1: \begin{abstract}

2:   Message-passing (MP) is a powerful tool for finding an

3:   approximate solution in optimization.

4:   We generalize it to nonlinear product-sum form, and numerically

5:   show the fast convergence for the minimum feedback vertex set

6:   and the minimum vertex cover known as NP-harp problems.

7:   From the linearity of MP in a logarithmic space,

8:   it is derived that an equilibrium solution exists in

9:   a neighborhood of random initial values.

10:   These results will give one of the reason why

11:   the convergence is very fast in collective computation

12:   based on a common mathematical background.

13: \end{abstract}

14: