The collection of results, conjectures and heuristic correspondences that link statistical properties of spectra arising in number theory (zeros of L‑functions, eigenvalues of families of operators) to eigenvalue statistics of random matrix ensembles (GUE, GOE, GSE and families), often after appropriate scaling.

Principle

Universality and symmetry classification: local statistics in the large‑scale limit depend primarily on symmetry type, so families of L‑functions with the same symmetry class exhibit matching scaling limits with corresponding random matrix ensembles; moments and correlation functions are predicted by ensemble averages.

Demonstration

Montgomery's pair correlation conjecture and Odlyzko's numerical verification: high zeros of the Riemann zeta function show spacing statistics matching the Gaussian Unitary Ensemble (GUE); Katz–Sarnak conjectures predict symmetry types for families of L‑functions over finite fields matching classical compact groups.

Misapplication

Assuming random matrix predictions determine individual zero locations or treating finite‑conductor arithmetic data as if it were already in the asymptotic scaling regime without verifying the limit, leading to incorrect pointwise claims.

Consequence

Provides precise conjectural descriptions for spacing distributions, value distribution of L‑functions, moments, and extreme value statistics; guides numerical experiments and formulation of refined conjectures about zeros and central values.

Reversal

Arithmetic data about families of L‑functions (e.g. monodromy groups, sign of functional equations) can determine the appropriate random matrix ensemble and thus suggest universality classes; conversely RMT suggests which statistical tests are informative for detecting arithmetic symmetry.

Boundary

Connections typically refer to asymptotic, local statistics in the high‑scale or large‑conductor limit and require averaging over families; they do not constitute proofs for individual L‑functions in finite ranges and do not replace rigorous, domain‑specific analytic methods.

Semantic Tension

Tension with classical probabilistic models (e.g. Poisson statistics) and with deterministic explanations: RMT gives a statistical, often universal description, which can conflict with models predicting independence or arithmetic rigidity in certain regimes.

Synthesis

Random matrix theory connections assert that, after appropriate normalization and averaging, the local statistical behavior of zeros and related arithmetic spectra matches universal ensembles determined by symmetry, providing a predictive framework for spacing, moments and extreme statistics while respecting its asymptotic and averaged scope.