LINCOA

LINCOA (LINearly Constrained Optimization Algorithm) is a numerical optimization algorithm by Michael J. D. Powell. It is also the name of Powell's Fortran 77 implementation of the algorithm.

LINCOA solves linearly constrained optimization problems without using derivatives of the objective function, which makes it a derivative-free algorithm. The algorithm solves the problem using a trust region method that forms quadratic models by interpolation. One new point is computed on each iteration, usually by solving a trust region subproblem subject to the linear constraints, or alternatively, by choosing a point to replace an interpolation point that may be too far away for reliability. In the second case, the new point may not satisfy the linear constraints.

The same as NEWUOA, LINCOA constructs the quadratic models by the least Frobenius norm updating ^[1] technique. A model function is determined by interpolating the objective function at (an integer between and ) points; the remaining freedom, if any, is taken up by minimizing the Frobenius norm of the change to the model's Hessian (with respect to the last iteration).

LINCOA software was released on December 6, 2013.^[2] In the comment of the source code,^[3] it is said that LINCOA is not suitable for very large numbers of variables (which is typically true for algorithms not using derivatives), but "a few calculations with 1000 variables, however, have been run successfully overnight, and the performance of LINCOA is satisfactory usually for small numbers of variables."^[3] It is also pointed out that the author's typical choices of are and , the latter "being recommended for a start", and "larger values tend to be highly inefficent when the number of variables is substantial, due to the amount of work and extra difficulty of adjusting more points."^[3]

The trust region subproblem is solved by the truncated conjugate gradient method described in Powell's report,^[4] but Powell did not write a report on the other details of LINCOA.

The LINCOA software is distributed under The GNU Lesser General Public License (LGPL).^[3]

See also

• TOLMIN
• COBYLA
• UOBYQA
• NEWUOA
• BOBYQA

References

External links

• Optimization software by Professor M. J. D. Powell at CCPForge

Optimization: Algorithms, methods, and heuristics
Unconstrained nonlinear: Methods calling … … functions Golden section search Interpolation methods Line search Nelder–Mead method Successive parabolic interpolation … and gradients Convergence Trust region Wolfe conditions Quasi–Newton BFGS and L-BFGS DFP Symmetric rank-one (SR1) Other methods Gauss–Newton Gradient Levenberg–Marquardt Conjugate gradient Truncated Newton … and Hessians Newton's method
Constrained nonlinear General Barrier methods Penalty methods Differentiable Augmented Lagrangian methods Sequential quadratic programming Successive linear programming
Convex optimization Convex minimization Cutting-plane method Reduced gradient (Frank–Wolfe) Subgradient method Linear and quadratic Interior point Affine scaling Ellipsoid algorithm of Khachiyan Projective algorithm of Karmarkar Basis-Exchange Simplex algorithm of Dantzig Revised simplex algorithm Criss-cross algorithm Principal pivoting algorithm of Lemke
Combinatorial Paradigms Approximation algorithm Dynamic programming Greedy algorithm Integer programming Branch & bound or cut Graph algorithms Minimum spanning tree Bellman–Ford Borůvka Dijkstra Floyd–Warshall Johnson Kruskal Network flows Dinic Edmonds–Karp Ford–Fulkerson Push–relabel maximum flow
Metaheuristics Evolutionary algorithm Hill climbing Local search Simulated annealing Tabu search
Categories Algorithms and methods Heuristics Software