This paper proves that some quadratic Dirichlet L-functions take much larger values at the central point than previously shown, under the assumption of the Generalized Riemann Hypothesis (GRH). Concretely, for large X the authors show there is a fundamental discriminant d with X<|d|≤2X for which L(1/2, χ_d) is at least exp((1+o(1)) sqrt((log X)(log_3 X)/(log_2 X))). Here log_2 X means log log X and log_3 X means log log log X.

The goal is to produce a lower bound for the maximum central value in the family of quadratic Dirichlet L-functions. The authors adapt the resonator method, a way to amplify large values by pairing L-values with a carefully chosen short Dirichlet polynomial. They use an approximate functional equation that expresses L(1/2, χ_d) as a weighted sum of the quadratic character χ_d(n) over n. By summing the square of the resonator and the resonator times L(1/2, χ_d) over discriminants d, they reduce the problem to estimating two averages and then comparing them.

Two technical inputs are crucial. First, under GRH the paper uses conditional mean-value estimates for quadratic characters to control off-diagonal terms when averaging over d. Second, the authors build the resonator from a specially chosen set M of square-free integers. The effectiveness of this choice rests on a GCD (greatest common divisor) sum estimate from earlier work of de la Bretèche and Tenenbaum. Combined, these tools make the average ratio large enough to force the existence of an individual d with the claimed large L-value.

Why this matters: extreme values of L-functions probe the limits of our knowledge about these central objects in number theory. Improving the constant from 1/2 to 1 in the exponent (compared with the recent work of Darbar and Maiti) is a clear quantitative advance for this family. It brings the known lower bound closer to analogous results for other families of L-functions and highlights the role of resonator constructions and GCD sums.