SIMULATED ANNEALING AND NEURAL NETWORKS AS ALTERNATIVE METHODS FOR NONLINEAR CONSTRAINED OPTIMIZATION

By highlighting a particular example of track fitting with geometrical and kinematical constraints of events from a particle physics experim ent simulated annealing and backprop neural networks are investigated as alternative optimization methods. The two methods are different in the way that simulated annealing suffers from the enormous amount of f unction evaluations required while a trained neural network requires n egligible computing time. This makes the former unpractical to use in analysis of large amounts of data, but we find that using any of these methods to compute start values has significant impact on the converg ence of the constrained fits, Further, it enables us to reduce and est imate the effects introduced by the optimization algorithm. For our ca se of kinematic and geometric fitting of <p(p)over bar> --> K(0)K(+/-) pi(+/-) Monte Carlo events we find that the rate of non-convergence, i n the case that the standard analytical minimization did not work usin g the measured values as the star? values, changes up to 40% when maki ng a second attempt with start values from a global optimizer.