The paper introduces a new algorithm called Very Fast Simulated Re-Annealing (VFR) for statistically finding the best global fit of a nonlinear non-convex cost-function over a \(D\)-dimensional space. The algorithm is designed to be faster than previous methods such as Boltzmann annealing and fast Cauchy annealing. The key innovation is the exponential decrease in the "temperature" \(T\) with annealing time \(k\), given by \(T = T_0 \exp(-c k^{1/D})\). This allows for adaptive annealing schedules that can handle changing sensitivities in the multi-dimensional parameter space. The paper also discusses the application of VFR to fitting empirical data to Lagrangians representing nonlinear Gaussian-Markovian systems, including examples from physics, neuroscience, and combat scenarios. The method is particularly useful for problems with finite parameter ranges and varying sensitivities, and it provides a sensitive separation of alternative models based on long-time correlations. The authors conclude that VFR is a powerful tool for fitting complex nonlinear models to empirical data.The paper introduces a new algorithm called Very Fast Simulated Re-Annealing (VFR) for statistically finding the best global fit of a nonlinear non-convex cost-function over a \(D\)-dimensional space. The algorithm is designed to be faster than previous methods such as Boltzmann annealing and fast Cauchy annealing. The key innovation is the exponential decrease in the "temperature" \(T\) with annealing time \(k\), given by \(T = T_0 \exp(-c k^{1/D})\). This allows for adaptive annealing schedules that can handle changing sensitivities in the multi-dimensional parameter space. The paper also discusses the application of VFR to fitting empirical data to Lagrangians representing nonlinear Gaussian-Markovian systems, including examples from physics, neuroscience, and combat scenarios. The method is particularly useful for problems with finite parameter ranges and varying sensitivities, and it provides a sensitive separation of alternative models based on long-time correlations. The authors conclude that VFR is a powerful tool for fitting complex nonlinear models to empirical data.