Path integrals for stochastic hybrid reaction-diffusion processes (2103.07759v2)

Abstract: We construct path integrals for stochastic hybrid reaction-diffusion (RD) processes, in which the reaction terms depend on the discrete state of a randomly switching environment. We proceed by spatially discretizing a given RD system and using a spinor representation of the environmental states to derive a path integral for the lattice model. In the case of large molecular numbers, the corresponding continuum path integral action is expressed in terms of an effective Hamiltonian, which involves a concentration field $u(\x,t)$, $\x\in \R^d$, a conjugate field $v(\x,t)$, and $M$ auxiliary conjugate pairs $(c_m(t),\phi_m(t))$, where $M$ is the number of discrete environmental states. The variable $c_m(t)$ determines the effective probability that a sample path is exposed to the $m$-th environmental state at time $t$, with $\sum_{m=1}^Mc_m(t)=1$. We then consider the semi-classical (adiabatic) limit $\epsilon \rightarrow 0$, where $\epsilon^{-1}$ determines the rate of switching between the environmental states. We show how the auxiliary variables can be eliminated to yield an action functional for the fields $u$ and $v$ alone. The associated Hamiltonian is the sum of a diffusion term and the Perron or principal eigenvalue of a functional linear operator involving the reaction terms and the matrix generator of the switching process. The reduced path integral is then used to derive a functional Hamilton-Jacobi equation for least action paths and to obtain a Gaussian noise approximation of the stochastic hybrid RD system in the adiabatic limit. Finally, the path integral in the case of low molecular numbers is constructed by considering a corresponding RD master equation. It is now necessary to take into account two sources of noise, one due to the switching environment and the other due to fluctuations in molecular numbers.