The paper tackles two classic problems in number theory: finding the smallest prime number in an arithmetic progression (a sequence like 3, 7, 11, 15... where numbers are evenly spaced), and finding the smallest quadratic non-residue modulo a number (roughly, the smallest number that cannot be expressed as a perfect square in a certain modular arithmetic sense). For decades, the best known bounds for these problems have come from assuming the Generalized Riemann Hypothesis, or GRH, which is a famous unproven conjecture about where certain mathematical objects called the zeros of L-functions can be located along a vertical line in the complex plane.
The key insight of the paper is that the GRH only constrains where these zeros sit horizontally, but says nothing about how they are vertically spaced relative to one another. The vertical spacing of zeros is studied through something called "pair-correlation," which describes statistical patterns in the gaps between zeros. By bringing in additional assumptions about this vertical structure, the authors gain extra information that the GRH alone does not provide, and they use this to derive tighter estimates for both problems than were previously achievable under GRH alone.
In practical terms, the results mean that if you are willing to assume both the GRH and a reasonable hypothesis about how zeros are spaced, you can get sharper guarantees about how far out you need to search before finding the smallest prime in a given arithmetic progression, or the smallest quadratic non-residue. The work establishes a new and concrete connection between the statistical behavior of L-function zeros and these two longstanding classical problems, suggesting that deeper knowledge of zero spacing could unlock further improvements in many related areas of analytic number theory.