CeVNSR: Global optimization algorithm for MINLPs based on VNS using a...

View source: R/CeVNSR.R

CeVNSRR Documentation

Global optimization algorithm for MINLPs based on VNS using a Cooperative Strategy

Description

Solves optimization problems with intenger variables. Using several cooperative instances of VNS.

Usage

	CeVNSR(	problem, opts, max_eval = Inf, max_time = Inf,
			n_iter = 1, is_parallel = TRUE, type = "SOCKS",
			global_save_list = NULL, ...)

Arguments

problem

List containing problem settings.

opts

A list of n_threads lists containing options for each cooperative instance of essR.

max_eval

Maximum bumber of evaluations. Default is Inf.

max_time

Maximum time, default is Inf.

n_iter

Number of cooperative iterations. Default is 0 which is the same as running multiple single thread (as many as n_cpus) optimization runs.

is_parallel

Default is TRUE. Sometimes this it is useful to use as FALSE for debugging.

type

Choose between "SOCKS" and "MPI". Default is "SOCKS" (socket-connection). If you are using "SOCKS" option and you want to run multiple cpus in different machines you must specify the adress of each machine in hosts.

"MPI" mode requires you to have Rmpi installed.

global_save_list

Specify the names of global variables to be exported.

...

Additional variables.

Details

problem[[ith_thread]]=VNS_problem; opts[[ith_thread]]=VNS_opts;

VNS_problem and VNS_opts correspond to lists as seen in the rvnds_hamming documentation.

Value

f_mean

Vector with size of n_iter+1 containing the mean value of the objective function in each iteration.

f_sd

Vector with size of n_iter+1 containing the standard deviation value of the objective function in each iteration.

fbest

Vector with size of n_iter+1 containing the best value of the objective function in each iteration.

iteration_res

A list containing the results from every VNS instance initialized. It follows the format: results$iteration_res[[iteration+1]][[thread_number]].

numeval

Vector with size of n_iter+1 containing the number objective function evaluations at the end of each iteration.

time

Vector with size of n_iter+1 containing the time spent at the end of an iteration.

x_sd

A list containing the standard deviation of decision each variable at the end of an iteration. It follows the format: results$iteration_res[[iteration+1]][[thread_number]]

xbest

A list containing the best set of decision variables found and the end of each iteration.

See Also

rvnds_hamming MEIGO


jaegea/MEIGOR documentation built on April 8, 2024, 9:36 a.m.