1 Implementation of the globally-convergent method-of-moving-asymptotes (MMA)
2 algorithm for gradient-based local optimization, as described in:
4 Krister Svanberg, "A class of globally convergent optimization
5 methods based on conservative convex separable approximations,"
6 SIAM J. Optim. 12 (2), p. 555-573 (2002).
8 In fact, this algorithm is much more general than most of the other
9 algorithms in NLopt, in that it handles an arbitrary set of nonlinear
10 (differentiable) constraints as well, in a very efficient manner.
11 I've implemented the full nonlinear-constrained MMA algorithm, and it
12 is exported under the nlopt_minimize_constrained API.
14 It is under the same MIT license as the rest of my code in NLopt (see