Exact closed-form recurrence probabilities for biased random walks at any step number

Abstract: We report on a closed-form expression for the survival probability of a discrete 1D biased random walk to not return to its origin after N steps. Our expression is exact for any N, including the elusive intermediate range, thereby allowing one to study its convergence to the large N limit. In that limit we recover Polya's recurrence probability, i.e. the survival probability equals the magnitude of the bias. We then obtain a closed-form expression for the probability of last return. In contrast to the bimodal behavior for the unbiased case, we show that the probability of last return decays monotonically throughout the walk beyond a critical bias. We obtain a simple expression for the critical bias as a function of the walk length, and show that it saturates at $1/\sqrt{3}$ for infinitely long walks. This property is missed when using expressions developed for the large N limit. Finally, we discuss application to molecular motors' biased random walks along microtubules, which are of intermediate step number.