Behaviour of the sequence $\vartheta_n = \vartheta(p_n)$

Abstract

The well-known sequence $\vartheta_n = \vartheta(p_n) = \sum_{i=1}^n \ln p_i= \ln\left([p_n]\#\right)$ exhibits numerous extremely interesting properties. Since $p_n = \exp(\vartheta_n - \vartheta_{n-1})$, it is immediately clear that the two sequences $p_n \longleftrightarrow \vartheta_n$ must ultimately encode exactly the same information. But the sequence $\vartheta_n$, while being extremely closely correlated with the primes, (in fact, $\vartheta_n \sim p_n$), is very much better behaved than the primes themselves. Using numerous suitable extensions of various reasonably standard results, I shall demonstrate that the sequence $\vartheta_n$ satisfies suitably defined $\vartheta$-analogues of the usual Cramer, Andrica, Legendre, Oppermann, Brocard, Firoozbakht, Fourges, Nicholson, and Farhadian conjectures. (So these $\vartheta$-analogues are not conjectures, they are instead theorems.) The crucial key to enabling this pleasant behaviour is the regularity (and relative smallness) of the $\theta$-gaps $\mathfrak{g}_n = \vartheta_{n+1}-\vartheta_n= \ln p_{n+1}$. While superficially these results bear close resemblance to some recently derived results for the averaged primes, $\bar p_n = {1\over n} \sum_{i=1}^n p_i$, both the broad outline and the technical details of the arguments given and proofs presented are quite radically distinct.