R/234_reductions_solvers_constant_solver.R

In CVXR: Disciplined Convex Optimization

#####
## DO NOT EDIT THIS FILE!! EDIT THE SOURCE INSTEAD: rsrc_tree/reductions/solvers/constant_solver.R
#####

## CVXPY SOURCE: reductions/solvers/constant_solver.py
## ConstantSolver -- handles problems with zero variables
##
## When a problem has no variables, we evaluate constraints directly
## without going through the full reduction pipeline.


CONSTANT_SOLVER <- "CONSTANT_SOLVER"

ConstantSolver <- new_class("ConstantSolver", parent = Solver, package = "CVXR",
  constructor = function() {
    new_object(S7_object(),
      .cache = new.env(parent = emptyenv()),
      MIP_CAPABLE = TRUE,
      BOUNDED_VARIABLES = FALSE
    )
  }
)

method(solver_name, ConstantSolver) <- function(x) CONSTANT_SOLVER

method(reduction_accepts, ConstantSolver) <- function(x, problem, ...) {
  length(variables(problem)) == 0L
}

## apply: pass the problem through unchanged
## CVXPY SOURCE: constant_solver.py lines 15-16
method(reduction_apply, ConstantSolver) <- function(x, problem, ...) {
  list(problem, list())
}

## invert: return solution unchanged
## CVXPY SOURCE: constant_solver.py lines 18-19
method(reduction_invert, ConstantSolver) <- function(x, solution, inverse_data, ...) {
  solution
}

## solve_via_data: evaluate constraints on the constant problem
## CVXPY SOURCE: constant_solver.py lines 30-37
method(solve_via_data, ConstantSolver) <- function(x, data, warm_start = FALSE, verbose = FALSE,
                                                   solver_opts = list(), ...) {
  ## data is the Problem object (passed through from apply())
  problem <- data
  feasible <- all(vapply(problem@constraints, function(c) value(c), logical(1L)))
  if (feasible) {
    opt_val <- value(problem@objective)
    Solution(OPTIMAL, opt_val, list(), list(), list())
  } else {
    Solution(INFEASIBLE, NULL, list(), list(), list())
  }
}