Preprint D76/2010
A variant of projected gradient method for quasi-convex optimization problems with a competitive search strategy

L. R. Lucambio Perez | Bello Cruz, J. Y.

**Keywords: **
Quasi-convex optimization | Armijo search | Projected gradient method

We present the projected gradient method for solving constrained
quasi-convex minimization problem with a competitive search
strategy, i.e., an appropriate stepsize rule through an
Armijo-search along feasible direction obtaining global
convergence properties. Differently from other similar stepsize
rule, we perform only one projection onto the feasible set per
iteration, rather than one projection for each tentative step
during the search of the stepsize, which represents a considerable saving when the projection is computationally expensive.