Skip to contents

Unified solver wrapper

Source: R/solvers.R
solve_optimization.Rd

A unified interface for solving the optimization problem using either Gurobi or HiGHS solver.

Usage

solve_optimization(
  Z,
  block,
  weight,
  coeflists,
  p,
  ms_list,
  exact = TRUE,
  block.sum = NULL,
  solver = "auto"
)

Arguments

Z

An n-dimensional treatment assignment vector.

block

An n-dimensional vector specifying block of each unit.

weight

A B-dimensional vector of block weights.

coeflists

A list of H elements, each containing B matrices with block-specific test statistics for different values of k.

p

Upper bound on the number of units with effects greater than c.

ms_list

A list containing mu (means) and sigma (standard deviations) for each test statistic.

exact

Logical; if TRUE, solve as integer linear program (ILP). If FALSE, solve as linear program (LP relaxation).

block.sum

Optional pre-computed block summary from summary_block().

solver

Character string specifying the solver to use. Options are:

  • "auto": Automatically select available solver (default)

  • "highs": Use HiGHS solver (open-source)

  • "gurobi": Use Gurobi solver (commercial, requires license)

Value

A list with:

sol

Solution vector

obj

Optimal objective value

solver

The solver that was used

Details

This function provides a unified interface for both solvers. When solver = "auto", it will prefer Gurobi if available (for backward compatibility), otherwise fall back to HiGHS.

Both solvers produce equivalent results for the same optimization problem, though there may be minor numerical differences (typically < 1e-6) due to different internal algorithms.

See also

HiGHS_sol_com, Gurobi_sol_com, solver_available, get_default_solver

Examples

if (FALSE) { # \dontrun{
# Automatically select solver
result <- solve_optimization(Z, block, weight, coeflists, p, ms_list)

# Explicitly use HiGHS
result <- solve_optimization(Z, block, weight, coeflists, p, ms_list, solver = "highs")

# Explicitly use Gurobi
result <- solve_optimization(Z, block, weight, coeflists, p, ms_list, solver = "gurobi")
} # }