Dice Question Streamline Icon: https://streamlinehq.com

Type I information at level X^{1/2} for product-form weights in Q(√−n)

Establish Type I estimates at level L = X^{1/2} for the Type I sum defined by ∑_{Nd ∼ L} |∑_{d | a, N a ≤ X} w(a)| in the imaginary quadratic field K = Q(√−n), where the weight function w is of product form w = f ⊠_ℓ f′ as in Definition 2.1 and f, f′ are supported on [±2n X^{1/2}]. In particular, determine whether analogous bounds at level L = X^{1/2} can be obtained without the absolute values in the inner sum, namely for ∑_{Nd ∼ L} ∑_{d | a, N a ≤ X} w(a), corresponding to Type I₂ information.

Information Square Streamline Icon: https://streamlinehq.com

Background

In their sieve framework over the number field K = Q(√−n), the authors rely on Type I and Type II information to control sums over prime ideals associated with primes of the form x2 + n y2. Type I sums are defined by equation (3.1) as ∑{Nd ∼ L} |∑{d | a} w(a)|, with L referred to as the level. Achieving sufficiently high levels for Type I and Type II estimates is crucial to run the Duke–Friedlander–Iwaniec sieve and derive asymptotics.

The paper proves Type I information up to level L = X{1/2 − o(1)} and Type II up to L between X{o(1)} and X{1/2 − o(1)}, which suffices for their main theorem. However, they highlight that reaching the critical level L = X{1/2} for Type I would be a substantive advance. Moreover, even proving bounds at that level without absolute values in the Type I sum (a form of Type I₂ information) appears difficult; achieving this would enable arbitrary logarithmic savings in their main asymptotic for primes of the form x2 + n y2.

This open problem is specifically about product-form weights w = f ⊠_ℓ f′ (Definition 2.1) tied to their Gaussian and number field setting, and directly concerns improving the sieve inputs beyond what is currently established in the paper.

References

We remark that obtaining Type I information at level $X{1/2}$ in our setting remains an interesting open question. In fact, even obtaining bounds at this level without the absolute values in \cref{typei-def} seems challenging.

Primes of the form $p^2 + nq^2$ (2410.04189 - Green et al., 5 Oct 2024) in Section 1.2 (Outline of the argument)