From the statement we have
τ(nan+1n+(n+1)ann+1)⊢n.(1)
Let a1=1 and a2=2. Suppose that we have constructed the numbers a1,a2,…,am so that they are different natural numbers and (1) satisfied for each n=1,2,…,m−1. Let L be the smallest positive integer that does not occur among the numbers a1,a2,…,am. We set am+2=L and choose am+1 so that am+1>a1+a2+⋯+am+L and (1) is true for n=m and n=m+1. Repeating this process we obtain a permutation of all positive integers satisfying (1) for any positive integer n.
For a prime p and a positive integer k, we denote by ψp(k) the largest integer l such that k is divisible by pl.
Lemma. Given positive integers a,b,k,l,M. Then there is a prime number p>M and a positive integer x such that ψp(axk+b)=l.
Proof. Let d=M+k+b and p be a prime divisor of abk−1(d!)k+1. Then p>d>M. By induction on N, we prove that for any positive integer N there exists a positive integer x such that the number axk+b is divisible by pN.
Base: for N=1 the number x=d!b.
Suppose that for N there exist positive integers x and t such that axk+b=pNt. Let z be a positive integer such that the number axn−1zk+t is divisible by p (such a number exists, since the number p is coprime with d!a, and hence with the number akx). Then
a(x+zpN)k+b≡axk+axk−1zpNk+b≡pNt+axk−1zpNk≡pN(axk−1zk+t)≡0(modpN+1),
and the induction step is proved.
Therefore, there are positive integers x,t with axk+b=pl+1t. Then
a(x+pl)k+b≡axk+axk−1pk+b≡axk−1kpl(modpl+1),
i.e. ψp(a(x+pl)k+b)=l, and the lemma is proved.
From the lemma for a=m,b=(m+1)amm+1,M=1,k=m,l=m−1 there is a prime p>1 and a positive integer x such that ψp(mxm+(m+1)amm+1)=m−1 and from the lemma for a=m+2,b=(m+1)km+1,M=p,k=m+2,l=m there is a prime q>p and a positive integer y such that ψq((m+1)km+1+(m+2)ym+2)=m. By the Chinese remainder theorem, there is a positive integer z such that z>a1+a2+⋯+am+L,z≡x(modpm) and z≡y(modqm+1). Then it is easy to see that am+1=z works.