paper

Subleading-order theory for condensation transitions in large deviations of sums of independent and identically distributed random variables

arXiv:2503.19581 · doi:10.1088/1742-5468/ade5fa

Abstract

We study the full distribution of sums where are independent and identically distributed random variables each sampled from a given distribution with a subexponential tail. We consider two particular cases: (I) the one-sided stretched exponential distribution where , (II) the two-sided stretched exponential distribution where . We assume (in both cases). As follows immediately from known theorems, for both cases (i) typical fluctuations of are described by the central-limit theorem, (ii) the tail is described by the big-jump principle , and (iii) in between these two regimes there is a nontrivial intermediate regime which displays anomalous scaling with anomalous exponents and large-deviation function that are all exactly known. In practice, although these theoretical predictions of work very well in regimes (i) and (ii), they often perform quite poorly in the intermediate regime (ii), with errors of several orders of magnitude for as large as . We calculate subleading order corrections to the theoretical predictions in the intermediate regime. We find that for , these corrections scale as power laws in , while for they scale as stretched exponentials, where the threshold value is in case (I) and in case (II). This difference between the two cases is a result of the mirror symmetry which holds only in the latter case.

19 pages, 7 figures

References in corpus (23)

Cited by in corpus (1)