Package {ROI.plugin.coinclp}

Type: Package
Title: 'COIN-OR' 'Clp' Plugin for the 'R' Optimization Infrastructure
Version: 0.1.0
Description: Registers the 'COIN-OR' 'Clp' linear programming solver, through the 'coinclp' package, with the 'R' Optimization Infrastructure ('ROI'). Linear programs with continuous variables are then solved by ROI_solve(op, solver = "coinclp"), keeping sparse constraint matrices sparse all the way to the solver, and returning dual values, reduced costs and row activities alongside the primal solution.
License: EPL
URL: https://github.com/SamLovick/ROI.plugin.coinclp
BugReports: https://github.com/SamLovick/ROI.plugin.coinclp/issues
Encoding: UTF-8
Depends: R (≥ 4.0)
Imports: coinclp (≥ 0.1.0), methods, ROI (≥ 1.0-0), slam, stats
Suggests: Matrix, ROI.plugin.highs
NeedsCompilation: no
Config/roxygen2/version: 8.1.0
Packaged: 2026-09-15 15:10:23 UTC; samlo
Author: Sam Lovick [aut, cre]
Maintainer: Sam Lovick <sam@lovickconsulting.com>
Repository: CRAN
Date/Publication: 2026-09-28 08:10:13 UTC

Solve linear programs in ROI with COIN-OR Clp

Description

Registers the COIN-OR Clp solver, through the coinclp package, with the R Optimization Infrastructure. Loading this package is all the setup there is: linear programs can then be solved with ROI_solve(op, solver = "coinclp"), and "coinclp" joins the solvers ROI_solve() picks from automatically.

Details

The plugin accepts a linear objective, linear constraints, continuous variables and variable bounds, minimising or maximising. Constraints given as a simple_triplet_matrix, which is how ROI stores them, are passed to Clp in sparse form without being densified.

Solver controls

Pass these to ROI_solve as named arguments or in control:

log_level

How much Clp prints, 0 (silent, the default) to 4. Also accepted as amount or verbosity_level, the names the earlier ROI.plugin.clp used.

max_iterations

Iteration limit; also iterations.

max_seconds

Time limit in seconds; also seconds.

algorithm

One of "auto" (the default), "primal", "dual", "barrier" or "barrier_nocross".

presolve

Run Clp's presolve, TRUE by default.

primal_tolerance, dual_tolerance

Simplex tolerances.

scaling

Scaling mode: 0 off, 1 equilibrium, 2 geometric, 3 auto, 4 dynamic.

Solution

Besides the primal solution and the objective value, solution(x, "dual") gives the dual value of each constraint and solution(x, "aux") the row activities and reduced costs. solution(x, "msg") returns the whole coinclp result, including the iteration count.

Relation to ROI.plugin.clp

This package replaces ROI.plugin.clp, which was archived from CRAN in January 2022 because the clpAPI package it depended on had been archived. The code here is independent of that plugin; the control names amount, verbosity_level, iterations and seconds are kept so that scripts written for it still work, but the solver is now named "coinclp" rather than "clp".

Author(s)

Maintainer: Sam Lovick sam@lovickconsulting.com

Authors:

See Also

Useful links:

Examples

library(ROI)
op <- OP(objective = c(143, 60),
         constraints = L_constraint(rbind(c(120, 210), c(110, 30), c(1, 1)),
                                    rep("<=", 3), c(15000, 4000, 75)),
         maximum = TRUE)
result <- ROI_solve(op, solver = "coinclp", log_level = 0)
solution(result)
solution(result, "objval")

Dual values and auxiliary quantities

Description

solution(x, "dual") returns the dual value of each constraint, the shadow price of its right hand side. solution(x, "aux") returns a list with the row activities as primal and the reduced costs of the variables as dual.

Usage

## S3 method for class 'coinclp_solution'
ROI_plugin_solution_dual(x)

## S3 method for class 'coinclp_solution'
ROI_plugin_solution_aux(x)

Arguments

x

A solution returned by ROI_solve with solver = "coinclp".

Details

Signs follow Clp's own convention for the direction that was solved, so a maximisation and the equivalent minimisation of the negated objective report duals of opposite sign.

Value

A numeric vector of dual values, or a list of auxiliary vectors.

Examples

library(ROI)
op <- OP(objective = c(143, 60),
         constraints = L_constraint(rbind(c(120, 210), c(110, 30), c(1, 1)),
                                    rep("<=", 3), c(15000, 4000, 75)),
         maximum = TRUE)
result <- ROI_solve(op, solver = "coinclp")
solution(result, "dual")
solution(result, "aux")

Solve a linear program with Clp

Description

The method ROI dispatches to when a linear program is solved with solver = "coinclp". It is not called directly; use ROI_solve.

Usage

solve_OP(x, control = list())

Arguments

x

An object of class "OP" holding a linear program.

control

A named list of solver controls, see ROI.plugin.coinclp for the ones recognised.

Value

An object of class "coinclp_solution", which solution unpacks.

Examples

library(ROI)
op <- OP(objective = c(143, 60),
         constraints = L_constraint(rbind(c(120, 210), c(110, 30), c(1, 1)),
                                    rep("<=", 3), c(15000, 4000, 75)),
         maximum = TRUE)
result <- ROI_solve(op, solver = "coinclp")
solution(result)
solution(result, "dual")