Wang-Landau sampling of equations of state – a new approach to input parameter optimization and uncertainty quantification

The construction of analytic equations of state (EOS) presents a formidable challenge due to the vast and complex high-dimensional parameter space involved. Traditional optimization methods struggle to navigate such landscapes efficiently, necessitating more advanced sampling techniques. Inspired by its success in computational statistical physics, we explore Wang-Landau sampling [1] as a novel approach to EOS parameter optimization and uncertainty quantification. Originally developed to determine the density of states in physical systems with complex free energy landscapes, Wang-Landau sampling systematically explores parameter spaces and avoids trapping in local minima. We demonstrate that this method offers a complementary perspective to conventional optimization techniques — enhancing the robustness of EOS modeling by providing deeper insights into parameter uncertainty and the structure of the error landscape — and can serve as a powerful tool for improving the accuracy and reliability of EOS models.

[1] F. Wang and D. P. Landau, Phys. Rev. Lett. 86, 2050 (2001)