PRACTICAL STEPS TOWARD REALIZING THE POTENTIAL OF MONTE-CARLO AUTOMATIC STATICS

This work develops a number of innovations to the simulated annealing Monte Carlo automatic statics technique. Frequency-domain processing reduces the need for computer memory and helps provide rich sampling. Realistic search limits drastically reduce the computational load and prevent wasted processing. A modification to the objective function provides more stability. Finally, a fast and simple algorithm is introduced to remove null space and long wavelength components from successive solutions. This algorithm is used in an automatic temperature control process to improve convergence. Together, these innovations make Monte Carlo automatic statics more efficient and easier to use.