First-passage exponents of multiple random walks

We investigate first-passage statistics of an ensemble of N noninteracting random walks on a line. Starting from a configuration in which all particles are located in the positive half-line, we study S-n(t), the probability that the nth rightmost particle remains in the positive half-line up to time t. This quantity decays algebraically, S-n(t) similar to t(-beta n), in the long-time limit. Interestingly, there is a family of nontrivial first-passage exponents, beta(1) < beta(2) < ... < beta(N-1); the only exception is the two-particle case where beta(1) = 1/3. In the N -> infinity limit, however, the exponents attain a scaling form, beta(n)(N) -> beta(z) with z = (n - N/2)/root N. We also demonstrate that the smallest exponent decays exponentially with N. We deduce these results from first-passage kinetics of a random walk in an N-dimensional cone and confirm them using numerical simulations. Additionally, we investigate the family of exponents that characterizes leadership statistics of multiple random walks and find that in this case, the cone provides an excellent approximation.