Abstract
A new optimization protocol is proposed which is based on a combination of a mean field approximation and simulated annealing. Instead of optimizing the energy of the real system the energy of a new mean field system is minimized. The global minimum of the new system and the original system is the same. The mean‐field optimization is advantageous to the optimization of the real system since (a) More statistics are obtained for alternative solutions and (b) the barrier heights separating the minima are reduced compared to the real system. Computational examples are provided for placement of side chains in tetrapeptides and in a small protein BPTI.