1: \begin{abstract}

2: The radiation (reaction, Robin) boundary condition for the diffusion

3: equation is widely used in chemical and biological applications to

4: express reactive boundaries. The underlying trajectories of the

5: diffusing particles are believed to be partially absorbed and

6: partially reflected at the reactive boundary, however, the relation

7: between the reaction constant in the Robin boundary condition and

8: the reflection probability is not well defined. In this paper we

9: define the partially reflected process as a limit of the Markovian

10: jump process generated by the Euler scheme for the underlying It\^o

11: dynamics with partial boundary reflection. Trajectories that cross

12: the boundary are terminated with probability $P\sqrt{\Delta t}$ and

13: otherwise are reflected in a normal or oblique direction. We use

14: boundary layer analysis of the corresponding master equation to

15: resolve the non-uniform convergence of the probability density

16: function of the numerical scheme to the solution of the

17: Fokker-Planck equation in a half space, with the Robin constant

18: $\kappa$. The boundary layer equation is of the Wiener-Hopf type. We

19: show that the Robin boundary condition is recovered if and only if

20: trajectories are reflected in the co-normal direction

21: $\mb{\sigma}\mb{n}$, where $\mb{\sigma}$ is the (possibly

22: anisotropic) constant diffusion matrix and $\n$ is the unit normal

23: to the boundary. Otherwise, the density satisfies an oblique

24: derivative boundary condition. The constant $\kappa$ is related to

25: $P$ by $\kappa= rP\sqrt{\sigma_n}$, where $r=1/\sqrt{\pi}$ and

26: $\sigma_n=\n^T\mb{\sigma}\n$. The reflection law and the relation

27: are new for diffusion in higher-dimensions. %The constant $r$ for the

28: %Euler scheme is not the same as that for other schemes, e.g., for a

29: %discrete random walk with radiation boundaries, $r=1/\sqrt{2}$.

30: \end{abstract}

31: