1: \begin{abstract}

2: Approximate message-passing (AMP) method is a simple and efficient

3: framework for the linear inverse problems. In this letter, we

4: propose a faster AMP to solve the \emph{$L_1$-Split-Analysis} for

5: the 2D sparsity separation, which is referred to as \emph{MixAMP}.

6: We develop the MixAMP  based on the factor graphical modeling and

7: the min-sum message-passing. Then, we examine MixAMP for two types

8: of the sparsity separation: separation of the direct-and-group

9: sparsity, and that of the direct-and-finite-difference sparsity.

10: This case study shows that the MixAMP method offers computational

11: advantages over the conventional first-order method, TFOCS.

12: \end{abstract}

13: