Maths Olympiad Prep

Library / /426 of 520

Number theory Difficulty 6.8 National olympiad Prove it

Theorem 3 Let pp be an odd prime, αN\alpha \in \mathbf{N}^{*}, then a primitive root modulo pαp^{\alpha} exists.

Solution

To prove a fundamental idea is "translation".
First, prove: there exists a primitive root gg modulo pp such that
gρ1≢1(modp2)g^{\rho-1} \not \equiv 1\left(\bmod p^{2}\right)

In fact, take any primitive root gg modulo pp. If gg does not satisfy (1), we say g+pg+p is a primitive root modulo pp that satisfies (1).

First, using the binomial theorem and the fact that gg is a primitive root modulo pp, we know that g+pg+p is a primitive root modulo pp.

Second, we have
(g+p)ρ1gρ1+p(p1)gρ21pgρ2≢1(modp2)\begin{aligned} (g+p)^{\rho-1} & \equiv g^{\rho-1}+p(p-1) g^{\rho-2} \\ & \equiv 1-p g^{\rho-2} \not \equiv 1\left(\bmod p^{2}\right) \end{aligned}

Therefore, there exists a primitive root modulo pp that satisfies (1).
Next, prove: if gg is a primitive root modulo pp that satisfies (1), then for any αN\alpha \in \mathbf{N}^{*}, gg is a primitive root modulo pαp^{\alpha}.

First, we have the following conclusion: for any βN\beta \in \mathbf{N}^{*}, we can set
gα(pβ)=1+pβkβg^{\alpha\left(p^{\beta}\right)}=1+p^{\beta} \cdot k_{\beta}

where pkβp \nmid k_{\beta}. In fact, when β=1\beta=1, by the choice of gg, (2) holds. Now assume (2) holds for β\beta, then
gρ(pβ+1)=(gρ(pβ))p=(1+pβkβ)p1+pβ+1kβ(modpβ+2)\begin{aligned} g^{\rho\left(p^{\beta+1}\right)} & =\left(g^{\rho\left(p^{\beta}\right)}\right)^{p}=\left(1+p^{\beta} \cdot k_{\beta}\right)^{p} \\ & \equiv 1+p^{\beta+1} \cdot k_{\beta}\left(\bmod p^{\beta+2}\right) \end{aligned}

Combining pkβp \nmid k_{\beta}, we know that (2) holds for β+1\beta+1. Therefore, (2) holds for βN\beta \in \mathbf{N}^{*}.
Second, let δ=δpα(g)\delta=\delta_{p^{\alpha}}(g), then by Euler's theorem, we know δpα1(p1)\delta \mid p^{\alpha-1}(p-1). And by gg being a primitive root modulo pp, and gδ1(modpα)g^{\delta} \equiv 1\left(\bmod p^{\alpha}\right) (of course, also gδ1(modp)g^{\delta} \equiv 1(\bmod p)), we know (p1)δ(p-1) \mid \delta. Therefore, we can set δ=pβ1(p1)\delta=p^{\beta-1}(p-1), where 1βα1 \leqslant \beta \leqslant \alpha.

Now using (2), we know
gρ(pβ)1(modpβ+1), i.e., gδ1(modpβ+1).g^{\rho\left(p^{\beta}\right)} \neq \equiv 1\left(\bmod p^{\beta+1}\right) \text{, i.e., } g^{\delta} \neq 1\left(\bmod p^{\beta+1}\right) \text{.}

Combining gδ1(modpα)g^{\delta} \equiv 1\left(\bmod p^{\alpha}\right), we know βα\beta \geqslant \alpha.
In summary, β=α\beta=\alpha, i.e.,
δpα(g)=pα1(p1)=φ(pα)\delta_{p^{\alpha}}(g)=p^{\alpha-1}(p-1)=\varphi\left(p^{\alpha}\right)

Thus, gg is a primitive root modulo pαp^{\alpha}.

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.