Maths Olympiad Prep

Library / /469 of 520

Number theory Difficulty 7.1 National olympiad, round 2 Prove it

Theorem 8.9. Let pp be an odd prime, then pkp^{k} has a primitive root for all positive integers kk. Moreover, if rr is a primitive root modulo p2p^{2}, then rr is a primitive root modulo pkp^{k}, for all positive integers kk.

Solution

Proof. From Theorem 8.8, we know that pp has a primitive root rr that is also a primitive root modulo p2p^{2}, so that
rp1≢1(modp2)r^{p-1} \not \equiv 1\left(\bmod p^{2}\right)

Using mathematical induction, we will prove that for this primitive root r ,
rpk2(p1)≢1(modpk)r^{p^{k-2}(p-1)} \not \equiv 1\left(\bmod p^{k}\right)
for all positive integers k . Once we have established this congruence, we can show that rr is also a primitive root modulo pkp^{k} by the following reasoning. Let
n= ord pεn=\text { ord }_{p^{*}} \varepsilon \text {. }

From Theorem 6.8 , we know that nϕ(pk)=pk1(p1)n \mid \phi\left(p^{k}\right)=p^{k-1}(p-1). On the other hand, since
rn1(modpk)r^{n} \equiv 1\left(\bmod p^{k}\right)
we also know that
rn1(modp)r^{n} \equiv 1(\bmod p)

From Theorem 8.1, we see that p1=ϕ(p)np-1=\phi(p) \mid n. Because (p1)n(p-1) \mid n, and npk1(p1)n \mid p^{k-1}(p-1), we know that n=pt(p1)n=p^{t}(p-1), where tt is an integer such that 0tk10 \leqslant t \leqslant k-1. If n=pt(p1)n=p^{t}(p-1) with tk2t \leqslant k-2, then
rpk2(p1)=(rp(p1))pk211(modpk)r^{p^{k-2}(p-1)}=\left(r^{p^{\prime}(p-1)}\right)^{p^{k-2-1}} \equiv 1\left(\bmod p^{k}\right)
which would contradict (8.3). Hence, ord kr=pk1(p1)=ϕ(pk){ }^{k} r=p^{k-1}(p-1)=\phi\left(p^{k}\right). Consequently, rr is also a primitive root modulo pkp^{k}.

All that remains is to prove (8.3) using mathematical induction. The case of k=2k=2 follows from (8.2). Let us assume the assertion is true for the positive integer k2k \geqslant 2. Then
rpk2(p1)≢1(modpk)r^{p^{k-2}(p-1)} \not \equiv 1\left(\bmod p^{k}\right)

Since (r,p)=1(r, p)=1, we know that (r,pk1)=1\left(r, p^{k-1}\right)=1. Consequently, from Euler's theorem, we know that
rpk2(p1)rϕ(pk1)r^{p^{k-2}(p-1)} \equiv r^{\phi\left(p^{k-1}\right)}

Therefore, there is an integer dd such that
rpk2(p1)=1+dpk1r^{p^{k-2}(p-1)}=1+d p^{k-1}
where p\dp \backslash d, since by hypothesis rpt2(p1)≢1(modpk)r^{p^{t-2}(p-1)} \not \equiv 1\left(\bmod p^{k}\right). We take the pp th power of both sides of the above equation, to obtain, via the binomial theorem,
rpk1(p1)=(1+dpk1)p=1+p(dpk1)+(p2)p2(dpk1)2++(dpk1)p1+dpk(modpk+1)\begin{aligned} r^{p^{k-1}(p-1)} & =\left(1+d p^{k-1}\right)^{p} \\ & =1+p\left(d p^{k-1}\right)+\binom{p}{2} p^{2}\left(d p^{k-1}\right)^{2}+\cdots+\left(d p^{k-1}\right)^{p} \\ & \equiv 1+d p^{k}\left(\bmod p^{k+1}\right) \end{aligned}

Since pdp \nmid d, we can conclude that
rpk1(p1)≢1(modpk+1)r^{p^{k-1}(p-1)} \not \equiv 1\left(\bmod p^{k+1}\right)

This completes the proof by induction.

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.