Notice that with positive integers a,m,n and a>1, we have
gcd(am−1,an−1)=agcd(m,n)−1.
Let f(r)=p−1pr−1 with p a prime and r a positive integer.
1.
Let x=gcd(pr−1,pp−1) and y=lcm(pr−1,pp−1), then
(pr−1)(pp−1)=xy.
From the above lemma, we get x=pgcd(r,p)−1=p−1. Hence (pr−1)(pp−1)=(p−1)y, which implies that F(r)=yppr−1. But pr−1∣ppr−1 and pp−1∣ppr−1, so y∣ppr−1 leads to F(r) being a positive integer.
2.
We can see that
gcd(prp−1,psp−1)=pgcd(rp,sp)−1=pp−1.
Let prp−1=(pp−1)r1, psp−1=(pp−1)s1 with gcd(r1,s1)=1, then
gcd(F(r),F(s))=gcd((pr−1)(pp−1)(pp−1)r1(p−1),(ps−1)(pp−1)(pp−1)s1(p−1))=gcd(f(r)r1,f(s)s1)=1
3.
At first, we will show that with any prime divisor q of F(r), we always have p∣q−1. (*)
Indeed, from Fermat's theorem, q∣pq−1−1 and because q∣(pr−1)f(p)prp−1, so q∣prp−1.
These imply that
q∣gcd(pq−1−1,prp−1)=pgcd(q−1,rp)−1
Since r,p are two primes, d=gcd(q−1,rp)∈{1,r,p,rp}. We have to consider 4 following cases:
1. If d=p or d=rp, we get p∣q−1 and (*) follows.
2. If d=1, we have q∣p−1 or p≡1(modq). We also have
0≡pp−1prp−1=pp(r−1)+pp(r−2)+⋯+pp+1≡p≡1(modq),
contradiction.
3. If d=r, we have q∣pr−1 or pr≡1(modq). We also have
0≡pr−1prp−1=pr(p−1)+pr(p−2)+⋯+pr+1≡r(modq),
contradiction.
Hence (*) is true. Finally, assume that for all prime q∣F(r), we always have p2∣q−1.
Because all divisors of F(r) are congruent to 1 modulo p2, then F(r)≡1(modp2).
Note that
F(r)(pp−1)(pr−1)=(prp−1)(p−1)=prp+1−prp−(p−1)≡−(p−1)(modp2)
and
F(r)(pp−1)(pr−1)=F(r)(pp+r−pr−pp+1)≡F(r)≡1(modp2).
These imply that
1≡−(p−1)(modp2)⇔p≡0(modp2),
which is a contradiction.
Therefore, there exists a prime satisfying all the given conditions. □