pipette
ENEnglish

Minimax and adaptive inference for branching ratios under long memory

Mauricio Herrera-Mar\'in

Preprint

En palabras de los autores

The branching ratio of a self-exciting point process determines the branching margin to the critical boundary. We study inference on from a finite record for stationary Hawkes processes with completely monotone kernels whose slow relaxation mass obeys an envelope near zero rate. Slow excitation can then be exchanged for exogenous immigration; a spectral bound on the Kullback--Leibler rate and a coupling argument for the unobserved past give a minimax lower bound of order for a power-law envelope with tail exponent . A block-dispersion estimator whose block length grows as attains this rate. A quantitative central limit theorem, obtained from the cluster representation and a second-order Poincar\'e inequality on Poisson space, yields honest confidence intervals of nearly minimax expected length. For unknown , a Lepski-type rule adapts in probability at a logarithmic cost; free adaptation between memory classes is impossible, and honest intervals cannot adapt to lighter memory. Simulations are consistent with the rates.

Resultado principalLimitación que admiten los autores

Apareció: jueves, 1 de octubre. arXiv. Preprint, todavía sin revisión por pares.