Find shortest prefix to generate original string by overlapping

Question

Given a string $S$, I want to find the prefix string $P$ of shortest length, such that the original string $S$ can be generated by concatenating copies of $P$ (where overlapping is allowed).

For example, if $S = atgatgatatgat$, I want to find $P = atgat$; $P$ is the smallest prefix of $S$ that can be concatenated (in this case three times, starting at indices $\{0,3,8\}$ of $S$, where the first and second copies overlap but the second and third copies do not overlap) to equal $S$.

Obviously, there is an $\mathcal{O}(n^2)$ algorithm by checking each prefix of $S$, but a colleague mentioned it might be possible to do it in $\mathcal{O}(n \log n)$. I'm thinking of using suffix arrays for different prefixes of $S$ but haven't quite been able to proceed from there.

Answer 1

You are asking about "quasiperiodicity". This is a well-studied topic and a google scholar search will turn up many papers about it. For example, there is an $O(n (\log n)^2)$ algorithm here and an $O(n \log n)$ algorithm here.