# Partial sums of the Möbius function on arithmetic progressions

There is a result of Soundararajan on the upper bound of the partial sums of the Möbius function assuming GRH here. Suger and Halupczok find an analogous bound for $$\displaystyle \sum_{\substack{n\leq x\\n \equiv a \bmod q}}\mu(n)$$, also assuming GRH. However, without assuming GRH, what are some non-trivial upper bounds on $$\displaystyle \sum_{\substack{n\leq x\\n \equiv a \bmod q}}\mu(n)?$$ It does not have to be the most optimal one (although I would appreciate being linked to such a bound). In fact, I don't even know of a non-trivial bound not assuming GRH on $$\displaystyle \sum_{n\leq x}\mu(n)$$. Thanks!

A variant of the Siegel-Walfisz theorem states that there is a constant $$c>0$$ with the following property. For any $$A>0$$ and $$q\leq(\log x)^A$$, we have $$\displaystyle \sum_{\substack{n\leq x\\n \equiv a \bmod q}}\mu(n)\ll_A x\exp\left(-c\sqrt{\log x}\right).$$ See Exercise 13 for Section 11.3 of Montgomery-Vaughan: Multiplicative number theory I.
• Many thanks for your answer. Are there also some results when $q\ge (\log x)^A$? Actually, I am quite satisfied as long as I can be sure that $\displaystyle \sum_{\substack{n\leq x\\n \equiv a \bmod q}}\mu(n)=o(x)$ for all $a\in \mathbb{Z}$ and $q\in \mathbb{N}$. – Remarkably Unremarkable Apr 21 at 20:10
• @RemarkablyUnremarkable: One can go up to $q\leq\exp(2c\sqrt{\log x})$, with the same $c>0$ as before, but there will be a main term depending on a possible Siegel zero for quadratic Dirichlet characters modulo $q$. If the Siegel zero does not exist, then the same bound holds true as for $q\leq(\log x)^A$. See Exercises 12-13 for Section 11.3. – GH from MO Apr 22 at 3:07