1} y_{k} - y_{1} x_{k})^{2} = p^{k+1} \\ (x_{1} x_{k} - y_{1} y_{k})^{2} + (x_{1} y_{k} + y_{1} x_{k})^{2} = p^{k+1} \end{array}

At this point,
d1=(x1xk+y1yk,x1yky1xk)=1d_{1} = (x_{1} x_{k} + y_{1} y_{k}, x_{1} y_{k} - y_{1} x_{k}) = 1

or
d2=(x1xky1yk,x1yk+y1xk)=1d_{2} = (x_{1} x_{k} - y_{1} y_{k}, x_{1} y_{k} + y_{1} x_{k}) = 1

at least one of them holds. If not, d1>1d_{1} > 1 and d2>1d_{2} > 1, then from equations (10) and (11), we know pd1p \mid d_{1} and pd2p \mid d_{2}. Therefore, p2x1xkp \mid 2 x_{1} x_{k}, so at least one of px1p \mid x_{1} or pxkp \mid x_{k} must hold, but from equations (8) and (9), this is impossible. Therefore, when d1=1d_{1} = 1, from equation (10), the conclusion holds for α=k+1\alpha = k + 1. When d2=1d_{2} = 1, from equation (11), the conclusion holds for α=k+1\alpha = k + 1. This proves that for any α1\alpha \geq 1, the indeterminate equation (7) always has a solution.

Next, we prove that if (n1,n2)=1(n_{1}, n_{2}) = 1 and
x12+y12=n1,(x1,y1)=1x22+y22=n2,(x2,y2)=1\begin{array}{ll} x_{1}^{2} + y_{1}^{2} = n_{1}, & (x_{1}, y_{1}) = 1 \\ x_{2}^{2} + y_{2}^{2} = n_{2}, & (x_{2}, y_{2}) = 1 \end{array}

then when n=n1n2n = n_{1} n_{2}, the indeterminate equation (6) must have a solution satisfying (x,y)=1(x, y) = 1. From equation (5), we have
(x1x2+y1y2)2+(x1y2y1x2)2=n1n2\left(x_{1} x_{2} + y_{1} y_{2}\right)^{2} + \left(x_{1} y_{2} - y_{1} x_{2}\right)^{2} = n_{1} n_{2}

If d=(x1x2+y1y2,x1y2y1x2)>1d = \left(x_{1} x_{2} + y_{1} y_{2}, x_{1} y_{2} - y_{1} x_{2}\right) > 1, then there must be a prime qdq \mid d. Therefore,
x1x2y1y2(modq),x1y2y1x2(modq)\begin{array}{l} x_{1} x_{2} \equiv -y_{1} y_{2} \pmod{q}, \\ x_{1} y_{2} \equiv y_{1} x_{2} \pmod{q} \end{array}

At this point, qx1q \nmid x_{1}. Otherwise, if qx1q \mid x_{1}, from the above two equations, we get qy1y2q \mid y_{1} y_{2} and qy1x2q \mid y_{1} x_{2}. From (x1,y1)=1(x_{1}, y_{1}) = 1, we deduce qy2q \mid y_{2} and qx2q \mid x_{2}, which contradicts (x2,y2)=1(x_{2}, y_{2}) = 1. Similarly, we can prove that qq cannot divide y1y_{1}, x2x_{2}, or y2y_{2}. From equations (15) and (16), we easily get
x1(x22+y22)0(modq),x2(x12+y12)0(modq)\begin{array}{l} x_{1} \left(x_{2}^{2} + y_{2}^{2}\right) \equiv 0 \pmod{q}, \\ x_{2} \left(x_{1}^{2} + y_{1}^{2}\right) \equiv 0 \pmod{q} \end{array}

From this and qx1x2q \nmid x_{1} x_{2}, we get (note: qq is a prime)
x12+y12x22+y220(modq)x_{1}^{2} + y_{1}^{2} \equiv x_{2}^{2} + y_{2}^{2} \equiv 0 \pmod{q}

From this and equations (12) and (13), we get n1n20(modq)n_{1} \equiv n_{2} \equiv 0 \pmod{q}, which contradicts (n1,n2)=1(n_{1}, n_{2}) = 1. Therefore, d=1d = 1. From this and equation (14), we have proven the desired conclusion.

Combining all the above discussions, we have proven the sufficiency. Proof complete.

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: NuminaMath-1.5, licensed Apache-2.0. Statement and solution reproduced as published; topic and difficulty added by this site.