1: \begin{subequations}

2: \label{reducedmodel2}

3: \begin{equation}

4: \label{reactiony_reg2}

5: \ce{$S_-$  <=>[k_-][k_+] $S_+$}, \qquad

6: \ce{$\emptyset$  <=>[{\overline k_5}(S)/\tau_y][k_6/\tau_y] $Y$}

7: \end{equation}

8: where 

9: the effective production rate takes one of two values depending on

10: whether $X$ is in the left or right well, 

11: \begin{equation}

12:  {\overline k_5}(S_\pm) = \int_{\Omega_\pm } k_5 (x) \rho_\pm(x) \, \ud x;

13: \end{equation}

14: here $\rho_{\pm}$ is the normalised stationary density of $X$

15: conditional on being in the left (minus) or right (plus) well. This definition corresponds to imposing a reflecting boundary condition on $x_*$ in \eqref{fokker-planck_x}.\footnote{This is in contrast to the Parallel Replica Algorithm,\cite{LeBris:2012et} which defines quasi-stationary distributions using absorbing boundary conditions.}

16: A formal derivation of this model is given in \ref{sec:regime2}.

17: We denote this reduced system by \emph{reduced model 2} or RM2. The rate constants $k_\pm$ of the process for $S(t)$ are the inverse of the mean switching times $T_\pm$ for $X$ to jump from one well to the other [the switching timescale $\tau_s$ is such that $T_\pm \sim O (\tau_s)$]. Determining these rates accurately is one of the main challenges of Regime 2; we will return to this issue in section \ref{sec:mst}. For the chemical $X(t)$, we will find that 

18: \begin{equation}

19: \label{mst_chem1}

20: k_- = 0.0931, \qquad k_+ = 0.0835. 

21: \end{equation}

22: \end{subequations}

23: