The prime geodesic theorem is an analogue of the classical prime number theorem, but instead of counting prime numbers, it counts closed geodesics (shortest looping paths) on certain geometric spaces called Bianchi manifolds. These are three-dimensional hyperbolic spaces built from complex numbers with special arithmetic properties. Just as the prime number theorem comes with an error term measuring how precisely we can count primes, the prime geodesic theorem has an error term, and a central goal in the field is to make that error term as small as possible.
To improve the error bound, the authors attack a related problem in analytic number theory involving objects called symmetric square L-functions. These are complex-analytic functions that encode deep arithmetic information, and controlling their average size is key to the geometric counting problem. The authors prove what is called the "mean Lindelof hypothesis" for the second moment of these L-functions, meaning they establish the best plausible average bound for how large these functions can be. This is a significant result in its own right, as the Lindelof hypothesis itself remains unproven in general.
By combining this analytic result with careful techniques in spectral theory and automorphic forms, the authors derive an improved unconditional error bound in the prime geodesic theorem on Bianchi manifolds. The new bound, expressed as x to the power 36/23, improves on previous results in this three-dimensional hyperbolic setting. Because it is unconditional, it does not rely on any unproven conjectures, making it a concrete and rigorous advance in understanding the geometry and arithmetic of these manifolds.