# What error bound would an epsilon closer to the Riemann hypothesis give?

• $s=1$ line gives: $$\psi(x) = x(1+o(1))$$

• classical zero free region gives: $$\psi(x) = x + O(x e^{-c \sqrt{\log x}})$$ for some positive constant $\delta$

• RH gives: $$\psi(x) = x + O(\sqrt{x}\log(x)^2)$$

So I was wondering what error term would we get if someone proved there were no zeros with real part $< 1-\varepsilon$?

and why does Terry Tao say understanding the error term is so important?

-
I have seen this type of assumption referred to as a "quasi-Riemann hypothesis". See e.g. p. 5 of homepages.math.uic.edu/~cojocaru/twin-primes-march05.pdf. (Although it looks weaker, it is not clear that it is any more tractable than RH itself!) – Pete L. Clark Feb 16 '13 at 19:50

Instead of the exponent $1/2$ as in the $\sqrt x$ in the error, you get the larger $x^{1-\epsilon}$