Gain optimization for enhanced solution quality of optoelectronic Ising machines

The Ising machine is an emerging hardware platform for combinatorial optimization, offering orders-of-magnitude speedups over conventional computing. However, due to its reliance on intrinsic physical dynamics, systematic strategies for improving solution quality remain limited, making optimal operation difficult to identify. Here, we investigate the optoelectronic Ising machine and systematically analyze the impact of pump gain on solution quality, considering both the final gain and the gain control scheme. Based on the system's evolutionary dynamics, we explain the experimentally observed behaviors and propose a gain-optimization strategy to determine the optimal operating point. Using this strategy, the solution quality of the optoelectronic Ising machine is improved for Max-Cut problems with graph densities of 10%, 50%, and 100%.