Asignación como un problema de flujo de costo mínimo

Puedes usar la resolución de flujo de costo mínimo para resolver casos especiales de problema de asignación.

De hecho, el flujo de costo mínimo a menudo puede devolver una solución más rápido que el MIP Resolución de problemas de CP-SAT. Sin embargo, MIP y CP-SAT pueden resolver una clase de problemas mayor que la de costo mínimo, así que, en la mayoría de los casos, MIP o CP-SAT son las mejores opciones.

En las siguientes secciones, se presentan los programas de Python que resuelven las siguientes situaciones: problemas de asignación con la resolución de flujo de costo mínimo:

• Un ejemplo de asignación lineal mínimo.
• Un problema de asignación con equipos de trabajadores.

Ejemplo de asignación lineal

En esta sección, se muestra cómo resolver el ejemplo descrito en la sección Herramienta de resolución de asignaciones lineales, como un minuto de flujo de costos.

Importa las bibliotecas

Con el siguiente código, se importa la biblioteca requerida.

Python

from ortools.graph.python import min_cost_flow

C++

#include <cstdint>
#include <vector>

#include "ortools/graph/min_cost_flow.h"

Java

import com.google.ortools.Loader;
import com.google.ortools.graph.MinCostFlow;
import com.google.ortools.graph.MinCostFlowBase;

C#

using System;
using Google.OrTools.Graph;

Cómo declarar la herramienta de resolución

Con el siguiente código, se crea el solucionador de flujo de costos mínimos.

Python

# Instantiate a SimpleMinCostFlow solver.
smcf = min_cost_flow.SimpleMinCostFlow()

C++

// Instantiate a SimpleMinCostFlow solver.
SimpleMinCostFlow min_cost_flow;

Java

// Instantiate a SimpleMinCostFlow solver.
MinCostFlow minCostFlow = new MinCostFlow();

C#

// Instantiate a SimpleMinCostFlow solver.
MinCostFlow minCostFlow = new MinCostFlow();

Crea los datos

El diagrama de flujo del problema consta del grafo de bipartito para el costo matriz (consulta la resumen de las asignaciones para un porcentaje ejemplo diferente), con una fuente y un receptor agregados.

Nota: La numeración de los trabajadores y las tareas es un poco diferente a la de la sección Herramienta de resolución de asignaciones lineales porque el solucionador de flujo de costos mínimo requiere que todos los nodos del grafo estén numerados de forma distinta.

Los datos contienen los siguientes cuatro arrays, que corresponden a los nodos de inicio: los nodos finales, las capacidades y los costos del problema. La longitud de cada array es la cantidad de arcos del gráfico.

Python

# Define the directed graph for the flow.
start_nodes = (
    [0, 0, 0, 0] + [1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4] + [5, 6, 7, 8]
)
end_nodes = (
    [1, 2, 3, 4] + [5, 6, 7, 8, 5, 6, 7, 8, 5, 6, 7, 8, 5, 6, 7, 8] + [9, 9, 9, 9]
)
capacities = (
    [1, 1, 1, 1] + [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1] + [1, 1, 1, 1]
)
costs = (
    [0, 0, 0, 0]
    + [90, 76, 75, 70, 35, 85, 55, 65, 125, 95, 90, 105, 45, 110, 95, 115]
    + [0, 0, 0, 0]
)

source = 0
sink = 9
tasks = 4
supplies = [tasks, 0, 0, 0, 0, 0, 0, 0, 0, -tasks]

C++

// Define four parallel arrays: sources, destinations, capacities,
// and unit costs between each pair. For instance, the arc from node 0
// to node 1 has a capacity of 15.
const std::vector<int64_t> start_nodes = {0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2,
                                          3, 3, 3, 3, 4, 4, 4, 4, 5, 6, 7, 8};
const std::vector<int64_t> end_nodes = {1, 2, 3, 4, 5, 6, 7, 8, 5, 6, 7, 8,
                                        5, 6, 7, 8, 5, 6, 7, 8, 9, 9, 9, 9};
const std::vector<int64_t> capacities = {1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
                                         1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1};
const std::vector<int64_t> unit_costs = {0,  0,   0,  0,   90,  76, 75, 70,
                                         35, 85,  55, 65,  125, 95, 90, 105,
                                         45, 110, 95, 115, 0,   0,  0,  0};

const int64_t source = 0;
const int64_t sink = 9;
const int64_t tasks = 4;
// Define an array of supplies at each node.
const std::vector<int64_t> supplies = {tasks, 0, 0, 0, 0, 0, 0, 0, 0, -tasks};

Java

// Define four parallel arrays: sources, destinations, capacities, and unit costs
// between each pair.
int[] startNodes =
    new int[] {0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 5, 6, 7, 8};
int[] endNodes =
    new int[] {1, 2, 3, 4, 5, 6, 7, 8, 5, 6, 7, 8, 5, 6, 7, 8, 5, 6, 7, 8, 9, 9, 9, 9};
int[] capacities =
    new int[] {1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1};
int[] unitCosts = new int[] {
    0, 0, 0, 0, 90, 76, 75, 70, 35, 85, 55, 65, 125, 95, 90, 105, 45, 110, 95, 115, 0, 0, 0, 0};

int source = 0;
int sink = 9;
int tasks = 4;
// Define an array of supplies at each node.
int[] supplies = new int[] {tasks, 0, 0, 0, 0, 0, 0, 0, 0, -tasks};

C#

// Define four parallel arrays: sources, destinations, capacities, and unit costs
// between each pair.
int[] startNodes = { 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 5, 6, 7, 8 };
int[] endNodes = { 1, 2, 3, 4, 5, 6, 7, 8, 5, 6, 7, 8, 5, 6, 7, 8, 5, 6, 7, 8, 9, 9, 9, 9 };
int[] capacities = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 };
int[] unitCosts = { 0,   0,  0,  0,   90, 76,  75, 70,  35, 85, 55, 65,
                    125, 95, 90, 105, 45, 110, 95, 115, 0,  0,  0,  0 };

int source = 0;
int sink = 9;
int tasks = 4;
// Define an array of supplies at each node.
int[] supplies = { tasks, 0, 0, 0, 0, 0, 0, 0, 0, -tasks };

Para aclarar cómo se configuran los datos, cada array se divide en tres subarrays:

• El primer array corresponde a arcos que salen de la fuente.
• El segundo array corresponde a los arcos entre los trabajadores y las tareas. Para costs, esta es solo la matriz de costos (usado por el solucionador de asignaciones lineales), acoplado en un vector.
• El tercer array corresponde a los arcos que conducen al receptor.

Los datos también incluyen el vector supplies, que proporciona el valor en cada el nodo de inicio de sesión.

Cómo un problema de flujo de costos mínimo representa un problema de asignación

¿Cómo representa el problema de flujo de costo mínimo anterior un problema de asignación? Primero, como la capacidad de cada arco es 1, el suministro de 4 en la fuente fuerza a cada de los cuatro arcos que conducen a los trabajadores para que tengan un flujo de 1.

Luego, la condición flow-in-equals-flow-out obliga al flujo de cada trabajador a salir. para que sea 1. Si es posible, el solucionador dirigirá ese flujo a través del costo mínimo de código saliente de cada trabajador. Sin embargo, el solucionador no puede dirigir los flujos desde dos trabajadores diferentes a una sola tarea. Si así fuera, habría una combinación de 2 en esa tarea, que no se podía enviar a través del arco único con capacidad 1 de la tarea al receptor. Esto significa que el solucionador solo puede asignar una tarea a un trabajador, que requiere el problema de la asignación.

Por último, la condición flujo-equivalente al flujo de salida obliga a cada tarea a tener una saliente de 1, por lo que un trabajador realiza cada tarea.

Crea el gráfico y las restricciones

El siguiente código crea el gráfico y las restricciones.

Python

# Add each arc.
for i in range(len(start_nodes)):
    smcf.add_arc_with_capacity_and_unit_cost(
        start_nodes[i], end_nodes[i], capacities[i], costs[i]
    )
# Add node supplies.
for i in range(len(supplies)):
    smcf.set_node_supply(i, supplies[i])

C++

// Add each arc.
for (int i = 0; i < start_nodes.size(); ++i) {
  int arc = min_cost_flow.AddArcWithCapacityAndUnitCost(
      start_nodes[i], end_nodes[i], capacities[i], unit_costs[i]);
  if (arc != i) LOG(FATAL) << "Internal error";
}

// Add node supplies.
for (int i = 0; i < supplies.size(); ++i) {
  min_cost_flow.SetNodeSupply(i, supplies[i]);
}

Java

// Add each arc.
for (int i = 0; i < startNodes.length; ++i) {
  int arc = minCostFlow.addArcWithCapacityAndUnitCost(
      startNodes[i], endNodes[i], capacities[i], unitCosts[i]);
  if (arc != i) {
    throw new Exception("Internal error");
  }
}

// Add node supplies.
for (int i = 0; i < supplies.length; ++i) {
  minCostFlow.setNodeSupply(i, supplies[i]);
}

C#

// Add each arc.
for (int i = 0; i < startNodes.Length; ++i)
{
    int arc =
        minCostFlow.AddArcWithCapacityAndUnitCost(startNodes[i], endNodes[i], capacities[i], unitCosts[i]);
    if (arc != i)
        throw new Exception("Internal error");
}

// Add node supplies.
for (int i = 0; i < supplies.Length; ++i)
{
    minCostFlow.SetNodeSupply(i, supplies[i]);
}

Invocar el solucionador

El siguiente código invoca al solucionador y muestra la solución.

Python

# Find the minimum cost flow between node 0 and node 10.
status = smcf.solve()

C++

// Find the min cost flow.
int status = min_cost_flow.Solve();

Java

// Find the min cost flow.
MinCostFlowBase.Status status = minCostFlow.solve();

C#

// Find the min cost flow.
MinCostFlow.Status status = minCostFlow.Solve();

La solución consta de los arcos entre los trabajadores y las tareas a las que se les asigna del solucionador de problemas. (Los arcos conectados a la fuente o al receptor no forman parte del la solución).

El programa verifica cada arco para ver si tiene el flujo 1 y, de ser así, imprime los Tail (nodo de inicio) y Head (nodo final) del arco, que corresponden a un trabajador y tarea en la tarea.

Resultado del programa

Python

if status == smcf.OPTIMAL:
    print("Total cost = ", smcf.optimal_cost())
    print()
    for arc in range(smcf.num_arcs()):
        # Can ignore arcs leading out of source or into sink.
        if smcf.tail(arc) != source and smcf.head(arc) != sink:

            # Arcs in the solution have a flow value of 1. Their start and end nodes
            # give an assignment of worker to task.
            if smcf.flow(arc) > 0:
                print(
                    "Worker %d assigned to task %d.  Cost = %d"
                    % (smcf.tail(arc), smcf.head(arc), smcf.unit_cost(arc))
                )
else:
    print("There was an issue with the min cost flow input.")
    print(f"Status: {status}")

C++

if (status == MinCostFlow::OPTIMAL) {
  LOG(INFO) << "Total cost: " << min_cost_flow.OptimalCost();
  LOG(INFO) << "";
  for (std::size_t i = 0; i < min_cost_flow.NumArcs(); ++i) {
    // Can ignore arcs leading out of source or into sink.
    if (min_cost_flow.Tail(i) != source && min_cost_flow.Head(i) != sink) {
      // Arcs in the solution have a flow value of 1. Their start and end
      // nodes give an assignment of worker to task.
      if (min_cost_flow.Flow(i) > 0) {
        LOG(INFO) << "Worker " << min_cost_flow.Tail(i)
                  << " assigned to task " << min_cost_flow.Head(i)
                  << " Cost: " << min_cost_flow.UnitCost(i);
      }
    }
  }
} else {
  LOG(INFO) << "Solving the min cost flow problem failed.";
  LOG(INFO) << "Solver status: " << status;
}

Java

if (status == MinCostFlow.Status.OPTIMAL) {
  System.out.println("Total cost: " + minCostFlow.getOptimalCost());
  System.out.println();
  for (int i = 0; i < minCostFlow.getNumArcs(); ++i) {
    // Can ignore arcs leading out of source or into sink.
    if (minCostFlow.getTail(i) != source && minCostFlow.getHead(i) != sink) {
      // Arcs in the solution have a flow value of 1. Their start and end nodes
      // give an assignment of worker to task.
      if (minCostFlow.getFlow(i) > 0) {
        System.out.println("Worker " + minCostFlow.getTail(i) + " assigned to task "
            + minCostFlow.getHead(i) + " Cost: " + minCostFlow.getUnitCost(i));
      }
    }
  }
} else {
  System.out.println("Solving the min cost flow problem failed.");
  System.out.println("Solver status: " + status);
}

C#

if (status == MinCostFlow.Status.OPTIMAL)
{
    Console.WriteLine("Total cost: " + minCostFlow.OptimalCost());
    Console.WriteLine("");
    for (int i = 0; i < minCostFlow.NumArcs(); ++i)
    {
        // Can ignore arcs leading out of source or into sink.
        if (minCostFlow.Tail(i) != source && minCostFlow.Head(i) != sink)
        {
            // Arcs in the solution have a flow value of 1. Their start and end nodes
            // give an assignment of worker to task.
            if (minCostFlow.Flow(i) > 0)
            {
                Console.WriteLine("Worker " + minCostFlow.Tail(i) + " assigned to task " + minCostFlow.Head(i) +
                                  " Cost: " + minCostFlow.UnitCost(i));
            }
        }
    }
}
else
{
    Console.WriteLine("Solving the min cost flow problem failed.");
    Console.WriteLine("Solver status: " + status);
}

Este es el resultado del programa.

Total cost = 265

Worker 1 assigned to task 8.  Cost = 70
Worker 2 assigned to task 7.  Cost = 55
Worker 3 assigned to task 6.  Cost = 95
Worker 4 assigned to task 5.  Cost = 45

Time = 0.000245 seconds

El resultado es el mismo que para la de resolución de asignaciones lineales (excepto por las diferentes la numeración de los trabajadores y los costos). La resolución de asignaciones lineales es un poco más rápida que el flujo de costo mínimo: 0.000147 segundos en comparación con 0.000458 segundos.

Todo el programa

El programa completo se muestra a continuación.

Python

"""Linear assignment example."""
from ortools.graph.python import min_cost_flow


def main():
    """Solving an Assignment Problem with MinCostFlow."""
    # Instantiate a SimpleMinCostFlow solver.
    smcf = min_cost_flow.SimpleMinCostFlow()

    # Define the directed graph for the flow.
    start_nodes = (
        [0, 0, 0, 0] + [1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4] + [5, 6, 7, 8]
    )
    end_nodes = (
        [1, 2, 3, 4] + [5, 6, 7, 8, 5, 6, 7, 8, 5, 6, 7, 8, 5, 6, 7, 8] + [9, 9, 9, 9]
    )
    capacities = (
        [1, 1, 1, 1] + [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1] + [1, 1, 1, 1]
    )
    costs = (
        [0, 0, 0, 0]
        + [90, 76, 75, 70, 35, 85, 55, 65, 125, 95, 90, 105, 45, 110, 95, 115]
        + [0, 0, 0, 0]
    )

    source = 0
    sink = 9
    tasks = 4
    supplies = [tasks, 0, 0, 0, 0, 0, 0, 0, 0, -tasks]

    # Add each arc.
    for i in range(len(start_nodes)):
        smcf.add_arc_with_capacity_and_unit_cost(
            start_nodes[i], end_nodes[i], capacities[i], costs[i]
        )
    # Add node supplies.
    for i in range(len(supplies)):
        smcf.set_node_supply(i, supplies[i])

    # Find the minimum cost flow between node 0 and node 10.
    status = smcf.solve()

    if status == smcf.OPTIMAL:
        print("Total cost = ", smcf.optimal_cost())
        print()
        for arc in range(smcf.num_arcs()):
            # Can ignore arcs leading out of source or into sink.
            if smcf.tail(arc) != source and smcf.head(arc) != sink:

                # Arcs in the solution have a flow value of 1. Their start and end nodes
                # give an assignment of worker to task.
                if smcf.flow(arc) > 0:
                    print(
                        "Worker %d assigned to task %d.  Cost = %d"
                        % (smcf.tail(arc), smcf.head(arc), smcf.unit_cost(arc))
                    )
    else:
        print("There was an issue with the min cost flow input.")
        print(f"Status: {status}")


if __name__ == "__main__":
    main()

C++

#include <cstdint>
#include <vector>

#include "ortools/graph/min_cost_flow.h"

namespace operations_research {
// MinCostFlow simple interface example.
void AssignmentMinFlow() {
  // Instantiate a SimpleMinCostFlow solver.
  SimpleMinCostFlow min_cost_flow;

  // Define four parallel arrays: sources, destinations, capacities,
  // and unit costs between each pair. For instance, the arc from node 0
  // to node 1 has a capacity of 15.
  const std::vector<int64_t> start_nodes = {0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2,
                                            3, 3, 3, 3, 4, 4, 4, 4, 5, 6, 7, 8};
  const std::vector<int64_t> end_nodes = {1, 2, 3, 4, 5, 6, 7, 8, 5, 6, 7, 8,
                                          5, 6, 7, 8, 5, 6, 7, 8, 9, 9, 9, 9};
  const std::vector<int64_t> capacities = {1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
                                           1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1};
  const std::vector<int64_t> unit_costs = {0,  0,   0,  0,   90,  76, 75, 70,
                                           35, 85,  55, 65,  125, 95, 90, 105,
                                           45, 110, 95, 115, 0,   0,  0,  0};

  const int64_t source = 0;
  const int64_t sink = 9;
  const int64_t tasks = 4;
  // Define an array of supplies at each node.
  const std::vector<int64_t> supplies = {tasks, 0, 0, 0, 0, 0, 0, 0, 0, -tasks};

  // Add each arc.
  for (int i = 0; i < start_nodes.size(); ++i) {
    int arc = min_cost_flow.AddArcWithCapacityAndUnitCost(
        start_nodes[i], end_nodes[i], capacities[i], unit_costs[i]);
    if (arc != i) LOG(FATAL) << "Internal error";
  }

  // Add node supplies.
  for (int i = 0; i < supplies.size(); ++i) {
    min_cost_flow.SetNodeSupply(i, supplies[i]);
  }

  // Find the min cost flow.
  int status = min_cost_flow.Solve();

  if (status == MinCostFlow::OPTIMAL) {
    LOG(INFO) << "Total cost: " << min_cost_flow.OptimalCost();
    LOG(INFO) << "";
    for (std::size_t i = 0; i < min_cost_flow.NumArcs(); ++i) {
      // Can ignore arcs leading out of source or into sink.
      if (min_cost_flow.Tail(i) != source && min_cost_flow.Head(i) != sink) {
        // Arcs in the solution have a flow value of 1. Their start and end
        // nodes give an assignment of worker to task.
        if (min_cost_flow.Flow(i) > 0) {
          LOG(INFO) << "Worker " << min_cost_flow.Tail(i)
                    << " assigned to task " << min_cost_flow.Head(i)
                    << " Cost: " << min_cost_flow.UnitCost(i);
        }
      }
    }
  } else {
    LOG(INFO) << "Solving the min cost flow problem failed.";
    LOG(INFO) << "Solver status: " << status;
  }
}

}  // namespace operations_research

int main() {
  operations_research::AssignmentMinFlow();
  return EXIT_SUCCESS;
}

Java

package com.google.ortools.graph.samples;
import com.google.ortools.Loader;
import com.google.ortools.graph.MinCostFlow;
import com.google.ortools.graph.MinCostFlowBase;

/** Minimal Assignment Min Flow. */
public class AssignmentMinFlow {
  public static void main(String[] args) throws Exception {
    Loader.loadNativeLibraries();
    // Instantiate a SimpleMinCostFlow solver.
    MinCostFlow minCostFlow = new MinCostFlow();

    // Define four parallel arrays: sources, destinations, capacities, and unit costs
    // between each pair.
    int[] startNodes =
        new int[] {0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 5, 6, 7, 8};
    int[] endNodes =
        new int[] {1, 2, 3, 4, 5, 6, 7, 8, 5, 6, 7, 8, 5, 6, 7, 8, 5, 6, 7, 8, 9, 9, 9, 9};
    int[] capacities =
        new int[] {1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1};
    int[] unitCosts = new int[] {
        0, 0, 0, 0, 90, 76, 75, 70, 35, 85, 55, 65, 125, 95, 90, 105, 45, 110, 95, 115, 0, 0, 0, 0};

    int source = 0;
    int sink = 9;
    int tasks = 4;
    // Define an array of supplies at each node.
    int[] supplies = new int[] {tasks, 0, 0, 0, 0, 0, 0, 0, 0, -tasks};

    // Add each arc.
    for (int i = 0; i < startNodes.length; ++i) {
      int arc = minCostFlow.addArcWithCapacityAndUnitCost(
          startNodes[i], endNodes[i], capacities[i], unitCosts[i]);
      if (arc != i) {
        throw new Exception("Internal error");
      }
    }

    // Add node supplies.
    for (int i = 0; i < supplies.length; ++i) {
      minCostFlow.setNodeSupply(i, supplies[i]);
    }

    // Find the min cost flow.
    MinCostFlowBase.Status status = minCostFlow.solve();

    if (status == MinCostFlow.Status.OPTIMAL) {
      System.out.println("Total cost: " + minCostFlow.getOptimalCost());
      System.out.println();
      for (int i = 0; i < minCostFlow.getNumArcs(); ++i) {
        // Can ignore arcs leading out of source or into sink.
        if (minCostFlow.getTail(i) != source && minCostFlow.getHead(i) != sink) {
          // Arcs in the solution have a flow value of 1. Their start and end nodes
          // give an assignment of worker to task.
          if (minCostFlow.getFlow(i) > 0) {
            System.out.println("Worker " + minCostFlow.getTail(i) + " assigned to task "
                + minCostFlow.getHead(i) + " Cost: " + minCostFlow.getUnitCost(i));
          }
        }
      }
    } else {
      System.out.println("Solving the min cost flow problem failed.");
      System.out.println("Solver status: " + status);
    }
  }

  private AssignmentMinFlow() {}
}

C#

using System;
using Google.OrTools.Graph;

public class AssignmentMinFlow
{
    static void Main()
    {
        // Instantiate a SimpleMinCostFlow solver.
        MinCostFlow minCostFlow = new MinCostFlow();

        // Define four parallel arrays: sources, destinations, capacities, and unit costs
        // between each pair.
        int[] startNodes = { 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 5, 6, 7, 8 };
        int[] endNodes = { 1, 2, 3, 4, 5, 6, 7, 8, 5, 6, 7, 8, 5, 6, 7, 8, 5, 6, 7, 8, 9, 9, 9, 9 };
        int[] capacities = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 };
        int[] unitCosts = { 0,   0,  0,  0,   90, 76,  75, 70,  35, 85, 55, 65,
                            125, 95, 90, 105, 45, 110, 95, 115, 0,  0,  0,  0 };

        int source = 0;
        int sink = 9;
        int tasks = 4;
        // Define an array of supplies at each node.
        int[] supplies = { tasks, 0, 0, 0, 0, 0, 0, 0, 0, -tasks };

        // Add each arc.
        for (int i = 0; i < startNodes.Length; ++i)
        {
            int arc =
                minCostFlow.AddArcWithCapacityAndUnitCost(startNodes[i], endNodes[i], capacities[i], unitCosts[i]);
            if (arc != i)
                throw new Exception("Internal error");
        }

        // Add node supplies.
        for (int i = 0; i < supplies.Length; ++i)
        {
            minCostFlow.SetNodeSupply(i, supplies[i]);
        }

        // Find the min cost flow.
        MinCostFlow.Status status = minCostFlow.Solve();

        if (status == MinCostFlow.Status.OPTIMAL)
        {
            Console.WriteLine("Total cost: " + minCostFlow.OptimalCost());
            Console.WriteLine("");
            for (int i = 0; i < minCostFlow.NumArcs(); ++i)
            {
                // Can ignore arcs leading out of source or into sink.
                if (minCostFlow.Tail(i) != source && minCostFlow.Head(i) != sink)
                {
                    // Arcs in the solution have a flow value of 1. Their start and end nodes
                    // give an assignment of worker to task.
                    if (minCostFlow.Flow(i) > 0)
                    {
                        Console.WriteLine("Worker " + minCostFlow.Tail(i) + " assigned to task " + minCostFlow.Head(i) +
                                          " Cost: " + minCostFlow.UnitCost(i));
                    }
                }
            }
        }
        else
        {
            Console.WriteLine("Solving the min cost flow problem failed.");
            Console.WriteLine("Solver status: " + status);
        }
    }
}

Asignación con equipos de trabajadores

En esta sección, se presenta un problema de asignación más general. En este problema, seis y los trabajadores se dividen en dos equipos. El problema es asignar cuatro tareas al trabajadores para que la carga de trabajo esté equilibrada entre los equipos es decir, cada equipo realiza dos de las tareas.

Para ver una solución de resolución de MIP para este problema, consulta Asignación con equipos de trabajadores.

En las siguientes secciones, se describe un programa que resuelve el problema mediante la función de de flujo de costos.

Importa las bibliotecas

Con el siguiente código, se importa la biblioteca requerida.

Python

from ortools.graph.python import min_cost_flow

C++

#include <cstdint>
#include <vector>

#include "ortools/graph/min_cost_flow.h"

Java

import com.google.ortools.Loader;
import com.google.ortools.graph.MinCostFlow;
import com.google.ortools.graph.MinCostFlowBase;

C#

using System;
using Google.OrTools.Graph;

Cómo declarar la herramienta de resolución

Con el siguiente código, se crea el solucionador de flujo de costos mínimos.

Python

smcf = min_cost_flow.SimpleMinCostFlow()

C++

// Instantiate a SimpleMinCostFlow solver.
SimpleMinCostFlow min_cost_flow;

Java

// Instantiate a SimpleMinCostFlow solver.
MinCostFlow minCostFlow = new MinCostFlow();

C#

// Instantiate a SimpleMinCostFlow solver.
MinCostFlow minCostFlow = new MinCostFlow();

Crea los datos

El siguiente código crea los datos para el programa.

Python

# Define the directed graph for the flow.
team_a = [1, 3, 5]
team_b = [2, 4, 6]

start_nodes = (
    # fmt: off
  [0, 0]
  + [11, 11, 11]
  + [12, 12, 12]
  + [1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 6, 6, 6, 6]
  + [7, 8, 9, 10]
    # fmt: on
)
end_nodes = (
    # fmt: off
  [11, 12]
  + team_a
  + team_b
  + [7, 8, 9, 10, 7, 8, 9, 10, 7, 8, 9, 10, 7, 8, 9, 10, 7, 8, 9, 10, 7, 8, 9, 10]
  + [13, 13, 13, 13]
    # fmt: on
)
capacities = (
    # fmt: off
  [2, 2]
  + [1, 1, 1]
  + [1, 1, 1]
  + [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
  + [1, 1, 1, 1]
    # fmt: on
)
costs = (
    # fmt: off
  [0, 0]
  + [0, 0, 0]
  + [0, 0, 0]
  + [90, 76, 75, 70, 35, 85, 55, 65, 125, 95, 90, 105, 45, 110, 95, 115, 60, 105, 80, 75, 45, 65, 110, 95]
  + [0, 0, 0, 0]
    # fmt: on
)

source = 0
sink = 13
tasks = 4
# Define an array of supplies at each node.
supplies = [tasks, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, -tasks]

C++

// Define the directed graph for the flow.
const std::vector<int64_t> team_A = {1, 3, 5};
const std::vector<int64_t> team_B = {2, 4, 6};

const std::vector<int64_t> start_nodes = {
    0, 0, 11, 11, 11, 12, 12, 12, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3,
    3, 3, 4,  4,  4,  4,  5,  5,  5, 5, 6, 6, 6, 6, 7, 8, 9, 10};
const std::vector<int64_t> end_nodes = {
    11, 12, 1, 3, 5, 2,  4, 6, 7, 8,  9, 10, 7, 8,  9,  10, 7,  8,
    9,  10, 7, 8, 9, 10, 7, 8, 9, 10, 7, 8,  9, 10, 13, 13, 13, 13};
const std::vector<int64_t> capacities = {2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
                                         1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
                                         1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1};
const std::vector<int64_t> unit_costs = {
    0,  0,   0,  0,  0,   0,  0,   0,   90, 76,  75, 70,
    35, 85,  55, 65, 125, 95, 90,  105, 45, 110, 95, 115,
    60, 105, 80, 75, 45,  65, 110, 95,  0,  0,   0,  0};

const int64_t source = 0;
const int64_t sink = 13;
const int64_t tasks = 4;
// Define an array of supplies at each node.
const std::vector<int64_t> supplies = {tasks, 0, 0, 0, 0, 0, 0,
                                       0,     0, 0, 0, 0, 0, -tasks};

Java

// Define the directed graph for the flow.
// int[] teamA = new int[] {1, 3, 5};
// int[] teamB = new int[] {2, 4, 6};

int[] startNodes = new int[] {0, 0, 11, 11, 11, 12, 12, 12, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3,
    4, 4, 4, 4, 5, 5, 5, 5, 6, 6, 6, 6, 7, 8, 9, 10};
int[] endNodes = new int[] {11, 12, 1, 3, 5, 2, 4, 6, 7, 8, 9, 10, 7, 8, 9, 10, 7, 8, 9, 10, 7,
    8, 9, 10, 7, 8, 9, 10, 7, 8, 9, 10, 13, 13, 13, 13};
int[] capacities = new int[] {2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1};
int[] unitCosts = new int[] {0, 0, 0, 0, 0, 0, 0, 0, 90, 76, 75, 70, 35, 85, 55, 65, 125, 95,
    90, 105, 45, 110, 95, 115, 60, 105, 80, 75, 45, 65, 110, 95, 0, 0, 0, 0};

int source = 0;
int sink = 13;
int tasks = 4;
// Define an array of supplies at each node.
int[] supplies = new int[] {tasks, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, -tasks};

C#

// Define the directed graph for the flow.
int[] teamA = { 1, 3, 5 };
int[] teamB = { 2, 4, 6 };

// Define four parallel arrays: sources, destinations, capacities, and unit costs
// between each pair.
int[] startNodes = { 0, 0, 11, 11, 11, 12, 12, 12, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3,
                     3, 3, 4,  4,  4,  4,  5,  5,  5, 5, 6, 6, 6, 6, 7, 8, 9, 10 };
int[] endNodes = { 11, 12, 1, 3, 5, 2,  4, 6, 7, 8,  9, 10, 7, 8,  9,  10, 7,  8,
                   9,  10, 7, 8, 9, 10, 7, 8, 9, 10, 7, 8,  9, 10, 13, 13, 13, 13 };
int[] capacities = { 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 };
int[] unitCosts = { 0,  0,   0,  0,   0,  0,   0,  0,   90, 76, 75, 70, 35,  85, 55, 65, 125, 95,
                    90, 105, 45, 110, 95, 115, 60, 105, 80, 75, 45, 65, 110, 95, 0,  0,  0,   0 };

int source = 0;
int sink = 13;
int tasks = 4;
// Define an array of supplies at each node.
int[] supplies = { tasks, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, -tasks };

Los trabajadores corresponden a los nodos 1 a 6. El Equipo A consta de los trabajadores 1, 3 y 5, y el equipo B consta de los trabajadores 2, 4 y 6. Las tareas están numeradas del 7 al 10.

Hay dos nodos nuevos, 11 y 12, entre la fuente y los trabajadores. El Nodo 11 es se conecta a los nodos del equipo A, y el Nodo 12 se conecta a los nodos del equipo equipo B, con arcos de capacidad 1. En el siguiente gráfico, se muestran solo los nodos y arcos desde la fuente hasta los trabajadores.

La clave para equilibrar la carga de trabajo es que la fuente 0 esté conectada a los nodos 11. y 12 por arcos de capacidad 2. Esto significa que los nodos 11 y 12 (y, por lo tanto, los equipos A y B) pueden tener un flujo máximo de 2. Como resultado, cada equipo puede realizar como máximo dos de las tareas.

Crea las restricciones

Python

# Add each arc.
for i in range(0, len(start_nodes)):
    smcf.add_arc_with_capacity_and_unit_cost(
        start_nodes[i], end_nodes[i], capacities[i], costs[i]
    )

# Add node supplies.
for i in range(0, len(supplies)):
    smcf.set_node_supply(i, supplies[i])

C++

// Add each arc.
for (int i = 0; i < start_nodes.size(); ++i) {
  int arc = min_cost_flow.AddArcWithCapacityAndUnitCost(
      start_nodes[i], end_nodes[i], capacities[i], unit_costs[i]);
  if (arc != i) LOG(FATAL) << "Internal error";
}

// Add node supplies.
for (int i = 0; i < supplies.size(); ++i) {
  min_cost_flow.SetNodeSupply(i, supplies[i]);
}

Java

// Add each arc.
for (int i = 0; i < startNodes.length; ++i) {
  int arc = minCostFlow.addArcWithCapacityAndUnitCost(
      startNodes[i], endNodes[i], capacities[i], unitCosts[i]);
  if (arc != i) {
    throw new Exception("Internal error");
  }
}

// Add node supplies.
for (int i = 0; i < supplies.length; ++i) {
  minCostFlow.setNodeSupply(i, supplies[i]);
}

C#

// Add each arc.
for (int i = 0; i < startNodes.Length; ++i)
{
    int arc =
        minCostFlow.AddArcWithCapacityAndUnitCost(startNodes[i], endNodes[i], capacities[i], unitCosts[i]);
    if (arc != i)
        throw new Exception("Internal error");
}

// Add node supplies.
for (int i = 0; i < supplies.Length; ++i)
{
    minCostFlow.SetNodeSupply(i, supplies[i]);
}

Invocar el solucionador

Python

# Find the minimum cost flow between node 0 and node 10.
status = smcf.solve()

C++

// Find the min cost flow.
int status = min_cost_flow.Solve();

Java

// Find the min cost flow.
MinCostFlowBase.Status status = minCostFlow.solve();

C#

// Find the min cost flow.
MinCostFlow.Status status = minCostFlow.Solve();

Resultado del programa

Python

if status == smcf.OPTIMAL:
    print("Total cost = ", smcf.optimal_cost())
    print()
    for arc in range(smcf.num_arcs()):
        # Can ignore arcs leading out of source or intermediate, or into sink.
        if (
            smcf.tail(arc) != source
            and smcf.tail(arc) != 11
            and smcf.tail(arc) != 12
            and smcf.head(arc) != sink
        ):

            # Arcs in the solution will have a flow value of 1.
            # There start and end nodes give an assignment of worker to task.
            if smcf.flow(arc) > 0:
                print(
                    "Worker %d assigned to task %d.  Cost = %d"
                    % (smcf.tail(arc), smcf.head(arc), smcf.unit_cost(arc))
                )
else:
    print("There was an issue with the min cost flow input.")
    print(f"Status: {status}")

C++

if (status == MinCostFlow::OPTIMAL) {
  LOG(INFO) << "Total cost: " << min_cost_flow.OptimalCost();
  LOG(INFO) << "";
  for (std::size_t i = 0; i < min_cost_flow.NumArcs(); ++i) {
    // Can ignore arcs leading out of source or intermediate nodes, or into
    // sink.
    if (min_cost_flow.Tail(i) != source && min_cost_flow.Tail(i) != 11 &&
        min_cost_flow.Tail(i) != 12 && min_cost_flow.Head(i) != sink) {
      // Arcs in the solution have a flow value of 1. Their start and end
      // nodes give an assignment of worker to task.
      if (min_cost_flow.Flow(i) > 0) {
        LOG(INFO) << "Worker " << min_cost_flow.Tail(i)
                  << " assigned to task " << min_cost_flow.Head(i)
                  << " Cost: " << min_cost_flow.UnitCost(i);
      }
    }
  }
} else {
  LOG(INFO) << "Solving the min cost flow problem failed.";
  LOG(INFO) << "Solver status: " << status;
}

Java

if (status == MinCostFlow.Status.OPTIMAL) {
  System.out.println("Total cost: " + minCostFlow.getOptimalCost());
  System.out.println();
  for (int i = 0; i < minCostFlow.getNumArcs(); ++i) {
    // Can ignore arcs leading out of source or intermediate nodes, or into sink.
    if (minCostFlow.getTail(i) != source && minCostFlow.getTail(i) != 11
        && minCostFlow.getTail(i) != 12 && minCostFlow.getHead(i) != sink) {
      // Arcs in the solution have a flow value of 1. Their start and end nodes
      // give an assignment of worker to task.
      if (minCostFlow.getFlow(i) > 0) {
        System.out.println("Worker " + minCostFlow.getTail(i) + " assigned to task "
            + minCostFlow.getHead(i) + " Cost: " + minCostFlow.getUnitCost(i));
      }
    }
  }
} else {
  System.out.println("Solving the min cost flow problem failed.");
  System.out.println("Solver status: " + status);
}

C#

if (status == MinCostFlow.Status.OPTIMAL)
{
    Console.WriteLine("Total cost: " + minCostFlow.OptimalCost());
    Console.WriteLine("");
    for (int i = 0; i < minCostFlow.NumArcs(); ++i)
    {
        // Can ignore arcs leading out of source or into sink.
        if (minCostFlow.Tail(i) != source && minCostFlow.Tail(i) != 11 && minCostFlow.Tail(i) != 12 &&
            minCostFlow.Head(i) != sink)
        {
            // Arcs in the solution have a flow value of 1. Their start and end nodes
            // give an assignment of worker to task.
            if (minCostFlow.Flow(i) > 0)
            {
                Console.WriteLine("Worker " + minCostFlow.Tail(i) + " assigned to task " + minCostFlow.Head(i) +
                                  " Cost: " + minCostFlow.UnitCost(i));
            }
        }
    }
}
else
{
    Console.WriteLine("Solving the min cost flow problem failed.");
    Console.WriteLine("Solver status: " + status);
}

A continuación, se muestra el resultado del programa.

Total cost = 250

Worker 1 assigned to task 9.  Cost =  75
Worker 2 assigned to task 7.  Cost =  35
Worker 5 assigned to task 10.  Cost =  75
Worker 6 assigned to task 8.  Cost =  65

Time = 0.00031 seconds

Al equipo A se le asignan las tareas 9 y 10, mientras que al equipo B se le asignan las tareas 7 y 8.

Ten en cuenta que la resolución de flujo de costo mínimo es más rápida para este problema que para de resolución MIP, que tarda alrededor de 0.006 segundos.

Todo el programa

El programa completo se muestra a continuación.

Python

"""Assignment with teams of workers."""
from ortools.graph.python import min_cost_flow


def main():
    """Solving an Assignment with teams of worker."""
    smcf = min_cost_flow.SimpleMinCostFlow()

    # Define the directed graph for the flow.
    team_a = [1, 3, 5]
    team_b = [2, 4, 6]

    start_nodes = (
        # fmt: off
      [0, 0]
      + [11, 11, 11]
      + [12, 12, 12]
      + [1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 6, 6, 6, 6]
      + [7, 8, 9, 10]
        # fmt: on
    )
    end_nodes = (
        # fmt: off
      [11, 12]
      + team_a
      + team_b
      + [7, 8, 9, 10, 7, 8, 9, 10, 7, 8, 9, 10, 7, 8, 9, 10, 7, 8, 9, 10, 7, 8, 9, 10]
      + [13, 13, 13, 13]
        # fmt: on
    )
    capacities = (
        # fmt: off
      [2, 2]
      + [1, 1, 1]
      + [1, 1, 1]
      + [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
      + [1, 1, 1, 1]
        # fmt: on
    )
    costs = (
        # fmt: off
      [0, 0]
      + [0, 0, 0]
      + [0, 0, 0]
      + [90, 76, 75, 70, 35, 85, 55, 65, 125, 95, 90, 105, 45, 110, 95, 115, 60, 105, 80, 75, 45, 65, 110, 95]
      + [0, 0, 0, 0]
        # fmt: on
    )

    source = 0
    sink = 13
    tasks = 4
    # Define an array of supplies at each node.
    supplies = [tasks, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, -tasks]

    # Add each arc.
    for i in range(0, len(start_nodes)):
        smcf.add_arc_with_capacity_and_unit_cost(
            start_nodes[i], end_nodes[i], capacities[i], costs[i]
        )

    # Add node supplies.
    for i in range(0, len(supplies)):
        smcf.set_node_supply(i, supplies[i])

    # Find the minimum cost flow between node 0 and node 10.
    status = smcf.solve()

    if status == smcf.OPTIMAL:
        print("Total cost = ", smcf.optimal_cost())
        print()
        for arc in range(smcf.num_arcs()):
            # Can ignore arcs leading out of source or intermediate, or into sink.
            if (
                smcf.tail(arc) != source
                and smcf.tail(arc) != 11
                and smcf.tail(arc) != 12
                and smcf.head(arc) != sink
            ):

                # Arcs in the solution will have a flow value of 1.
                # There start and end nodes give an assignment of worker to task.
                if smcf.flow(arc) > 0:
                    print(
                        "Worker %d assigned to task %d.  Cost = %d"
                        % (smcf.tail(arc), smcf.head(arc), smcf.unit_cost(arc))
                    )
    else:
        print("There was an issue with the min cost flow input.")
        print(f"Status: {status}")


if __name__ == "__main__":
    main()

C++

#include <cstdint>
#include <vector>

#include "ortools/graph/min_cost_flow.h"

namespace operations_research {
// MinCostFlow simple interface example.
void BalanceMinFlow() {
  // Instantiate a SimpleMinCostFlow solver.
  SimpleMinCostFlow min_cost_flow;

  // Define the directed graph for the flow.
  const std::vector<int64_t> team_A = {1, 3, 5};
  const std::vector<int64_t> team_B = {2, 4, 6};

  const std::vector<int64_t> start_nodes = {
      0, 0, 11, 11, 11, 12, 12, 12, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3,
      3, 3, 4,  4,  4,  4,  5,  5,  5, 5, 6, 6, 6, 6, 7, 8, 9, 10};
  const std::vector<int64_t> end_nodes = {
      11, 12, 1, 3, 5, 2,  4, 6, 7, 8,  9, 10, 7, 8,  9,  10, 7,  8,
      9,  10, 7, 8, 9, 10, 7, 8, 9, 10, 7, 8,  9, 10, 13, 13, 13, 13};
  const std::vector<int64_t> capacities = {2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
                                           1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
                                           1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1};
  const std::vector<int64_t> unit_costs = {
      0,  0,   0,  0,  0,   0,  0,   0,   90, 76,  75, 70,
      35, 85,  55, 65, 125, 95, 90,  105, 45, 110, 95, 115,
      60, 105, 80, 75, 45,  65, 110, 95,  0,  0,   0,  0};

  const int64_t source = 0;
  const int64_t sink = 13;
  const int64_t tasks = 4;
  // Define an array of supplies at each node.
  const std::vector<int64_t> supplies = {tasks, 0, 0, 0, 0, 0, 0,
                                         0,     0, 0, 0, 0, 0, -tasks};

  // Add each arc.
  for (int i = 0; i < start_nodes.size(); ++i) {
    int arc = min_cost_flow.AddArcWithCapacityAndUnitCost(
        start_nodes[i], end_nodes[i], capacities[i], unit_costs[i]);
    if (arc != i) LOG(FATAL) << "Internal error";
  }

  // Add node supplies.
  for (int i = 0; i < supplies.size(); ++i) {
    min_cost_flow.SetNodeSupply(i, supplies[i]);
  }

  // Find the min cost flow.
  int status = min_cost_flow.Solve();

  if (status == MinCostFlow::OPTIMAL) {
    LOG(INFO) << "Total cost: " << min_cost_flow.OptimalCost();
    LOG(INFO) << "";
    for (std::size_t i = 0; i < min_cost_flow.NumArcs(); ++i) {
      // Can ignore arcs leading out of source or intermediate nodes, or into
      // sink.
      if (min_cost_flow.Tail(i) != source && min_cost_flow.Tail(i) != 11 &&
          min_cost_flow.Tail(i) != 12 && min_cost_flow.Head(i) != sink) {
        // Arcs in the solution have a flow value of 1. Their start and end
        // nodes give an assignment of worker to task.
        if (min_cost_flow.Flow(i) > 0) {
          LOG(INFO) << "Worker " << min_cost_flow.Tail(i)
                    << " assigned to task " << min_cost_flow.Head(i)
                    << " Cost: " << min_cost_flow.UnitCost(i);
        }
      }
    }
  } else {
    LOG(INFO) << "Solving the min cost flow problem failed.";
    LOG(INFO) << "Solver status: " << status;
  }
}

}  // namespace operations_research

int main() {
  operations_research::BalanceMinFlow();
  return EXIT_SUCCESS;
}

Java

package com.google.ortools.graph.samples;
import com.google.ortools.Loader;
import com.google.ortools.graph.MinCostFlow;
import com.google.ortools.graph.MinCostFlowBase;

/** Minimal Assignment Min Flow. */
public class BalanceMinFlow {
  public static void main(String[] args) throws Exception {
    Loader.loadNativeLibraries();
    // Instantiate a SimpleMinCostFlow solver.
    MinCostFlow minCostFlow = new MinCostFlow();

    // Define the directed graph for the flow.
    // int[] teamA = new int[] {1, 3, 5};
    // int[] teamB = new int[] {2, 4, 6};

    int[] startNodes = new int[] {0, 0, 11, 11, 11, 12, 12, 12, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3,
        4, 4, 4, 4, 5, 5, 5, 5, 6, 6, 6, 6, 7, 8, 9, 10};
    int[] endNodes = new int[] {11, 12, 1, 3, 5, 2, 4, 6, 7, 8, 9, 10, 7, 8, 9, 10, 7, 8, 9, 10, 7,
        8, 9, 10, 7, 8, 9, 10, 7, 8, 9, 10, 13, 13, 13, 13};
    int[] capacities = new int[] {2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1};
    int[] unitCosts = new int[] {0, 0, 0, 0, 0, 0, 0, 0, 90, 76, 75, 70, 35, 85, 55, 65, 125, 95,
        90, 105, 45, 110, 95, 115, 60, 105, 80, 75, 45, 65, 110, 95, 0, 0, 0, 0};

    int source = 0;
    int sink = 13;
    int tasks = 4;
    // Define an array of supplies at each node.
    int[] supplies = new int[] {tasks, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, -tasks};

    // Add each arc.
    for (int i = 0; i < startNodes.length; ++i) {
      int arc = minCostFlow.addArcWithCapacityAndUnitCost(
          startNodes[i], endNodes[i], capacities[i], unitCosts[i]);
      if (arc != i) {
        throw new Exception("Internal error");
      }
    }

    // Add node supplies.
    for (int i = 0; i < supplies.length; ++i) {
      minCostFlow.setNodeSupply(i, supplies[i]);
    }

    // Find the min cost flow.
    MinCostFlowBase.Status status = minCostFlow.solve();

    if (status == MinCostFlow.Status.OPTIMAL) {
      System.out.println("Total cost: " + minCostFlow.getOptimalCost());
      System.out.println();
      for (int i = 0; i < minCostFlow.getNumArcs(); ++i) {
        // Can ignore arcs leading out of source or intermediate nodes, or into sink.
        if (minCostFlow.getTail(i) != source && minCostFlow.getTail(i) != 11
            && minCostFlow.getTail(i) != 12 && minCostFlow.getHead(i) != sink) {
          // Arcs in the solution have a flow value of 1. Their start and end nodes
          // give an assignment of worker to task.
          if (minCostFlow.getFlow(i) > 0) {
            System.out.println("Worker " + minCostFlow.getTail(i) + " assigned to task "
                + minCostFlow.getHead(i) + " Cost: " + minCostFlow.getUnitCost(i));
          }
        }
      }
    } else {
      System.out.println("Solving the min cost flow problem failed.");
      System.out.println("Solver status: " + status);
    }
  }

  private BalanceMinFlow() {}
}

C#

using System;
using Google.OrTools.Graph;

public class BalanceMinFlow
{
    static void Main()
    {
        // Instantiate a SimpleMinCostFlow solver.
        MinCostFlow minCostFlow = new MinCostFlow();

        // Define the directed graph for the flow.
        int[] teamA = { 1, 3, 5 };
        int[] teamB = { 2, 4, 6 };

        // Define four parallel arrays: sources, destinations, capacities, and unit costs
        // between each pair.
        int[] startNodes = { 0, 0, 11, 11, 11, 12, 12, 12, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3,
                             3, 3, 4,  4,  4,  4,  5,  5,  5, 5, 6, 6, 6, 6, 7, 8, 9, 10 };
        int[] endNodes = { 11, 12, 1, 3, 5, 2,  4, 6, 7, 8,  9, 10, 7, 8,  9,  10, 7,  8,
                           9,  10, 7, 8, 9, 10, 7, 8, 9, 10, 7, 8,  9, 10, 13, 13, 13, 13 };
        int[] capacities = { 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
                             1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 };
        int[] unitCosts = { 0,  0,   0,  0,   0,  0,   0,  0,   90, 76, 75, 70, 35,  85, 55, 65, 125, 95,
                            90, 105, 45, 110, 95, 115, 60, 105, 80, 75, 45, 65, 110, 95, 0,  0,  0,   0 };

        int source = 0;
        int sink = 13;
        int tasks = 4;
        // Define an array of supplies at each node.
        int[] supplies = { tasks, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, -tasks };

        // Add each arc.
        for (int i = 0; i < startNodes.Length; ++i)
        {
            int arc =
                minCostFlow.AddArcWithCapacityAndUnitCost(startNodes[i], endNodes[i], capacities[i], unitCosts[i]);
            if (arc != i)
                throw new Exception("Internal error");
        }

        // Add node supplies.
        for (int i = 0; i < supplies.Length; ++i)
        {
            minCostFlow.SetNodeSupply(i, supplies[i]);
        }

        // Find the min cost flow.
        MinCostFlow.Status status = minCostFlow.Solve();

        if (status == MinCostFlow.Status.OPTIMAL)
        {
            Console.WriteLine("Total cost: " + minCostFlow.OptimalCost());
            Console.WriteLine("");
            for (int i = 0; i < minCostFlow.NumArcs(); ++i)
            {
                // Can ignore arcs leading out of source or into sink.
                if (minCostFlow.Tail(i) != source && minCostFlow.Tail(i) != 11 && minCostFlow.Tail(i) != 12 &&
                    minCostFlow.Head(i) != sink)
                {
                    // Arcs in the solution have a flow value of 1. Their start and end nodes
                    // give an assignment of worker to task.
                    if (minCostFlow.Flow(i) > 0)
                    {
                        Console.WriteLine("Worker " + minCostFlow.Tail(i) + " assigned to task " + minCostFlow.Head(i) +
                                          " Cost: " + minCostFlow.UnitCost(i));
                    }
                }
            }
        }
        else
        {
            Console.WriteLine("Solving the min cost flow problem failed.");
            Console.WriteLine("Solver status: " + status);
        }
    }
}