A path integral approximation of conditional probability densities with application to stochastic elastic rods

Abstract

In this work, we generalise Gelfand-Yaglom-type methods in the vector case for the computation of Gaussian path integrals. The extension we propose allows to consider general second variation operators subject to different boundary conditions and to regularise the divergence in presence of zero modes. The derived methods are exploited to study the statistical physics of polymers at thermodynamic equilibrium (e.g. DNA). The energy of equilibria combined with suitable Jacobi field determinants can be used to estimate the distribution of end-to-end displacements when the filament is interacting with a heat bath. In the continuum limit of Cosserat elastic rods, we demonstrate how to derive approximate conditional probability density functions governing the relative location and orientation of the two ends, first for the looping problem and second when the rod is subject to a prescribed external end-loading, in addition to external stochastic forcing. For isotropic Cosserat rods, certain looping boundary value problems admit non-isolated families of critical points of the energy due to an associated continuous symmetry, and the standard Laplace method fails for the presence of zero modes. Taking inspiration from (imaginary) path integral techniques, we show how a quantum mechanical probabilistic treatment of Goldstone modes in statistical rod mechanics sheds light on J-factor computations for isotropic rods in the semi-classical context. All the results are achieved exploiting appropriate Jacobi fields arising from Gaussian path integrals, and show good agreement when compared with intense Monte Carlo simulations for the target examples.