By the induction method we will show that divide in to two subset A,B such that ai+bi is prime number for i∣1,n.
If n=1 then {1,2}={1}∪{2}, 1+2=3 is prime number and {1}∩{2}=∅. Assume now that the result is proved for 1,2,3,...,n−1. Now we will show that for n.
By well known Bertrand's postulate there exists prime number p such that 2n<p<4n. Because of p>2n≥2 then p-odd prime number. Thus p=2n+m, m-odd positive integer. Now consider
(2n,m);(2n−1,m+1);…(22n+m+1,22n+m−1)
pairs and those pair's two numbers sum is p=2n+m. For other remaining numbers are 1,2,...,m−1 get new set that is {1,2,...,m−1}, which is by the induction method dividing into two subsets. Because m−1 is even number. Thus the set {1,2,3,...,2n} dividing into two subsets.
Let (a1+b1)...(an+bn)=p1α1p2α2...psαs, α1+...+αs=n, αi∈N.
Hence the number of divisors of (a1+b1)...(an+bn) is N:=(α1+1)...(αs+1).
If αj=1 for some j then
N=2⋅(α1+1)(α2+1)...(αj−1+1)(αj+1+1)...(αs+1).
So we need to prove that following inequality
αj+1∏(αj+1)≤2n−1.
This inequality is same as property (iii). Hence we can assume that αj+1 for arbitrary i∈{1,2,...,s}. Using the Cauchy's inequality, we get that
i=1∏s(αi+1)≤(s∑i=1s(αi+1))s=(sn+s)s.
Let us consider f(x)=(xn+x)x function on R.
Taking derivative, we get that
f′(x)=(1+xn)xln(1+xn)−xn(1+xn)n−1==(1+xn)x−1⋅((1+xn)ln(1+xn)−xn).
Observe that x≤2n⇔xn≥2. So ln(1+xn)≥ln3>1, thus we concluded that (1+xn)ln(1+xn)−xn>1>0. So f is increasing function on (0;n/2]. Finally, the f(x) function get maximum value the point x=2n. Hence we can see that f(2n)=3n/2<2n. Proof is completed.