Optex.Solution (Optex v0.1.0)

The result of a solve: a decoded status, the objective value, primal variable values, and, for LPs, dual information.

values is keyed by variable id from the solver backend; Optex.optimize/2 rekeys it (and reduced_costs) by the user-facing variable names, and rekeys duals by constraint name (name: option on constraint), falling back to the constraint id in declaration order for unnamed rows. duals and reduced_costs are nil when the solver produced no dual solution, which is always the case for models with integer or binary variables. duals covers linear rows only: quadratic constraint duals live in qcon_duals, keyed by qconstraint name (id in the qconstraint id space as the fallback), and are populated only when the solve was run with qcp_duals: true on a backend that supports it (Gurobi); otherwise qcon_duals is nil (always nil for MIQCP, where no dual solution exists). Auxiliary variables introduced by defined variables (named {name, :arg}) appear in values like any other variable.

Summary

Types

stats()

Solve statistics: wall-clock :solve_time in seconds, :simplex_iterations, branch-and-bound :nodes, and the relative :mip_gap actually achieved (nil for pure LPs, where the concept does not apply).

status()

t()

Types

stats()

@type stats() :: %{
  solve_time: float(),
  simplex_iterations: non_neg_integer(),
  nodes: non_neg_integer(),
  mip_gap: float() | nil
}

status()

@type status() ::
  :optimal
  | :infeasible
  | :unbounded
  | :unbounded_or_infeasible
  | :time_limit
  | :interrupted
  | {:other, integer()}

t()

@type t() :: %Optex.Solution{
  duals: %{required(term()) => float()} | nil,
  objective: float() | nil,
  qcon_duals: %{required(term()) => float()} | nil,
  reduced_costs: %{required(term()) => float()} | nil,
  stats: stats(),
  status: status(),
  values: %{required(term()) => float()}
}