Expected maximum of bridge random walks & Lévy flights
Benjamin De Bruyne,Satya N. Majumdar,Gregory Schehr
2021-05-20
Abstract:We consider one-dimensional discrete-time random walks (RWs) with arbitrary
symmetric and continuous jump distributions $f(\eta)$, including the case of
L\'evy flights. We study the expected maximum ${\mathbb E}[M_n]$ of bridge RWs,
i.e., RWs starting and ending at the origin after $n$ steps. We obtain an exact
analytical expression for ${\mathbb E}[M_n]$ valid for any $n$ and jump
distribution $f(\eta)$, which we then analyze in the large $n$ limit up to
second leading order term. For jump distributions whose Fourier transform
behaves, for small $k$, as $\hat f(k) \sim 1 - |a\, k|^\mu$ with a L\'evy index
$0<\mu \leq 2$ and an arbitrary length scale $a>0$, we find that, at leading
order for large $n$, ${\mathbb E}[M_n]\sim a\, h_1(\mu)\, n^{1/\mu}$. We obtain
an explicit expression for the amplitude $h_1(\mu)$ and find that it carries
the signature of the bridge condition, being different from its counterpart for
the free random walk. For $\mu=2$, we find that the second leading order term
is a constant, which, quite remarkably, is the same as its counterpart for the
free RW. For generic $0< \mu < 2$, this second leading order term is a growing
function of $n$, which depends non-trivially on further details of $\hat f
(k)$, beyond the L\'evy index $\mu$. Finally, we apply our results to compute
the mean perimeter of the convex hull of the $2d$ Rouse polymer chain and of
the $2d$ run-and-tumble particle, as well as to the computation of the survival
probability in a bridge version of the well-known "lamb-lion" capture problem.
Probability,Statistical Mechanics,Mathematical Physics