Skip to contents

Set an objective function

Source: R/objective.R
lp_objective.Rd

Minimize of maximize an expression.

Usage

lp_minimize(.problem, objective)

lp_maximize(.problem, objective)

lp_min(.problem, objective)

lp_max(.problem, objective)

Arguments

.problem

An lp_problem().

objective

Expression to optimize, which must evaluate to an lp_variable object. Alternatively, set objective = 0 to make the solver find a feasible solution instead of optimizing, just like lp_find_feasible() does.

Value

The .problem with the updated $objective function, a list with these fields:

  • $coef : Vector with the coefficients for each variable.

  • $add : Numeric, addend to the final value. It is not used in the solver.

  • $direction : String, goal of the solver. Can be "minimize" or "maximize".

  • $expr : String, expression that defined the objective function.

Details

If objective evaluates to a multivariate variable instead of a scalar, it will apply sum(objective) and display a message. Suppress this message by writing the sum yourself.

Examples

profit   <- c(Phone = 60, Tablet = 20, eBook = 10)
max_made <- c(Phone = 500, Tablet = 300, eBook = 950)
product  <- names(profit)
fix_cost <- 11e3

p <- lp_problem() |>
    lp_variable(made[product], lower = 0, upper = max_made, integer = TRUE) |>
    lp_alias(total_profit = sum(made * profit)) |>
    lp_maximize(total_profit - fix_cost) |>
    lp_constraint(sum(made) <= 1500)

p$objective
#> maximize linear function:
#> total_profit - fix_cost
#> 

library(ROI) |> suppressMessages()
(s <- lp_solve(p))
#> $variables
#> $variables$made
#> product
#>  Phone Tablet  eBook 
#>    500    300    700 
#> 
#> 
#> $aliases
#> $aliases$total_profit
#> [1] 43000
#> 
#> 
#> $objective
#> [1] 32000
#> 
#> $status$code = 0  (Optimal)
#> 
#> Fields:
#> -- $objective --
#> -- $variables --
#> -- $aliases --
#> -- $variables_vec --
#> -- $status --
#> -- $message --
#> -- $model --

s$aliases$total_profit
#> [1] 43000
sum(p$objective$coef * s$variables_vec)
#> [1] 43000

s$objective
#> [1] 32000
sum(p$objective$coef * s$variables_vec) + p$objective$add
#> [1] 32000