1: \begin{abstract}

2: In this paper, a domain decomposition method for the Poisson-Boltzmann (PB) solvation model that is widely used in computational chemistry is proposed.

3: This method, called ddLPB for short, solves the linear Poisson-Boltzmann (LPB) equation defined in $\mathbb R^3$ using the van der Waals cavity as the solute cavity.

4: The Schwarz domain decomposition method is used to formulate local problems  by decomposing the cavity into overlapping balls and only solving a set of coupled sub-equations in balls.

5: %{In the ddLPB, the LPB equation is first transformed into a system of two coupled equations defined in the (bounded) solute cavity, including a Poisson equation and a homogeneous screened Poisson (HSP) equation.}

6: %Then, a particular Schwarz domain decomposition method is used to solve these two equations, by decomposing the solute cavity into overlapping balls and only solving a set of coupled sub-equations in balls.

7: %A Laplace solver and a HSP solver in the unit ball are consequently developed for these sub-equations, which uses the spherical harmonics as basis functions.

8: %\red Since both the Laplace equation and the HSP equation have been well-studied in a ball, this results in analytical solutions on a local level. \black

9: %Finally, a global linear system is obtained {\red after the coupling conditions are imposed and discretized,} and an iterative linear solver based on the GMRES method is proposed.

10: A series of numerical experiments is presented to test the robustness and the efficiency of this method including the comparisons with some existing methods.

11: We observe exponential convergence of the solvation energy with respect to the number of degrees of freedom which allows this method to reach the required level of accuracy  when coupling with quantum mechanical descriptions of the solute.

12: \end{abstract}

13: