@DanielWeber Excellent idea! I generated 4 random “control” curves with “fake” $\mu$ functions that coincide with $\mu(n)$ for $n\leq 12$ (I think $100$ is too much, so I went with $12$) and otherwise have the same probability distribution on $\{0,\pm 1\}$, like you suggested. So as not to put too much stuff on the question, I posted them here on Bluesky (no account necessary to view; also here on Twitter). Now I don't know what to think any more!
↧