6 - Algorithme de coupe énumérative pour la minimisation de fonction d'énergie

We study the 1-D reconstruction problem of real-valued data . This problem can be formulated by minimizing the energy functio n or hamiltonien . To solve this problem, many methods have been proposed such as, simulated annealing (SA), the graduate d nonconvexiiy (GNC), and mean field annealing (MFA) . We propose a new algorithm called "Discontinuity Position Sweep " (in French Balayage par Position de Discontinuite : (BPD)) . This new algorithm comes from cutting enumerative methods i n combinatorial optimization and its main advantages over the aforementioned methods are that it is parallelisable together, it i s faster and it finds the better solution .