GLOBAL CONTINUATION FOR DISTANCE GEOMETRY PROBLEMS

Distance geometry problems arise in the determination of protein struc ture. We consider the case where only a subset of the distances betwee n atoms is given and formulate this distance geometry problem as a glo bal minimization problem with special structure. We show that global s moothing techniques and a continuation approach for global optimizatio n can be used to determine global solutions of this problem reliably a nd efficiently. The global continuation approach determines a global s olution with less computational effort than is required by a standard multistart algorithm. Moreover, the continuation approach usually find s the global solution from any given starting point, while the multist art algorithm tends to fail.