Boundary singularities and boundary conditions for the Fokker-Planck equations

The boundary conditions for the Fokker-Planck equations, forward and backward ones are directly derived from the Chapman-Kolmogorov equation for M-dimensional region with boundaries. The boundaries are assumed, in addition, to be able to absorb wandering particles or to give rise to fast surface transport. It is demonstrated that the boundaries break down the symmetry of random walks in their vicinity, leading to the boundary singularities in the corresponding kinetic coefficients. Eliminating these singularities we get the desired boundary conditions. As it must be the boundary condition for the forward Fokker-Planck equation matches the mass conservation.