学术文献库

We implement a discretization of the one-dimensional branching Brownian motion in the form of a Monte Carlo event generator, designed to efficiently produce ensembles of realizations in which the rightmost lead particle at the final time $T$ is constrained to have a position $X$ larger than some predefined value $X_{\text{min}}$. The latter may be chosen arbitrarily far from the expectation value of $X$, and the evolution time after which observables on the particle density near the lead particle are measured may be as large as $T\sim 10^4$. We then calculate numerically the probability distribution $p_n(Δx)$ of the number $n$ of particles in the interval $[X-Δx,X]$ as a function of $Δx$. When $X_{\text{min}}$ is significantly smaller than the expectation value of the position of the rightmost lead particle, i.e. when $X$ is effectively unconstrained, we check that both the mean and the typical values of $n$ grow exponentially with $Δx$, up to a linear prefactor and to finite-$T$ corrections. When $X_{\text{min}}$ is picked far ahead of the latter but within a region extending over a size of order $\sqrt{T}$ to its right, the mean value of the particle number still grows exponentially with $Δx$, but its typical value is lower by a multiplicative factor consistent with $e^{-ζΔx^{2/3}}$, where $ζ$ is a number of order unity. These numerical results bring strong support to recent analytical calculations and conjectures in the infinite-time limit.