Abstract: A Mach-Zehnder (MZ) modulator-based Ising machine simulation algorithm was proposed to efficiently solve combinatorial optimization problems, addressing the exponential growth of computation time with problem size in traditional enumeration algorithms. The interference output characteristics of MZ modulators were simulated using cosine functions, Gaussian random numbers were introduced to simulate system noise, and numerical iterations were employed to simulate the dynamic evolution process. Maximum-cut problems were tested on regular networks, small-world networks, and random networks with 16 and 100 vertices. The results demonstrate that the algorithm achieves 100% success rate for 16-vertex networks and maintains 88% success rate for 100-vertex random networks. In terms of computational efficiency, the Ising algorithm requires only 0.42 s for 25-vertex regular networks, showing significant advantages compared to the 29.93 s needed by the enumeration method. This study provides not only an efficient solution for complex network optimization problems but also theoretical references for the experimental design of MZ Ising machines.