FICO
FICO Xpress Optimization Examples Repository
FICO Optimization Community FICO Xpress Optimization Home
Back to examples browserPrevious exampleNext example

Set controls of Xpress

Description

Learn about setting controls of Xpress



Further explanation of this example: Xpress R Reference Manual

set_controls_Rmd.zip[download all files]

Source Files
By clicking on a file name, a preview is opened at the bottom of this page.
set_controls.R[download]

Data Files





set_controls.R

#####################################
# This file is part of the          #
# Xpress-R interface examples       #
#                                   #
#   (c) 2021 Fair Isaac Corporation #
#####################################
#' ---
#' title: "Set Controls"
#' author: Gregor Hendel
#' date: Dec. 2020
#' ---
#'

#'
#' This example shows how to set controls of the FICO Xpress Optimizer. A list of
#' all public controls can be found in the [Controls Reference](reference_controls.html).
#'
#' One important thing first: You can access
#' all controls and attributes with a special syntax in R: Use `xpress:::ROWS`
#' (yes, 3 colons) to access the integer ID of the rows attribute of the optimizer.
#' We provide the integer IDs for controls and attributes in this way such that
#' they do not pollute the namespace upon loading the `xpress` package.
#'
#' If you haven't done so already, please familiarize yourself with the Facility
#' Location Example, which we use throughout our quick examples.
#'
#' We read the Facility location problem from the introductory example, and print
#' it to verify its dimensions.
#'
## ----Reading the Facility Location Problem------------------------------------
suppressMessages(library(xpress))
p <- createprob()
print(readprob(p, "flp.lp"))

#'
#' # Setting Limits
#'
#' Most often, users seek to limit the solution process, for example a time limit,
#' node limit, or LP iterations limit.
#'
#' Here is how to set a limit on:
#'
#' - the maximum allowed wallclock time of 10 seconds, or
#' - the number of allowed branch-and-bound nodes, or
#' - the number of solutions.
#'
## ----Setting Limits on Time, Nodes, and Solutions-----------------------------
setintcontrol(p, xpress:::MAXTIME, -10L)
setintcontrol(p, xpress:::MAXNODE, 1L)
setintcontrol(p, xpress:::MAXMIPSOL, 1L)
dumpcontrols(p)

#'
#' The alert reader will notice that we set a *negative* MAXTIME of -10 seconds.
#' This is the right approach to stop the Optimizer after 10 seconds *regardless of
#' whether a solution has been found or not*. A positive limit of 10 seconds stops
#' the optimizer after 10 seconds if a solution was found by then, or if not, as
#' soon as a solution is found.
#'
#'
## ----Starting the Optimization Process----------------------------------------
# enable logging to stdout
setoutput(p)

summary(mipoptimize(p)) # we could also call "xprs_optimize()"
print(getintattrib(p, xpress:::MIPSTATUS))

#' In this case, we hit the solution limit and interrupt the solution process.
#' The first solution has an objective value of
#' `r format(getdblattrib(p, xpress:::MIPOBJVAL))`,
#' which is
#' `r format(getdblattrib(p, xpress:::MIPOBJVAL) - getdblattrib(p, xpress:::BESTBOUND))`
#' larger than the
#' dual bound of `r format(getdblattrib(p, xpress:::BESTBOUND))`.
#'
#' # Resetting Controls
#'
#' We decide to set the solution and node limit back to their default values, and
#' continue the search:
#'
## ----Setting Default Controls-------------------------------------------------
setdefaultcontrol(p, xpress:::MAXMIPSOL)
setdefaultcontrol(p, xpress:::MAXNODE)

## continue the solution process
summary(mipoptimize(p))

#' In order to set all controls to their default values, we invoke
## ----Resetting all controls---------------------------------------------------
setdefaults(p)

#'
#' We verify that no controls are set via `dumpcontrols`
## ----Output User-set Controls to the Console----------------------------------
dumpcontrols(p)

#'
#' # Change Solver Behavior
#'
#' After we familiarized ourselves with limits on the solution process, we can change some algorithmic behavior.
#'
#' - the [HEURSEARCHFREQ](https://www.fico.com/fico-xpress-optimization/docs/latest/solver/optimizer/HTML/HEURSEARCHFREQ.html)
#'   parameter specifies the frequency at which heuristics are used in the tree
#'   search
#' - the [CUTSTRATEGY](https://www.fico.com/fico-xpress-optimization/docs/latest/solver/optimizer/HTML/CUTSTRATEGY.html)
#'   parameter specifies the aggressiveness with which cutting planes are separated
#' - the [SBITERLIMIT](https://www.fico.com/fico-xpress-optimization/docs/latest/solver/optimizer/HTML/SBITERLIMIT.html)
#'   parameter specifies a user limit on the number of Simplex iterations in each
#'   Strong Branching LP
#'
## ----Change Algorithmic Parameters--------------------------------------------

# the frequency at which heuristics are used in the tree search
setintcontrol(p, xpress:::HEURSEARCHFREQ, 2L)

# disable all cutting planes
setintcontrol(p, xpress:::CUTSTRATEGY, 0)

# use a user limit on the number of Simplex iterations in each Strong Branching LP
setintcontrol(p, xpress:::SBITERLIMIT, 200)
summary(mipoptimize(p))

#'

Back to examples browserPrevious exampleNext example