1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120

//! A solver that uses [minilp](https://docs.rs/minilp), a pure rust solver.

use crate::solvers::{ObjectiveDirection, ResolutionError, Solution, SolverModel};
use crate::variable::{UnsolvedProblem, VariableDefinition};
use crate::{Constraint, Variable};
use std::marker::PhantomData;

/// The [minilp](https://docs.rs/minilp) solver,
/// to be used with [UnsolvedProblem::using].
pub fn minilp<F>(to_solve: UnsolvedProblem<F>) -> MiniLpProblem<F> {
    let UnsolvedProblem {
        objective,
        direction,
        variables,
    } = to_solve;
    let mut problem = minilp::Problem::new(match direction {
        ObjectiveDirection::Maximisation => minilp::OptimizationDirection::Maximize,
        ObjectiveDirection::Minimisation => minilp::OptimizationDirection::Minimize,
    });
    let variables: Vec<minilp::Variable> = variables
        .iter_variables_with_def()
        .map(|(var, &VariableDefinition { min, max, .. })| {
            let coeff = *objective.linear.coefficients.get(&var).unwrap_or(&0.);
            problem.add_var(coeff, (min, max))
        })
        .collect();
    MiniLpProblem {
        problem,
        variables,
        variable_type: PhantomData,
    }
}

/// A minilp model
pub struct MiniLpProblem<F> {
    problem: minilp::Problem,
    variables: Vec<minilp::Variable>,
    variable_type: PhantomData<F>,
}

impl<T> MiniLpProblem<T> {
    /// Get the inner minilp model
    pub fn as_inner(&self) -> &minilp::Problem {
        &self.problem
    }
}

impl<T> SolverModel<T> for MiniLpProblem<T> {
    type Solution = MiniLpSolution<T>;
    type Error = ResolutionError;

    fn with(mut self, constraint: Constraint<T>) -> Self {
        let coefficients: Vec<(minilp::Variable, f64)> = constraint
            .expression
            .linear
            .coefficients
            .iter()
            .map(|(var, &coeff)| (self.variables[var.index()], coeff))
            .collect();
        let op = match constraint.is_equality {
            true => minilp::ComparisonOp::Eq,
            false => minilp::ComparisonOp::Le,
        };
        let constant = -constraint.expression.constant;
        self.problem.add_constraint(&coefficients, op, constant);
        self
    }

    fn solve(self) -> Result<Self::Solution, Self::Error> {
        match self.problem.solve() {
            Err(minilp::Error::Unbounded) => Err(ResolutionError::Unbounded),
            Err(minilp::Error::Infeasible) => Err(ResolutionError::Infeasible),
            Ok(solution) => Ok(MiniLpSolution {
                solution,
                variables: self.variables,
                _variable_type: PhantomData,
            }),
        }
    }
}

/// The solution to a minilp problem
pub struct MiniLpSolution<F> {
    solution: minilp::Solution,
    variables: Vec<minilp::Variable>,
    _variable_type: PhantomData<F>,
}

impl<F> MiniLpSolution<F> {
    /// Returns the MiniLP solution object. You can use it to dynamically add new constraints
    pub fn into_inner(self) -> minilp::Solution {
        self.solution
    }
}

impl<F> Solution<F> for MiniLpSolution<F> {
    fn value(&self, variable: Variable<F>) -> f64 {
        self.solution[self.variables[variable.index()]]
    }
}

#[cfg(test)]
mod tests {
    use super::minilp;
    use crate::{variable, variables, Solution, SolverModel};

    #[test]
    fn can_solve_easy() {
        let mut vars = variables!();
        let x = vars.add(variable().clamp(0, 2));
        let y = vars.add(variable().clamp(1, 3));
        let solution = vars
            .maximise(x + y)
            .using(minilp)
            .with(2 * x + y << 4)
            .solve()
            .unwrap();
        assert_eq!((solution.value(x), solution.value(y)), (0.5, 3.))
    }
}