ROI.plugin.alabama (version 0.3-1)

control: alabama

Description

This package provides a ROI plugin to the alabama package. The following description of the control parameters is mostly copied from the alabama manual.

  • [start:] The initial values for the parameter vector.

  • [method:] Unconstrained optimization algorithm for inner loop optimization. Allowed values are "Nelder-Mead", "BFGS", "CG", "L-BFGS-B", "SANN", "Brent" and "nlminb".

  • [lam0:] Initial value for the Lagrangian parameter.

  • [sig0:] A scaling parameter for penalty term that is augmented to the Lagrangian.

  • [tol:] Tolerance for convergence of outer iterations of the barrier and/or augmented lagrangian algorithm

  • [max_iter:] Maximum number of outer iterations.

  • [ilack.max:] Maximum number of outer iterations where no change in parameters is tolerated.

  • [verbose:] A logical variable indicating whether information on outer iterations should be printed out. If TRUE, at each outer iteration information is displayed on: (i) how well the inequality and equalities are satisfied, (ii) current parameter values, and (iii) current objective function value.

  • [NMinit:] A logical variable indicating whether "Nelder-Mead" algorithm should be used for the first outer iteration.

  • [i.scale:] A vector of length equal to number of inequalities that may be used to scale the inequalities or it can be a scalar in which case all the inequalities are scaled by the same value.

  • [e.scale:] A vector of length equal to number of equalities that may be used to scale the equalities or it can be a scalar in which case all the equalities are scaled by the same value.

  • [kkt2.check:] A logical variable (TRUE/FALSE) indicating whether the second-order KKT condition should be checked. Deafult is TRUE. It may be set to FALSE in problems where the Hessian computation can b etime consuming.

  • [control.optim:] A list of control parameters to be used by the unconstrained optimization algorithm in the inner loop. Identical to that used in optim or in nlminb.

Arguments

References

Ravi Varadhan (2015). alabama: Constrained Nonlinear Optimization. R package version 2015.3-1. https://CRAN.R-project.org/package=alabama

Examples

Run this code
# NOT RUN {
library(ROI)

n <- 2L
x <- OP(F_objective(sum, n = n), 
        bounds = V_bound(nobj = 2, ld = -1, ud = 1))

control_optim <- list(trace = 0, fnscale = 1, parscale = rep.int(1, n), 
                      ndeps = rep.int(0.001, n), maxit = 100L, abstol = -Inf, 
                      reltol = sqrt(.Machine$double.eps), alpha = 1, 
                      beta = 0.5, gamma = 2, REPORT = 10, type = 1, lmm = 5,
                      factr = 1e+07, pgtol = 0, tmax = 10, temp = 10)

control <- list(start = c(0, 0), method = "BFGS", lam0 = 10, sig0 = 100,  
                tol = 1e-07, max_iter = 50, verbose = FALSE, NMinit = FALSE, 
                ilack.max = 6, i.scale = 1, e.scale = 1, kkt2.check = TRUE,
                control.optim = control_optim)

s <- ROI_solve(x, solver = "alabama", control)
# }

Run the code above in your browser using DataCamp Workspace