A1.9

Number Theory | Part II, 2003

(i) Let pp be an odd prime and kk a strictly positive integer. Prove that the multiplicative group of relatively prime residue classes modulo pkp^{k} is cyclic.

[You may assume that the result is true for k=1k=1.]

(ii) Let n=p1p2prn=p_{1} p_{2} \ldots p_{r}, where r2r \geqslant 2 and p1,p2,,prp_{1}, p_{2}, \ldots, p_{r} are distinct odd primes. Let BB denote the set of all integers which are relatively prime to nn. We recall that nn is said to be an Euler pseudo-prime to the base bBb \in B if

b(n1)/2(bn)modn.b^{(n-1) / 2} \equiv\left(\frac{b}{n}\right) \bmod n .

If nn is an Euler pseudo-prime to the base b1Bb_{1} \in B, but is not an Euler pseudo-prime to the base b2Bb_{2} \in B, prove that nn is not an Euler pseudo-prime to the base b1b2b_{1} b_{2}. Let pp denote any of the primes p1,p2,,prp_{1}, p_{2}, \ldots, p_{r}. Prove that there exists a bBb \in B such that

(bp)=1 and b1modn/p\left(\frac{b}{p}\right)=-1 \quad \text { and } \quad b \equiv 1 \bmod n / p

and deduce that nn is not an Euler pseudo-prime to this base bb. Hence prove that nn is not an Euler pseudo-prime to the base bb for at least half of all the relatively prime residue classes bmodnb \bmod n.

Typos? Please submit corrections to this page on GitHub.