Organizacje, których pracownicy pracują na wielu zmianach, muszą zaplanować wystarczającą liczbę pracowników na każdą dzienną zmianę. Zwykle harmonogramy zawierają ograniczenia, np. „Żaden pracownik nie powinien pracować na 2 zmiany z rzędu”. Znalezienie harmonogramu, który spełniałby wszystkie ograniczenia, może być trudne do obliczenia.
W kolejnych sekcjach przedstawiamy 2 przykłady problemów z planowaniem pracowników oraz sposoby ich rozwiązywania za pomocą narzędzia do rozwiązywania problemów z CP-SAT.
Bardziej zaawansowany przykład znajdziesz w tym programie do planowania przesunięcia na GitHubie.
Problem z planem pracy pielęgniarki
W następnym przykładzie kierownik szpitala musi stworzyć harmonogram dla 4 pielęgniarzy w okresie 3 dni, uwzględniając te warunki:
- Każdy dzień jest podzielony na 3 8-godzinne zmiany.
- Codziennie na każdą zmianę przypisywana jest jedna pielęgniarka, która nie pracuje więcej niż na jednej zmianie.
- Każda pielęgniarka jest przypisywana na co najmniej 2 zmiany w trakcie trzydniowego okresu.
W sekcjach poniżej przedstawiamy rozwiązanie problemu z planowaniem pracy pielęgniarskiej.
Importowanie bibliotek
Ten kod importuje wymaganą bibliotekę.
Python
from ortools.sat.python import cp_model
C++
#include <stdlib.h> #include <atomic> #include <map> #include <numeric> #include <string> #include <tuple> #include <vector> #include "absl/strings/str_format.h" #include "ortools/base/logging.h" #include "ortools/sat/cp_model.h" #include "ortools/sat/cp_model.pb.h" #include "ortools/sat/cp_model_solver.h" #include "ortools/sat/model.h" #include "ortools/sat/sat_parameters.pb.h" #include "ortools/util/time_limit.h"
Java
import com.google.ortools.Loader; import com.google.ortools.sat.CpModel; import com.google.ortools.sat.CpSolver; import com.google.ortools.sat.CpSolverSolutionCallback; import com.google.ortools.sat.CpSolverStatus; import com.google.ortools.sat.LinearExpr; import com.google.ortools.sat.LinearExprBuilder; import com.google.ortools.sat.Literal; import java.util.ArrayList; import java.util.List; import java.util.stream.IntStream;
C#
using System; using System.Collections.Generic; using System.IO; using System.Linq; using Google.OrTools.Sat;
Dane dla przykładu
Dane na potrzeby przykładu zostaną utworzone za pomocą podanego niżej kodu.
Python
num_nurses = 4 num_shifts = 3 num_days = 3 all_nurses = range(num_nurses) all_shifts = range(num_shifts) all_days = range(num_days)
C++
const int num_nurses = 4; const int num_shifts = 3; const int num_days = 3; std::vector<int> all_nurses(num_nurses); std::iota(all_nurses.begin(), all_nurses.end(), 0); std::vector<int> all_shifts(num_shifts); std::iota(all_shifts.begin(), all_shifts.end(), 0); std::vector<int> all_days(num_days); std::iota(all_days.begin(), all_days.end(), 0);
Java
final int numNurses = 4; final int numDays = 3; final int numShifts = 3; final int[] allNurses = IntStream.range(0, numNurses).toArray(); final int[] allDays = IntStream.range(0, numDays).toArray(); final int[] allShifts = IntStream.range(0, numShifts).toArray();
C#
const int numNurses = 4; const int numDays = 3; const int numShifts = 3; int[] allNurses = Enumerable.Range(0, numNurses).ToArray(); int[] allDays = Enumerable.Range(0, numDays).ToArray(); int[] allShifts = Enumerable.Range(0, numShifts).ToArray();
Tworzenie modelu
Poniższy kod tworzy model.
Python
model = cp_model.CpModel()
C++
CpModelBuilder cp_model;
Java
CpModel model = new CpModel();
C#
CpModel model = new CpModel(); model.Model.Variables.Capacity = numNurses * numDays * numShifts;
Tworzenie zmiennych
Poniższy kod tworzy tablicę zmiennych.
Python
shifts = {}
for n in all_nurses:
for d in all_days:
for s in all_shifts:
shifts[(n, d, s)] = model.new_bool_var(f"shift_n{n}_d{d}_s{s}")
C++
std::map<std::tuple<int, int, int>, BoolVar> shifts;
for (int n : all_nurses) {
for (int d : all_days) {
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
shifts[key] = cp_model.NewBoolVar().WithName(
absl::StrFormat("shift_n%dd%ds%d", n, d, s));
}
}
}
Java
Literal[][][] shifts = new Literal[numNurses][numDays][numShifts];
for (int n : allNurses) {
for (int d : allDays) {
for (int s : allShifts) {
shifts[n][d][s] = model.newBoolVar("shifts_n" + n + "d" + d + "s" + s);
}
}
}
C#
Dictionary<(int, int, int), BoolVar> shifts =
new Dictionary<(int, int, int), BoolVar>(numNurses * numDays * numShifts);
foreach (int n in allNurses)
{
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
shifts.Add((n, d, s), model.NewBoolVar($"shifts_n{n}d{d}s{s}"));
}
}
}
Tablica definiuje przydziały dla pielęgniarek w taki sposób:
shifts[(n, d, s)] równa się 1, jeśli zmiana s jest przypisana do pielęgniarza/pielęgniarki n w dniu d, lub 0 w przeciwnym razie.
Przypisywanie pielęgniarek na zmiany
Następnie pokażemy, jak przypisywać pielęgniarkę na zmiany z uwzględnieniem tych ograniczeń:
- Każda zmiana jest przypisywana do jednej pielęgniarki dziennie.
- Każda pielęgniarka pracuje na maksymalnie 1 zmianie dziennie.
Oto kod, który tworzy pierwszy warunek.
Python
for d in all_days:
for s in all_shifts:
model.add_exactly_one(shifts[(n, d, s)] for n in all_nurses)
C++
for (int d : all_days) {
for (int s : all_shifts) {
std::vector<BoolVar> nurses;
for (int n : all_nurses) {
auto key = std::make_tuple(n, d, s);
nurses.push_back(shifts[key]);
}
cp_model.AddExactlyOne(nurses);
}
}
Java
for (int d : allDays) {
for (int s : allShifts) {
List<Literal> nurses = new ArrayList<>();
for (int n : allNurses) {
nurses.add(shifts[n][d][s]);
}
model.addExactlyOne(nurses);
}
}
C#
List<ILiteral> literals = new List<ILiteral>();
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
foreach (int n in allNurses)
{
literals.Add(shifts[(n, d, s)]);
}
model.AddExactlyOne(literals);
literals.Clear();
}
}
Ostatni wiersz mówi, że na każdą zmianę suma przypisanych do niej pielęgniarek wynosi 1.
Następnie kod, który wymaga, aby każda pielęgniarka pracowała nie częściej niż na jedną zmianę dziennie.
Python
for n in all_nurses:
for d in all_days:
model.add_at_most_one(shifts[(n, d, s)] for s in all_shifts)
C++
for (int n : all_nurses) {
for (int d : all_days) {
std::vector<BoolVar> work;
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
work.push_back(shifts[key]);
}
cp_model.AddAtMostOne(work);
}
}
Java
for (int n : allNurses) {
for (int d : allDays) {
List<Literal> work = new ArrayList<>();
for (int s : allShifts) {
work.add(shifts[n][d][s]);
}
model.addAtMostOne(work);
}
}
C#
foreach (int n in allNurses)
{
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
literals.Add(shifts[(n, d, s)]);
}
model.AddAtMostOne(literals);
literals.Clear();
}
}
Suma zmian przypisanych do każdej pielęgniarki wynosi maksymalnie 1 („co najwyżej”, ponieważ może ona mieć dzień wolny).
Przypisz zmiany równomiernie
Następnie pokazujemy, jak możliwie równomiernie przypisywać zmiany do pielęgniarek. Ponieważ w ciągu 3 dni obowiązuje 9 zmian, możemy przypisać każdą z tych 4 pielęgniarzy dwie zmiany. Potem pozostanie jedna zmiana, którą można przypisać dowolnej pielęgniarce.
Poniższy kod zapewnia, że każda pielęgniarka będzie pracować na co najmniej 2 zmiany w ciągu 3-dniowego okresu.
Python
# Try to distribute the shifts evenly, so that each nurse works
# min_shifts_per_nurse shifts. If this is not possible, because the total
# number of shifts is not divisible by the number of nurses, some nurses will
# be assigned one more shift.
min_shifts_per_nurse = (num_shifts * num_days) // num_nurses
if num_shifts * num_days % num_nurses == 0:
max_shifts_per_nurse = min_shifts_per_nurse
else:
max_shifts_per_nurse = min_shifts_per_nurse + 1
for n in all_nurses:
shifts_worked = []
for d in all_days:
for s in all_shifts:
shifts_worked.append(shifts[(n, d, s)])
model.add(min_shifts_per_nurse <= sum(shifts_worked))
model.add(sum(shifts_worked) <= max_shifts_per_nurse)
C++
// Try to distribute the shifts evenly, so that each nurse works
// min_shifts_per_nurse shifts. If this is not possible, because the total
// number of shifts is not divisible by the number of nurses, some nurses will
// be assigned one more shift.
int min_shifts_per_nurse = (num_shifts * num_days) / num_nurses;
int max_shifts_per_nurse;
if ((num_shifts * num_days) % num_nurses == 0) {
max_shifts_per_nurse = min_shifts_per_nurse;
} else {
max_shifts_per_nurse = min_shifts_per_nurse + 1;
}
for (int n : all_nurses) {
std::vector<BoolVar> shifts_worked;
for (int d : all_days) {
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
shifts_worked.push_back(shifts[key]);
}
}
cp_model.AddLessOrEqual(min_shifts_per_nurse,
LinearExpr::Sum(shifts_worked));
cp_model.AddLessOrEqual(LinearExpr::Sum(shifts_worked),
max_shifts_per_nurse);
}
Java
// Try to distribute the shifts evenly, so that each nurse works
// minShiftsPerNurse shifts. If this is not possible, because the total
// number of shifts is not divisible by the number of nurses, some nurses will
// be assigned one more shift.
int minShiftsPerNurse = (numShifts * numDays) / numNurses;
int maxShiftsPerNurse;
if ((numShifts * numDays) % numNurses == 0) {
maxShiftsPerNurse = minShiftsPerNurse;
} else {
maxShiftsPerNurse = minShiftsPerNurse + 1;
}
for (int n : allNurses) {
LinearExprBuilder shiftsWorked = LinearExpr.newBuilder();
for (int d : allDays) {
for (int s : allShifts) {
shiftsWorked.add(shifts[n][d][s]);
}
}
model.addLinearConstraint(shiftsWorked, minShiftsPerNurse, maxShiftsPerNurse);
}
C#
// Try to distribute the shifts evenly, so that each nurse works
// minShiftsPerNurse shifts. If this is not possible, because the total
// number of shifts is not divisible by the number of nurses, some nurses will
// be assigned one more shift.
int minShiftsPerNurse = (numShifts * numDays) / numNurses;
int maxShiftsPerNurse;
if ((numShifts * numDays) % numNurses == 0)
{
maxShiftsPerNurse = minShiftsPerNurse;
}
else
{
maxShiftsPerNurse = minShiftsPerNurse + 1;
}
List<IntVar> shiftsWorked = new List<IntVar>();
foreach (int n in allNurses)
{
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
shiftsWorked.Add(shifts[(n, d, s)]);
}
}
model.AddLinearConstraint(LinearExpr.Sum(shiftsWorked), minShiftsPerNurse, maxShiftsPerNurse);
shiftsWorked.Clear();
}
Liczba wszystkich korekt w okresie harmonogramu wynosi num_shifts * num_days, więc możesz przypisać co najmniej (num_shifts * num_days) // num_nurses
która zajmuje się konkretną pielęgniarką, ale niektóre mogą nie zostać zaliczone. // to operator dzielenia się liczbą całkowitą w języku Python, który zwraca minimalną wartość zwykłego ilorazu.
Przy podanych wartościach num_nurses = 4, num_shifts = 3 i num_days = 3 wyrażenie min_shifts_per_nurse zawiera wartość (3 * 3 // 4) = 2, więc do każdej pielęgniarki możesz przypisać co najmniej 2 zmiany. Jest to określone przez ograniczenie (w Pythonie)
model.add(min_shifts_per_nurse <= sum(num_shifts_worked))
Ponieważ w ciągu 3 dni jest łącznie 9 zmian, po przypisaniu każdej pielęgniarki pozostaje jedna zmiana. Dodatkową zmianę można przypisać dowolnej pielęgniarce.
Ostatni wiersz (w Pythonie)
model.add(sum(num_shifts_worked) <= max_shifts_per_nurse)
zapewnia, że żadnej pielęgniarce nie zostanie przydzielona więcej niż jedna dodatkowa zmiana.
W tym przypadku ograniczenie nie jest konieczne, ponieważ jest tylko jedno dodatkowe przesunięcie. Jednak w przypadku różnych wartości parametrów może wystąpić kilka dodatkowych przesunięcia, co oznacza, że konieczne jest wprowadzenie ograniczenia.
Aktualizowanie parametrów rozwiązania
W modelu bez optymalizacji możesz włączyć wyszukiwanie dla wszystkich rozwiązań.
Python
solver = cp_model.CpSolver() solver.parameters.linearization_level = 0 # Enumerate all solutions. solver.parameters.enumerate_all_solutions = True
C++
Model model; SatParameters parameters; parameters.set_linearization_level(0); // Enumerate all solutions. parameters.set_enumerate_all_solutions(true); model.Add(NewSatParameters(parameters));
Java
CpSolver solver = new CpSolver(); solver.getParameters().setLinearizationLevel(0); // Tell the solver to enumerate all solutions. solver.getParameters().setEnumerateAllSolutions(true);
C#
CpSolver solver = new CpSolver(); // Tell the solver to enumerate all solutions. solver.StringParameters += "linearization_level:0 " + "enumerate_all_solutions:true ";
Rejestrowanie wywołania zwrotnego rozwiązania
Musisz zarejestrować wywołanie zwrotne w rozwiązaniu, które będzie wywoływane przy każdym rozwiązaniu.
Python
class NursesPartialSolutionPrinter(cp_model.CpSolverSolutionCallback):
"""Print intermediate solutions."""
def __init__(self, shifts, num_nurses, num_days, num_shifts, limit):
cp_model.CpSolverSolutionCallback.__init__(self)
self._shifts = shifts
self._num_nurses = num_nurses
self._num_days = num_days
self._num_shifts = num_shifts
self._solution_count = 0
self._solution_limit = limit
def on_solution_callback(self):
self._solution_count += 1
print(f"Solution {self._solution_count}")
for d in range(self._num_days):
print(f"Day {d}")
for n in range(self._num_nurses):
is_working = False
for s in range(self._num_shifts):
if self.value(self._shifts[(n, d, s)]):
is_working = True
print(f" Nurse {n} works shift {s}")
if not is_working:
print(f" Nurse {n} does not work")
if self._solution_count >= self._solution_limit:
print(f"Stop search after {self._solution_limit} solutions")
self.stop_search()
def solutionCount(self):
return self._solution_count
# Display the first five solutions.
solution_limit = 5
solution_printer = NursesPartialSolutionPrinter(
shifts, num_nurses, num_days, num_shifts, solution_limit
)
C++
// Create an atomic Boolean that will be periodically checked by the limit.
std::atomic<bool> stopped(false);
model.GetOrCreate<TimeLimit>()->RegisterExternalBooleanAsLimit(&stopped);
const int kSolutionLimit = 5;
int num_solutions = 0;
model.Add(NewFeasibleSolutionObserver([&](const CpSolverResponse& r) {
LOG(INFO) << "Solution " << num_solutions;
for (int d : all_days) {
LOG(INFO) << "Day " << std::to_string(d);
for (int n : all_nurses) {
bool is_working = false;
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
if (SolutionIntegerValue(r, shifts[key])) {
is_working = true;
LOG(INFO) << " Nurse " << std::to_string(n) << " works shift "
<< std::to_string(s);
}
}
if (!is_working) {
LOG(INFO) << " Nurse " << std::to_string(n) << " does not work";
}
}
}
num_solutions++;
if (num_solutions >= kSolutionLimit) {
stopped = true;
LOG(INFO) << "Stop search after " << kSolutionLimit << " solutions.";
}
}));
Java
final int solutionLimit = 5;
class VarArraySolutionPrinterWithLimit extends CpSolverSolutionCallback {
public VarArraySolutionPrinterWithLimit(
int[] allNurses, int[] allDays, int[] allShifts, Literal[][][] shifts, int limit) {
solutionCount = 0;
this.allNurses = allNurses;
this.allDays = allDays;
this.allShifts = allShifts;
this.shifts = shifts;
solutionLimit = limit;
}
@Override
public void onSolutionCallback() {
System.out.printf("Solution #%d:%n", solutionCount);
for (int d : allDays) {
System.out.printf("Day %d%n", d);
for (int n : allNurses) {
boolean isWorking = false;
for (int s : allShifts) {
if (booleanValue(shifts[n][d][s])) {
isWorking = true;
System.out.printf(" Nurse %d work shift %d%n", n, s);
}
}
if (!isWorking) {
System.out.printf(" Nurse %d does not work%n", n);
}
}
}
solutionCount++;
if (solutionCount >= solutionLimit) {
System.out.printf("Stop search after %d solutions%n", solutionLimit);
stopSearch();
}
}
public int getSolutionCount() {
return solutionCount;
}
private int solutionCount;
private final int[] allNurses;
private final int[] allDays;
private final int[] allShifts;
private final Literal[][][] shifts;
private final int solutionLimit;
}
VarArraySolutionPrinterWithLimit cb =
new VarArraySolutionPrinterWithLimit(allNurses, allDays, allShifts, shifts, solutionLimit);
C#
Najpierw zdefiniuj klasę SolutionPrinter.
public class SolutionPrinter : CpSolverSolutionCallback
{
public SolutionPrinter(int[] allNurses, int[] allDays, int[] allShifts,
Dictionary<(int, int, int), BoolVar> shifts, int limit)
{
solutionCount_ = 0;
allNurses_ = allNurses;
allDays_ = allDays;
allShifts_ = allShifts;
shifts_ = shifts;
solutionLimit_ = limit;
}
public override void OnSolutionCallback()
{
Console.WriteLine($"Solution #{solutionCount_}:");
foreach (int d in allDays_)
{
Console.WriteLine($"Day {d}");
foreach (int n in allNurses_)
{
bool isWorking = false;
foreach (int s in allShifts_)
{
if (Value(shifts_[(n, d, s)]) == 1L)
{
isWorking = true;
Console.WriteLine($" Nurse {n} work shift {s}");
}
}
if (!isWorking)
{
Console.WriteLine($" Nurse {d} does not work");
}
}
}
solutionCount_++;
if (solutionCount_ >= solutionLimit_)
{
Console.WriteLine($"Stop search after {solutionLimit_} solutions");
StopSearch();
}
}
public int SolutionCount()
{
return solutionCount_;
}
private int solutionCount_;
private int[] allNurses_;
private int[] allDays_;
private int[] allShifts_;
private Dictionary<(int, int, int), BoolVar> shifts_;
private int solutionLimit_;
}
Następnie utwórz instancję za pomocą:
const int solutionLimit = 5; SolutionPrinter cb = new SolutionPrinter(allNurses, allDays, allShifts, shifts, solutionLimit);
Wywoływanie rozwiązania
Poniższy kod wywołuje funkcję rozwiązania i wyświetla pierwsze 5 rozwiązań.
Python
solver.solve(model, solution_printer)
C++
const CpSolverResponse response = SolveCpModel(cp_model.Build(), &model);
Java
CpSolverStatus status = solver.solve(model, cb);
System.out.println("Status: " + status);
System.out.println(cb.getSolutionCount() + " solutions found.");
C#
CpSolverStatus status = solver.Solve(model, cb);
Console.WriteLine($"Solve status: {status}");
Rozwiązania
Oto 5 pierwszych rozwiązań.
Solution 0
Day 0
Nurse 0 does not work
Nurse 1 works shift 0
Nurse 2 works shift 1
Nurse 3 works shift 2
Day 1
Nurse 0 works shift 2
Nurse 1 does not work
Nurse 2 works shift 1
Nurse 3 works shift 0
Day 2
Nurse 0 works shift 2
Nurse 1 works shift 1
Nurse 2 works shift 0
Nurse 3 does not work
Solution 1
Day 0
Nurse 0 works shift 0
Nurse 1 does not work
Nurse 2 works shift 1
Nurse 3 works shift 2
Day 1
Nurse 0 does not work
Nurse 1 works shift 2
Nurse 2 works shift 1
Nurse 3 works shift 0
Day 2
Nurse 0 works shift 2
Nurse 1 works shift 1
Nurse 2 works shift 0
Nurse 3 does not work
Solution 2
Day 0 Nurse 0 works shift 0
Nurse 1 does not work
Nurse 2 works shift 1
Nurse 3 works shift 2
Day 1
Nurse 0 works shift 1
Nurse 1 works shift 2
Nurse 2 does not work
Nurse 3 works shift 0
Day 2
Nurse 0 works shift 2
Nurse 1 works shift 1
Nurse 2 works shift 0
Nurse 3 does not work
Solution 3
Day 0 Nurse 0 does not work
Nurse 1 works shift 0
Nurse 2 works shift 1
Nurse 3 works shift 2
Day 1
Nurse 0 works shift 1
Nurse 1 works shift 2
Nurse 2 does not work
Nurse 3 works shift 0
Day 2
Nurse 0 works shift 2
Nurse 1 works shift 1
Nurse 2 works shift 0
Nurse 3 does not work
Solution 4
Day 0
Nurse 0 does not work
Nurse 1 works shift 0
Nurse 2 works shift 1
Nurse 3 works shift 2
Day 1
Nurse 0 works shift 2
Nurse 1 works shift 1
Nurse 2 does not work
Nurse 3 works shift 0
Day 2
Nurse 0 works shift 2
Nurse 1 works shift 1
Nurse 2 works shift 0
Nurse 3 does not work
Statistics
- conflicts : 5
- branches : 142
- wall time : 0.002484 s
- solutions found: 5
Łączna liczba rozwiązań to 5184. Przyczynę podano poniżej.
Po pierwsze są 4 opcje dla jednej pielęgniarki, która pracuje na dodatkową zmianę. Po wybraniu tej pielęgniarki można przypisać ją na 3 zmiany w każdym z 3 dni, więc liczba możliwych sposobów przypisania do niej z dodatkową zmianą to 4 · 33 = 108. Po przypisaniu tej pielęgniarki są w danym dniu 2 nieprzypisane zmiany.
Jedna z pozostałych 3 pielęgniarzy pracuje w dniach 0 i 1, jedna pracuje w dni 0 i 2, a druga pracuje w dniu 1 i 2. Jest ich 3. = 6 sposobów przypisywania personelu pielęgniarskiego, co pokazano na wykresie poniżej. (Te 3 pielęgniarki są oznaczone etykietami A, B i C, ale nie przypisaliśmy ich jeszcze do pracy zmianowej).
Day 0 Day 1 Day 2
A B A C B C
A B B C A C
A C A B B C
A C B C A B
B C A B A C
B C A C A B
W przypadku każdego wiersza na powyższym schemacie jest 23 = 8 sposobów na przydzielenie pozostałych zmian pielęgniarkom (2 możliwości dziennie). Łączna liczba możliwych przypisań to 108·6·8 = 5184.
Cały program
Oto cały program dotyczący harmonogramu pracy pielęgniarki.
Python
"""Example of a simple nurse scheduling problem."""
from ortools.sat.python import cp_model
def main() -> None:
# Data.
num_nurses = 4
num_shifts = 3
num_days = 3
all_nurses = range(num_nurses)
all_shifts = range(num_shifts)
all_days = range(num_days)
# Creates the model.
model = cp_model.CpModel()
# Creates shift variables.
# shifts[(n, d, s)]: nurse 'n' works shift 's' on day 'd'.
shifts = {}
for n in all_nurses:
for d in all_days:
for s in all_shifts:
shifts[(n, d, s)] = model.new_bool_var(f"shift_n{n}_d{d}_s{s}")
# Each shift is assigned to exactly one nurse in the schedule period.
for d in all_days:
for s in all_shifts:
model.add_exactly_one(shifts[(n, d, s)] for n in all_nurses)
# Each nurse works at most one shift per day.
for n in all_nurses:
for d in all_days:
model.add_at_most_one(shifts[(n, d, s)] for s in all_shifts)
# Try to distribute the shifts evenly, so that each nurse works
# min_shifts_per_nurse shifts. If this is not possible, because the total
# number of shifts is not divisible by the number of nurses, some nurses will
# be assigned one more shift.
min_shifts_per_nurse = (num_shifts * num_days) // num_nurses
if num_shifts * num_days % num_nurses == 0:
max_shifts_per_nurse = min_shifts_per_nurse
else:
max_shifts_per_nurse = min_shifts_per_nurse + 1
for n in all_nurses:
shifts_worked = []
for d in all_days:
for s in all_shifts:
shifts_worked.append(shifts[(n, d, s)])
model.add(min_shifts_per_nurse <= sum(shifts_worked))
model.add(sum(shifts_worked) <= max_shifts_per_nurse)
# Creates the solver and solve.
solver = cp_model.CpSolver()
solver.parameters.linearization_level = 0
# Enumerate all solutions.
solver.parameters.enumerate_all_solutions = True
class NursesPartialSolutionPrinter(cp_model.CpSolverSolutionCallback):
"""Print intermediate solutions."""
def __init__(self, shifts, num_nurses, num_days, num_shifts, limit):
cp_model.CpSolverSolutionCallback.__init__(self)
self._shifts = shifts
self._num_nurses = num_nurses
self._num_days = num_days
self._num_shifts = num_shifts
self._solution_count = 0
self._solution_limit = limit
def on_solution_callback(self):
self._solution_count += 1
print(f"Solution {self._solution_count}")
for d in range(self._num_days):
print(f"Day {d}")
for n in range(self._num_nurses):
is_working = False
for s in range(self._num_shifts):
if self.value(self._shifts[(n, d, s)]):
is_working = True
print(f" Nurse {n} works shift {s}")
if not is_working:
print(f" Nurse {n} does not work")
if self._solution_count >= self._solution_limit:
print(f"Stop search after {self._solution_limit} solutions")
self.stop_search()
def solutionCount(self):
return self._solution_count
# Display the first five solutions.
solution_limit = 5
solution_printer = NursesPartialSolutionPrinter(
shifts, num_nurses, num_days, num_shifts, solution_limit
)
solver.solve(model, solution_printer)
# Statistics.
print("\nStatistics")
print(f" - conflicts : {solver.num_conflicts}")
print(f" - branches : {solver.num_branches}")
print(f" - wall time : {solver.wall_time} s")
print(f" - solutions found: {solution_printer.solutionCount()}")
if __name__ == "__main__":
main()
C++
// Example of a simple nurse scheduling problem.
#include <stdlib.h>
#include <atomic>
#include <map>
#include <numeric>
#include <string>
#include <tuple>
#include <vector>
#include "absl/strings/str_format.h"
#include "ortools/base/logging.h"
#include "ortools/sat/cp_model.h"
#include "ortools/sat/cp_model.pb.h"
#include "ortools/sat/cp_model_solver.h"
#include "ortools/sat/model.h"
#include "ortools/sat/sat_parameters.pb.h"
#include "ortools/util/time_limit.h"
namespace operations_research {
namespace sat {
void NurseSat() {
const int num_nurses = 4;
const int num_shifts = 3;
const int num_days = 3;
std::vector<int> all_nurses(num_nurses);
std::iota(all_nurses.begin(), all_nurses.end(), 0);
std::vector<int> all_shifts(num_shifts);
std::iota(all_shifts.begin(), all_shifts.end(), 0);
std::vector<int> all_days(num_days);
std::iota(all_days.begin(), all_days.end(), 0);
// Creates the model.
CpModelBuilder cp_model;
// Creates shift variables.
// shifts[(n, d, s)]: nurse 'n' works shift 's' on day 'd'.
std::map<std::tuple<int, int, int>, BoolVar> shifts;
for (int n : all_nurses) {
for (int d : all_days) {
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
shifts[key] = cp_model.NewBoolVar().WithName(
absl::StrFormat("shift_n%dd%ds%d", n, d, s));
}
}
}
// Each shift is assigned to exactly one nurse in the schedule period.
for (int d : all_days) {
for (int s : all_shifts) {
std::vector<BoolVar> nurses;
for (int n : all_nurses) {
auto key = std::make_tuple(n, d, s);
nurses.push_back(shifts[key]);
}
cp_model.AddExactlyOne(nurses);
}
}
// Each nurse works at most one shift per day.
for (int n : all_nurses) {
for (int d : all_days) {
std::vector<BoolVar> work;
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
work.push_back(shifts[key]);
}
cp_model.AddAtMostOne(work);
}
}
// Try to distribute the shifts evenly, so that each nurse works
// min_shifts_per_nurse shifts. If this is not possible, because the total
// number of shifts is not divisible by the number of nurses, some nurses will
// be assigned one more shift.
int min_shifts_per_nurse = (num_shifts * num_days) / num_nurses;
int max_shifts_per_nurse;
if ((num_shifts * num_days) % num_nurses == 0) {
max_shifts_per_nurse = min_shifts_per_nurse;
} else {
max_shifts_per_nurse = min_shifts_per_nurse + 1;
}
for (int n : all_nurses) {
std::vector<BoolVar> shifts_worked;
for (int d : all_days) {
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
shifts_worked.push_back(shifts[key]);
}
}
cp_model.AddLessOrEqual(min_shifts_per_nurse,
LinearExpr::Sum(shifts_worked));
cp_model.AddLessOrEqual(LinearExpr::Sum(shifts_worked),
max_shifts_per_nurse);
}
Model model;
SatParameters parameters;
parameters.set_linearization_level(0);
// Enumerate all solutions.
parameters.set_enumerate_all_solutions(true);
model.Add(NewSatParameters(parameters));
// Display the first five solutions.
// Create an atomic Boolean that will be periodically checked by the limit.
std::atomic<bool> stopped(false);
model.GetOrCreate<TimeLimit>()->RegisterExternalBooleanAsLimit(&stopped);
const int kSolutionLimit = 5;
int num_solutions = 0;
model.Add(NewFeasibleSolutionObserver([&](const CpSolverResponse& r) {
LOG(INFO) << "Solution " << num_solutions;
for (int d : all_days) {
LOG(INFO) << "Day " << std::to_string(d);
for (int n : all_nurses) {
bool is_working = false;
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
if (SolutionIntegerValue(r, shifts[key])) {
is_working = true;
LOG(INFO) << " Nurse " << std::to_string(n) << " works shift "
<< std::to_string(s);
}
}
if (!is_working) {
LOG(INFO) << " Nurse " << std::to_string(n) << " does not work";
}
}
}
num_solutions++;
if (num_solutions >= kSolutionLimit) {
stopped = true;
LOG(INFO) << "Stop search after " << kSolutionLimit << " solutions.";
}
}));
const CpSolverResponse response = SolveCpModel(cp_model.Build(), &model);
// Statistics.
LOG(INFO) << "Statistics";
LOG(INFO) << CpSolverResponseStats(response);
LOG(INFO) << "solutions found : " << std::to_string(num_solutions);
}
} // namespace sat
} // namespace operations_research
int main() {
operations_research::sat::NurseSat();
return EXIT_SUCCESS;
}
Java
package com.google.ortools.sat.samples;
import com.google.ortools.Loader;
import com.google.ortools.sat.CpModel;
import com.google.ortools.sat.CpSolver;
import com.google.ortools.sat.CpSolverSolutionCallback;
import com.google.ortools.sat.CpSolverStatus;
import com.google.ortools.sat.LinearExpr;
import com.google.ortools.sat.LinearExprBuilder;
import com.google.ortools.sat.Literal;
import java.util.ArrayList;
import java.util.List;
import java.util.stream.IntStream;
/** Nurses problem. */
public class NursesSat {
public static void main(String[] args) {
Loader.loadNativeLibraries();
final int numNurses = 4;
final int numDays = 3;
final int numShifts = 3;
final int[] allNurses = IntStream.range(0, numNurses).toArray();
final int[] allDays = IntStream.range(0, numDays).toArray();
final int[] allShifts = IntStream.range(0, numShifts).toArray();
// Creates the model.
CpModel model = new CpModel();
// Creates shift variables.
// shifts[(n, d, s)]: nurse 'n' works shift 's' on day 'd'.
Literal[][][] shifts = new Literal[numNurses][numDays][numShifts];
for (int n : allNurses) {
for (int d : allDays) {
for (int s : allShifts) {
shifts[n][d][s] = model.newBoolVar("shifts_n" + n + "d" + d + "s" + s);
}
}
}
// Each shift is assigned to exactly one nurse in the schedule period.
for (int d : allDays) {
for (int s : allShifts) {
List<Literal> nurses = new ArrayList<>();
for (int n : allNurses) {
nurses.add(shifts[n][d][s]);
}
model.addExactlyOne(nurses);
}
}
// Each nurse works at most one shift per day.
for (int n : allNurses) {
for (int d : allDays) {
List<Literal> work = new ArrayList<>();
for (int s : allShifts) {
work.add(shifts[n][d][s]);
}
model.addAtMostOne(work);
}
}
// Try to distribute the shifts evenly, so that each nurse works
// minShiftsPerNurse shifts. If this is not possible, because the total
// number of shifts is not divisible by the number of nurses, some nurses will
// be assigned one more shift.
int minShiftsPerNurse = (numShifts * numDays) / numNurses;
int maxShiftsPerNurse;
if ((numShifts * numDays) % numNurses == 0) {
maxShiftsPerNurse = minShiftsPerNurse;
} else {
maxShiftsPerNurse = minShiftsPerNurse + 1;
}
for (int n : allNurses) {
LinearExprBuilder shiftsWorked = LinearExpr.newBuilder();
for (int d : allDays) {
for (int s : allShifts) {
shiftsWorked.add(shifts[n][d][s]);
}
}
model.addLinearConstraint(shiftsWorked, minShiftsPerNurse, maxShiftsPerNurse);
}
CpSolver solver = new CpSolver();
solver.getParameters().setLinearizationLevel(0);
// Tell the solver to enumerate all solutions.
solver.getParameters().setEnumerateAllSolutions(true);
// Display the first five solutions.
final int solutionLimit = 5;
class VarArraySolutionPrinterWithLimit extends CpSolverSolutionCallback {
public VarArraySolutionPrinterWithLimit(
int[] allNurses, int[] allDays, int[] allShifts, Literal[][][] shifts, int limit) {
solutionCount = 0;
this.allNurses = allNurses;
this.allDays = allDays;
this.allShifts = allShifts;
this.shifts = shifts;
solutionLimit = limit;
}
@Override
public void onSolutionCallback() {
System.out.printf("Solution #%d:%n", solutionCount);
for (int d : allDays) {
System.out.printf("Day %d%n", d);
for (int n : allNurses) {
boolean isWorking = false;
for (int s : allShifts) {
if (booleanValue(shifts[n][d][s])) {
isWorking = true;
System.out.printf(" Nurse %d work shift %d%n", n, s);
}
}
if (!isWorking) {
System.out.printf(" Nurse %d does not work%n", n);
}
}
}
solutionCount++;
if (solutionCount >= solutionLimit) {
System.out.printf("Stop search after %d solutions%n", solutionLimit);
stopSearch();
}
}
public int getSolutionCount() {
return solutionCount;
}
private int solutionCount;
private final int[] allNurses;
private final int[] allDays;
private final int[] allShifts;
private final Literal[][][] shifts;
private final int solutionLimit;
}
VarArraySolutionPrinterWithLimit cb =
new VarArraySolutionPrinterWithLimit(allNurses, allDays, allShifts, shifts, solutionLimit);
// Creates a solver and solves the model.
CpSolverStatus status = solver.solve(model, cb);
System.out.println("Status: " + status);
System.out.println(cb.getSolutionCount() + " solutions found.");
// Statistics.
System.out.println("Statistics");
System.out.printf(" conflicts: %d%n", solver.numConflicts());
System.out.printf(" branches : %d%n", solver.numBranches());
System.out.printf(" wall time: %f s%n", solver.wallTime());
}
private NursesSat() {}
}
C#
using System;
using System.Collections.Generic;
using System.IO;
using System.Linq;
using Google.OrTools.Sat;
public class NursesSat
{
public class SolutionPrinter : CpSolverSolutionCallback
{
public SolutionPrinter(int[] allNurses, int[] allDays, int[] allShifts,
Dictionary<(int, int, int), BoolVar> shifts, int limit)
{
solutionCount_ = 0;
allNurses_ = allNurses;
allDays_ = allDays;
allShifts_ = allShifts;
shifts_ = shifts;
solutionLimit_ = limit;
}
public override void OnSolutionCallback()
{
Console.WriteLine($"Solution #{solutionCount_}:");
foreach (int d in allDays_)
{
Console.WriteLine($"Day {d}");
foreach (int n in allNurses_)
{
bool isWorking = false;
foreach (int s in allShifts_)
{
if (Value(shifts_[(n, d, s)]) == 1L)
{
isWorking = true;
Console.WriteLine($" Nurse {n} work shift {s}");
}
}
if (!isWorking)
{
Console.WriteLine($" Nurse {d} does not work");
}
}
}
solutionCount_++;
if (solutionCount_ >= solutionLimit_)
{
Console.WriteLine($"Stop search after {solutionLimit_} solutions");
StopSearch();
}
}
public int SolutionCount()
{
return solutionCount_;
}
private int solutionCount_;
private int[] allNurses_;
private int[] allDays_;
private int[] allShifts_;
private Dictionary<(int, int, int), BoolVar> shifts_;
private int solutionLimit_;
}
public static void Main(String[] args)
{
const int numNurses = 4;
const int numDays = 3;
const int numShifts = 3;
int[] allNurses = Enumerable.Range(0, numNurses).ToArray();
int[] allDays = Enumerable.Range(0, numDays).ToArray();
int[] allShifts = Enumerable.Range(0, numShifts).ToArray();
// Creates the model.
CpModel model = new CpModel();
model.Model.Variables.Capacity = numNurses * numDays * numShifts;
// Creates shift variables.
// shifts[(n, d, s)]: nurse 'n' works shift 's' on day 'd'.
Dictionary<(int, int, int), BoolVar> shifts =
new Dictionary<(int, int, int), BoolVar>(numNurses * numDays * numShifts);
foreach (int n in allNurses)
{
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
shifts.Add((n, d, s), model.NewBoolVar($"shifts_n{n}d{d}s{s}"));
}
}
}
// Each shift is assigned to exactly one nurse in the schedule period.
List<ILiteral> literals = new List<ILiteral>();
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
foreach (int n in allNurses)
{
literals.Add(shifts[(n, d, s)]);
}
model.AddExactlyOne(literals);
literals.Clear();
}
}
// Each nurse works at most one shift per day.
foreach (int n in allNurses)
{
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
literals.Add(shifts[(n, d, s)]);
}
model.AddAtMostOne(literals);
literals.Clear();
}
}
// Try to distribute the shifts evenly, so that each nurse works
// minShiftsPerNurse shifts. If this is not possible, because the total
// number of shifts is not divisible by the number of nurses, some nurses will
// be assigned one more shift.
int minShiftsPerNurse = (numShifts * numDays) / numNurses;
int maxShiftsPerNurse;
if ((numShifts * numDays) % numNurses == 0)
{
maxShiftsPerNurse = minShiftsPerNurse;
}
else
{
maxShiftsPerNurse = minShiftsPerNurse + 1;
}
List<IntVar> shiftsWorked = new List<IntVar>();
foreach (int n in allNurses)
{
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
shiftsWorked.Add(shifts[(n, d, s)]);
}
}
model.AddLinearConstraint(LinearExpr.Sum(shiftsWorked), minShiftsPerNurse, maxShiftsPerNurse);
shiftsWorked.Clear();
}
CpSolver solver = new CpSolver();
// Tell the solver to enumerate all solutions.
solver.StringParameters += "linearization_level:0 " + "enumerate_all_solutions:true ";
// Display the first five solutions.
const int solutionLimit = 5;
SolutionPrinter cb = new SolutionPrinter(allNurses, allDays, allShifts, shifts, solutionLimit);
// Solve
CpSolverStatus status = solver.Solve(model, cb);
Console.WriteLine($"Solve status: {status}");
Console.WriteLine("Statistics");
Console.WriteLine($" conflicts: {solver.NumConflicts()}");
Console.WriteLine($" branches : {solver.NumBranches()}");
Console.WriteLine($" wall time: {solver.WallTime()}s");
}
}
Planowanie z użyciem próśb o zmianę
W tej sekcji wykorzystamy poprzedni przykład i dodamy prośby pielęgniarek o konkretne zmiany. Następnie szukamy harmonogramu, który maksymalizuje liczbę zrealizowanych żądań. W przypadku większości problemów z planowaniem najlepiej jest zoptymalizować funkcję obiektywną, ponieważ zwykle nie jest możliwe drukowanie wszystkich możliwych harmonogramów.
Ten przykład zawiera te same ograniczenia co w poprzednim przykładzie.
Importowanie bibliotek
Ten kod importuje wymaganą bibliotekę.
Python
from ortools.sat.python import cp_model
C++
#include <stdlib.h> #include <cstdint> #include <map> #include <numeric> #include <string> #include <tuple> #include <vector> #include "absl/strings/str_format.h" #include "ortools/base/logging.h" #include "ortools/sat/cp_model.h" #include "ortools/sat/cp_model.pb.h" #include "ortools/sat/cp_model_solver.h"
Java
import com.google.ortools.Loader; import com.google.ortools.sat.CpModel; import com.google.ortools.sat.CpSolver; import com.google.ortools.sat.CpSolverStatus; import com.google.ortools.sat.LinearExpr; import com.google.ortools.sat.LinearExprBuilder; import com.google.ortools.sat.Literal; import java.util.ArrayList; import java.util.List; import java.util.stream.IntStream;
C#
using System; using System.Collections.Generic; using System.Linq; using Google.OrTools.Sat;
Dane dla przykładu
Dane z tego przykładu są pokazane później.
Python
num_nurses = 5
num_shifts = 3
num_days = 7
all_nurses = range(num_nurses)
all_shifts = range(num_shifts)
all_days = range(num_days)
shift_requests = [
[[0, 0, 1], [0, 0, 0], [0, 0, 0], [0, 0, 0], [0, 0, 1], [0, 1, 0], [0, 0, 1]],
[[0, 0, 0], [0, 0, 0], [0, 1, 0], [0, 1, 0], [1, 0, 0], [0, 0, 0], [0, 0, 1]],
[[0, 1, 0], [0, 1, 0], [0, 0, 0], [1, 0, 0], [0, 0, 0], [0, 1, 0], [0, 0, 0]],
[[0, 0, 1], [0, 0, 0], [1, 0, 0], [0, 1, 0], [0, 0, 0], [1, 0, 0], [0, 0, 0]],
[[0, 0, 0], [0, 0, 1], [0, 1, 0], [0, 0, 0], [1, 0, 0], [0, 1, 0], [0, 0, 0]],
]
C++
const int num_nurses = 5;
const int num_days = 7;
const int num_shifts = 3;
std::vector<int> all_nurses(num_nurses);
std::iota(all_nurses.begin(), all_nurses.end(), 0);
std::vector<int> all_days(num_days);
std::iota(all_days.begin(), all_days.end(), 0);
std::vector<int> all_shifts(num_shifts);
std::iota(all_shifts.begin(), all_shifts.end(), 0);
std::vector<std::vector<std::vector<int64_t>>> shift_requests = {
{
{0, 0, 1},
{0, 0, 0},
{0, 0, 0},
{0, 0, 0},
{0, 0, 1},
{0, 1, 0},
{0, 0, 1},
},
{
{0, 0, 0},
{0, 0, 0},
{0, 1, 0},
{0, 1, 0},
{1, 0, 0},
{0, 0, 0},
{0, 0, 1},
},
{
{0, 1, 0},
{0, 1, 0},
{0, 0, 0},
{1, 0, 0},
{0, 0, 0},
{0, 1, 0},
{0, 0, 0},
},
{
{0, 0, 1},
{0, 0, 0},
{1, 0, 0},
{0, 1, 0},
{0, 0, 0},
{1, 0, 0},
{0, 0, 0},
},
{
{0, 0, 0},
{0, 0, 1},
{0, 1, 0},
{0, 0, 0},
{1, 0, 0},
{0, 1, 0},
{0, 0, 0},
},
};
Java
final int numNurses = 5;
final int numDays = 7;
final int numShifts = 3;
final int[] allNurses = IntStream.range(0, numNurses).toArray();
final int[] allDays = IntStream.range(0, numDays).toArray();
final int[] allShifts = IntStream.range(0, numShifts).toArray();
final int[][][] shiftRequests = new int[][][] {
{
{0, 0, 1},
{0, 0, 0},
{0, 0, 0},
{0, 0, 0},
{0, 0, 1},
{0, 1, 0},
{0, 0, 1},
},
{
{0, 0, 0},
{0, 0, 0},
{0, 1, 0},
{0, 1, 0},
{1, 0, 0},
{0, 0, 0},
{0, 0, 1},
},
{
{0, 1, 0},
{0, 1, 0},
{0, 0, 0},
{1, 0, 0},
{0, 0, 0},
{0, 1, 0},
{0, 0, 0},
},
{
{0, 0, 1},
{0, 0, 0},
{1, 0, 0},
{0, 1, 0},
{0, 0, 0},
{1, 0, 0},
{0, 0, 0},
},
{
{0, 0, 0},
{0, 0, 1},
{0, 1, 0},
{0, 0, 0},
{1, 0, 0},
{0, 1, 0},
{0, 0, 0},
},
};
C#
const int numNurses = 5;
const int numDays = 7;
const int numShifts = 3;
int[] allNurses = Enumerable.Range(0, numNurses).ToArray();
int[] allDays = Enumerable.Range(0, numDays).ToArray();
int[] allShifts = Enumerable.Range(0, numShifts).ToArray();
int[,,] shiftRequests = new int[,,] {
{
{ 0, 0, 1 },
{ 0, 0, 0 },
{ 0, 0, 0 },
{ 0, 0, 0 },
{ 0, 0, 1 },
{ 0, 1, 0 },
{ 0, 0, 1 },
},
{
{ 0, 0, 0 },
{ 0, 0, 0 },
{ 0, 1, 0 },
{ 0, 1, 0 },
{ 1, 0, 0 },
{ 0, 0, 0 },
{ 0, 0, 1 },
},
{
{ 0, 1, 0 },
{ 0, 1, 0 },
{ 0, 0, 0 },
{ 1, 0, 0 },
{ 0, 0, 0 },
{ 0, 1, 0 },
{ 0, 0, 0 },
},
{
{ 0, 0, 1 },
{ 0, 0, 0 },
{ 1, 0, 0 },
{ 0, 1, 0 },
{ 0, 0, 0 },
{ 1, 0, 0 },
{ 0, 0, 0 },
},
{
{ 0, 0, 0 },
{ 0, 0, 1 },
{ 0, 1, 0 },
{ 0, 0, 0 },
{ 1, 0, 0 },
{ 0, 1, 0 },
{ 0, 0, 0 },
},
};
Tworzenie modelu
Poniższy kod tworzy model.
Python
model = cp_model.CpModel()
C++
CpModelBuilder cp_model;
Java
CpModel model = new CpModel();
C#
CpModel model = new CpModel();
Tworzenie zmiennych
Poniżej znajduje się kod tablicy zmiennych zadania.
Oprócz zmiennych z poprzedniego przykładu dane zawierają też zestaw trójek, które odpowiadają 3 zmianom dziennie. Każdy element potrójnego ma wartość 0 lub 1, co wskazuje, czy zażądano przesunięcia. Na przykład potrójny [0, 0, 1] w piątej pozycji w wierszu 1 oznacza, że pielęgniarka nr 1 prosi o zmianę na 3 w dniu 5.
Python
shifts = {}
for n in all_nurses:
for d in all_days:
for s in all_shifts:
shifts[(n, d, s)] = model.new_bool_var(f"shift_n{n}_d{d}_s{s}")
C++
std::map<std::tuple<int, int, int>, BoolVar> shifts;
for (int n : all_nurses) {
for (int d : all_days) {
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
shifts[key] = cp_model.NewBoolVar().WithName(
absl::StrFormat("shift_n%dd%ds%d", n, d, s));
}
}
}
Java
Literal[][][] shifts = new Literal[numNurses][numDays][numShifts];
for (int n : allNurses) {
for (int d : allDays) {
for (int s : allShifts) {
shifts[n][d][s] = model.newBoolVar("shifts_n" + n + "d" + d + "s" + s);
}
}
}
C#
Dictionary<Tuple<int, int, int>, IntVar> shifts = new Dictionary<Tuple<int, int, int>, IntVar>();
foreach (int n in allNurses)
{
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
shifts.Add(Tuple.Create(n, d, s), model.NewBoolVar($"shifts_n{n}d{d}s{s}"));
}
}
}
Tworzenie ograniczeń
Ten kod tworzy ograniczenia problemu.
Python
for d in all_days:
for s in all_shifts:
model.add_exactly_one(shifts[(n, d, s)] for n in all_nurses)
C++
for (int d : all_days) {
for (int s : all_shifts) {
std::vector<BoolVar> nurses;
for (int n : all_nurses) {
auto key = std::make_tuple(n, d, s);
nurses.push_back(shifts[key]);
}
cp_model.AddExactlyOne(nurses);
}
}
Java
for (int d : allDays) {
for (int s : allShifts) {
List<Literal> nurses = new ArrayList<>();
for (int n : allNurses) {
nurses.add(shifts[n][d][s]);
}
model.addExactlyOne(nurses);
}
}
C#
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
IntVar[] x = new IntVar[numNurses];
foreach (int n in allNurses)
{
var key = Tuple.Create(n, d, s);
x[n] = shifts[key];
}
model.Add(LinearExpr.Sum(x) == 1);
}
}
Python
for n in all_nurses:
for d in all_days:
model.add_at_most_one(shifts[(n, d, s)] for s in all_shifts)
C++
for (int n : all_nurses) {
for (int d : all_days) {
std::vector<BoolVar> work;
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
work.push_back(shifts[key]);
}
cp_model.AddAtMostOne(work);
}
}
Java
for (int n : allNurses) {
for (int d : allDays) {
List<Literal> work = new ArrayList<>();
for (int s : allShifts) {
work.add(shifts[n][d][s]);
}
model.addAtMostOne(work);
}
}
C#
foreach (int n in allNurses)
{
foreach (int d in allDays)
{
IntVar[] x = new IntVar[numShifts];
foreach (int s in allShifts)
{
var key = Tuple.Create(n, d, s);
x[s] = shifts[key];
}
model.Add(LinearExpr.Sum(x) <= 1);
}
}
Python
# Try to distribute the shifts evenly, so that each nurse works
# min_shifts_per_nurse shifts. If this is not possible, because the total
# number of shifts is not divisible by the number of nurses, some nurses will
# be assigned one more shift.
min_shifts_per_nurse = (num_shifts * num_days) // num_nurses
if num_shifts * num_days % num_nurses == 0:
max_shifts_per_nurse = min_shifts_per_nurse
else:
max_shifts_per_nurse = min_shifts_per_nurse + 1
for n in all_nurses:
num_shifts_worked = 0
for d in all_days:
for s in all_shifts:
num_shifts_worked += shifts[(n, d, s)]
model.add(min_shifts_per_nurse <= num_shifts_worked)
model.add(num_shifts_worked <= max_shifts_per_nurse)
C++
// Try to distribute the shifts evenly, so that each nurse works
// min_shifts_per_nurse shifts. If this is not possible, because the total
// number of shifts is not divisible by the number of nurses, some nurses will
// be assigned one more shift.
int min_shifts_per_nurse = (num_shifts * num_days) / num_nurses;
int max_shifts_per_nurse;
if ((num_shifts * num_days) % num_nurses == 0) {
max_shifts_per_nurse = min_shifts_per_nurse;
} else {
max_shifts_per_nurse = min_shifts_per_nurse + 1;
}
for (int n : all_nurses) {
LinearExpr num_worked_shifts;
for (int d : all_days) {
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
num_worked_shifts += shifts[key];
}
}
cp_model.AddLessOrEqual(min_shifts_per_nurse, num_worked_shifts);
cp_model.AddLessOrEqual(num_worked_shifts, max_shifts_per_nurse);
}
Java
// Try to distribute the shifts evenly, so that each nurse works
// minShiftsPerNurse shifts. If this is not possible, because the total
// number of shifts is not divisible by the number of nurses, some nurses will
// be assigned one more shift.
int minShiftsPerNurse = (numShifts * numDays) / numNurses;
int maxShiftsPerNurse;
if ((numShifts * numDays) % numNurses == 0) {
maxShiftsPerNurse = minShiftsPerNurse;
} else {
maxShiftsPerNurse = minShiftsPerNurse + 1;
}
for (int n : allNurses) {
LinearExprBuilder numShiftsWorked = LinearExpr.newBuilder();
for (int d : allDays) {
for (int s : allShifts) {
numShiftsWorked.add(shifts[n][d][s]);
}
}
model.addLinearConstraint(numShiftsWorked, minShiftsPerNurse, maxShiftsPerNurse);
}
C#
// Try to distribute the shifts evenly, so that each nurse works
// minShiftsPerNurse shifts. If this is not possible, because the total
// number of shifts is not divisible by the number of nurses, some nurses will
// be assigned one more shift.
int minShiftsPerNurse = (numShifts * numDays) / numNurses;
int maxShiftsPerNurse;
if ((numShifts * numDays) % numNurses == 0)
{
maxShiftsPerNurse = minShiftsPerNurse;
}
else
{
maxShiftsPerNurse = minShiftsPerNurse + 1;
}
foreach (int n in allNurses)
{
IntVar[] numShiftsWorked = new IntVar[numDays * numShifts];
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
var key = Tuple.Create(n, d, s);
numShiftsWorked[d * numShifts + s] = shifts[key];
}
}
model.AddLinearConstraint(LinearExpr.Sum(numShiftsWorked), minShiftsPerNurse, maxShiftsPerNurse);
}
Cel przykładu
Chcemy zoptymalizować podaną niżej funkcję celu.
Python
model.maximize(
sum(
shift_requests[n][d][s] * shifts[(n, d, s)]
for n in all_nurses
for d in all_days
for s in all_shifts
)
)
C++
LinearExpr objective_expr;
for (int n : all_nurses) {
for (int d : all_days) {
for (int s : all_shifts) {
if (shift_requests[n][d][s] == 1) {
auto key = std::make_tuple(n, d, s);
objective_expr += shifts[key] * shift_requests[n][d][s];
}
}
}
}
cp_model.Maximize(objective_expr);
Java
LinearExprBuilder obj = LinearExpr.newBuilder();
for (int n : allNurses) {
for (int d : allDays) {
for (int s : allShifts) {
obj.addTerm(shifts[n][d][s], shiftRequests[n][d][s]);
}
}
}
model.maximize(obj);
C#
IntVar[] flatShifts = new IntVar[numNurses * numDays * numShifts];
int[] flatShiftRequests = new int[numNurses * numDays * numShifts];
foreach (int n in allNurses)
{
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
var key = Tuple.Create(n, d, s);
flatShifts[n * numDays * numShifts + d * numShifts + s] = shifts[key];
flatShiftRequests[n * numDays * numShifts + d * numShifts + s] = shiftRequests[n, d, s];
}
}
}
model.Maximize(LinearExpr.WeightedSum(flatShifts, flatShiftRequests));
Ponieważ shift_requests[n][d][s] * shifts[(n, d, s) ma wartość 1, jeśli zmiana s jest przypisana do pielęgniarki (n) d dnia oraz tej pielęgniarki zażądała tej zmiany (i 0 w innym przypadku), celem jest zmiana liczby projektów, które spełniają wymagania.
Wywoływanie rozwiązania
Ten kod wywołuje funkcję rozwiązania.
Python
solver = cp_model.CpSolver() status = solver.solve(model)
C++
const CpSolverResponse response = Solve(cp_model.Build());
Java
CpSolver solver = new CpSolver(); CpSolverStatus status = solver.solve(model);
C#
CpSolver solver = new CpSolver();
CpSolverStatus status = solver.Solve(model);
Console.WriteLine($"Solve status: {status}");
Wyświetl wyniki
Poniższy kod wyświetla dane wyjściowe, które zawierają optymalny harmonogram (choć nie jedyny). Dane wyjściowe pokazują, które przypisania zmian zostały zgłoszone, a także ile zrealizowano żądań.
Python
if status == cp_model.OPTIMAL:
print("Solution:")
for d in all_days:
print("Day", d)
for n in all_nurses:
for s in all_shifts:
if solver.value(shifts[(n, d, s)]) == 1:
if shift_requests[n][d][s] == 1:
print("Nurse", n, "works shift", s, "(requested).")
else:
print("Nurse", n, "works shift", s, "(not requested).")
print()
print(
f"Number of shift requests met = {solver.objective_value}",
f"(out of {num_nurses * min_shifts_per_nurse})",
)
else:
print("No optimal solution found !")
C++
if (response.status() == CpSolverStatus::OPTIMAL) {
LOG(INFO) << "Solution:";
for (int d : all_days) {
LOG(INFO) << "Day " << std::to_string(d);
for (int n : all_nurses) {
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
if (SolutionIntegerValue(response, shifts[key]) == 1) {
if (shift_requests[n][d][s] == 1) {
LOG(INFO) << " Nurse " << std::to_string(n) << " works shift "
<< std::to_string(s) << " (requested).";
} else {
LOG(INFO) << " Nurse " << std::to_string(n) << " works shift "
<< std::to_string(s) << " (not requested).";
}
}
}
}
LOG(INFO) << "";
}
LOG(INFO) << "Number of shift requests met = " << response.objective_value()
<< " (out of " << num_nurses * min_shifts_per_nurse << ")";
} else {
LOG(INFO) << "No optimal solution found !";
}
Java
if (status == CpSolverStatus.OPTIMAL || status == CpSolverStatus.FEASIBLE) {
System.out.printf("Solution:%n");
for (int d : allDays) {
System.out.printf("Day %d%n", d);
for (int n : allNurses) {
for (int s : allShifts) {
if (solver.booleanValue(shifts[n][d][s])) {
if (shiftRequests[n][d][s] == 1) {
System.out.printf(" Nurse %d works shift %d (requested).%n", n, s);
} else {
System.out.printf(" Nurse %d works shift %d (not requested).%n", n, s);
}
}
}
}
}
System.out.printf("Number of shift requests met = %f (out of %d)%n", solver.objectiveValue(),
numNurses * minShiftsPerNurse);
} else {
System.out.printf("No optimal solution found !");
}
C#
if (status == CpSolverStatus.Optimal || status == CpSolverStatus.Feasible)
{
Console.WriteLine("Solution:");
foreach (int d in allDays)
{
Console.WriteLine($"Day {d}");
foreach (int n in allNurses)
{
bool isWorking = false;
foreach (int s in allShifts)
{
var key = Tuple.Create(n, d, s);
if (solver.Value(shifts[key]) == 1L)
{
if (shiftRequests[n, d, s] == 1)
{
Console.WriteLine($" Nurse {n} work shift {s} (requested).");
}
else
{
Console.WriteLine($" Nurse {n} work shift {s} (not requested).");
}
}
}
}
}
Console.WriteLine(
$"Number of shift requests met = {solver.ObjectiveValue} (out of {numNurses * minShiftsPerNurse}).");
}
else
{
Console.WriteLine("No solution found.");
}
Po uruchomieniu programu wyświetli się ten wynik:
Day 0
Nurse 1 works shift 0 (not requested).
Nurse 2 works shift 1 (requested).
Nurse 3 works shift 2 (requested).
Day 1
Nurse 0 works shift 0 (not requested).
Nurse 2 works shift 1 (requested).
Nurse 4 works shift 2 (requested).
Day 2
Nurse 1 works shift 2 (not requested).
Nurse 3 works shift 0 (requested).
Nurse 4 works shift 1 (requested).
Day 3
Nurse 2 works shift 0 (requested).
Nurse 3 works shift 1 (requested).
Nurse 4 works shift 2 (not requested).
Day 4
Nurse 0 works shift 2 (requested).
Nurse 1 works shift 0 (requested).
Nurse 4 works shift 1 (not requested).
Day 5
Nurse 0 works shift 2 (not requested).
Nurse 2 works shift 1 (requested).
Nurse 3 works shift 0 (requested).
Day 6
Nurse 0 works shift 1 (not requested).
Nurse 1 works shift 2 (requested).
Nurse 4 works shift 0 (not requested).
Statistics
- Number of shift requests met = 13 (out of 20 )
- wall time : 0.003571 s
Cały program
Oto cały program planowania z użyciem próśb o zmianę.
Python
"""Nurse scheduling problem with shift requests."""
from ortools.sat.python import cp_model
def main() -> None:
# This program tries to find an optimal assignment of nurses to shifts
# (3 shifts per day, for 7 days), subject to some constraints (see below).
# Each nurse can request to be assigned to specific shifts.
# The optimal assignment maximizes the number of fulfilled shift requests.
num_nurses = 5
num_shifts = 3
num_days = 7
all_nurses = range(num_nurses)
all_shifts = range(num_shifts)
all_days = range(num_days)
shift_requests = [
[[0, 0, 1], [0, 0, 0], [0, 0, 0], [0, 0, 0], [0, 0, 1], [0, 1, 0], [0, 0, 1]],
[[0, 0, 0], [0, 0, 0], [0, 1, 0], [0, 1, 0], [1, 0, 0], [0, 0, 0], [0, 0, 1]],
[[0, 1, 0], [0, 1, 0], [0, 0, 0], [1, 0, 0], [0, 0, 0], [0, 1, 0], [0, 0, 0]],
[[0, 0, 1], [0, 0, 0], [1, 0, 0], [0, 1, 0], [0, 0, 0], [1, 0, 0], [0, 0, 0]],
[[0, 0, 0], [0, 0, 1], [0, 1, 0], [0, 0, 0], [1, 0, 0], [0, 1, 0], [0, 0, 0]],
]
# Creates the model.
model = cp_model.CpModel()
# Creates shift variables.
# shifts[(n, d, s)]: nurse 'n' works shift 's' on day 'd'.
shifts = {}
for n in all_nurses:
for d in all_days:
for s in all_shifts:
shifts[(n, d, s)] = model.new_bool_var(f"shift_n{n}_d{d}_s{s}")
# Each shift is assigned to exactly one nurse in .
for d in all_days:
for s in all_shifts:
model.add_exactly_one(shifts[(n, d, s)] for n in all_nurses)
# Each nurse works at most one shift per day.
for n in all_nurses:
for d in all_days:
model.add_at_most_one(shifts[(n, d, s)] for s in all_shifts)
# Try to distribute the shifts evenly, so that each nurse works
# min_shifts_per_nurse shifts. If this is not possible, because the total
# number of shifts is not divisible by the number of nurses, some nurses will
# be assigned one more shift.
min_shifts_per_nurse = (num_shifts * num_days) // num_nurses
if num_shifts * num_days % num_nurses == 0:
max_shifts_per_nurse = min_shifts_per_nurse
else:
max_shifts_per_nurse = min_shifts_per_nurse + 1
for n in all_nurses:
num_shifts_worked = 0
for d in all_days:
for s in all_shifts:
num_shifts_worked += shifts[(n, d, s)]
model.add(min_shifts_per_nurse <= num_shifts_worked)
model.add(num_shifts_worked <= max_shifts_per_nurse)
model.maximize(
sum(
shift_requests[n][d][s] * shifts[(n, d, s)]
for n in all_nurses
for d in all_days
for s in all_shifts
)
)
# Creates the solver and solve.
solver = cp_model.CpSolver()
status = solver.solve(model)
if status == cp_model.OPTIMAL:
print("Solution:")
for d in all_days:
print("Day", d)
for n in all_nurses:
for s in all_shifts:
if solver.value(shifts[(n, d, s)]) == 1:
if shift_requests[n][d][s] == 1:
print("Nurse", n, "works shift", s, "(requested).")
else:
print("Nurse", n, "works shift", s, "(not requested).")
print()
print(
f"Number of shift requests met = {solver.objective_value}",
f"(out of {num_nurses * min_shifts_per_nurse})",
)
else:
print("No optimal solution found !")
# Statistics.
print("\nStatistics")
print(f" - conflicts: {solver.num_conflicts}")
print(f" - branches : {solver.num_branches}")
print(f" - wall time: {solver.wall_time}s")
if __name__ == "__main__":
main()
C++
// Nurse scheduling problem with shift requests.
#include <stdlib.h>
#include <cstdint>
#include <map>
#include <numeric>
#include <string>
#include <tuple>
#include <vector>
#include "absl/strings/str_format.h"
#include "ortools/base/logging.h"
#include "ortools/sat/cp_model.h"
#include "ortools/sat/cp_model.pb.h"
#include "ortools/sat/cp_model_solver.h"
namespace operations_research {
namespace sat {
void ScheduleRequestsSat() {
const int num_nurses = 5;
const int num_days = 7;
const int num_shifts = 3;
std::vector<int> all_nurses(num_nurses);
std::iota(all_nurses.begin(), all_nurses.end(), 0);
std::vector<int> all_days(num_days);
std::iota(all_days.begin(), all_days.end(), 0);
std::vector<int> all_shifts(num_shifts);
std::iota(all_shifts.begin(), all_shifts.end(), 0);
std::vector<std::vector<std::vector<int64_t>>> shift_requests = {
{
{0, 0, 1},
{0, 0, 0},
{0, 0, 0},
{0, 0, 0},
{0, 0, 1},
{0, 1, 0},
{0, 0, 1},
},
{
{0, 0, 0},
{0, 0, 0},
{0, 1, 0},
{0, 1, 0},
{1, 0, 0},
{0, 0, 0},
{0, 0, 1},
},
{
{0, 1, 0},
{0, 1, 0},
{0, 0, 0},
{1, 0, 0},
{0, 0, 0},
{0, 1, 0},
{0, 0, 0},
},
{
{0, 0, 1},
{0, 0, 0},
{1, 0, 0},
{0, 1, 0},
{0, 0, 0},
{1, 0, 0},
{0, 0, 0},
},
{
{0, 0, 0},
{0, 0, 1},
{0, 1, 0},
{0, 0, 0},
{1, 0, 0},
{0, 1, 0},
{0, 0, 0},
},
};
// Creates the model.
CpModelBuilder cp_model;
// Creates shift variables.
// shifts[(n, d, s)]: nurse 'n' works shift 's' on day 'd'.
std::map<std::tuple<int, int, int>, BoolVar> shifts;
for (int n : all_nurses) {
for (int d : all_days) {
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
shifts[key] = cp_model.NewBoolVar().WithName(
absl::StrFormat("shift_n%dd%ds%d", n, d, s));
}
}
}
// Each shift is assigned to exactly one nurse in the schedule period.
for (int d : all_days) {
for (int s : all_shifts) {
std::vector<BoolVar> nurses;
for (int n : all_nurses) {
auto key = std::make_tuple(n, d, s);
nurses.push_back(shifts[key]);
}
cp_model.AddExactlyOne(nurses);
}
}
// Each nurse works at most one shift per day.
for (int n : all_nurses) {
for (int d : all_days) {
std::vector<BoolVar> work;
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
work.push_back(shifts[key]);
}
cp_model.AddAtMostOne(work);
}
}
// Try to distribute the shifts evenly, so that each nurse works
// min_shifts_per_nurse shifts. If this is not possible, because the total
// number of shifts is not divisible by the number of nurses, some nurses will
// be assigned one more shift.
int min_shifts_per_nurse = (num_shifts * num_days) / num_nurses;
int max_shifts_per_nurse;
if ((num_shifts * num_days) % num_nurses == 0) {
max_shifts_per_nurse = min_shifts_per_nurse;
} else {
max_shifts_per_nurse = min_shifts_per_nurse + 1;
}
for (int n : all_nurses) {
LinearExpr num_worked_shifts;
for (int d : all_days) {
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
num_worked_shifts += shifts[key];
}
}
cp_model.AddLessOrEqual(min_shifts_per_nurse, num_worked_shifts);
cp_model.AddLessOrEqual(num_worked_shifts, max_shifts_per_nurse);
}
LinearExpr objective_expr;
for (int n : all_nurses) {
for (int d : all_days) {
for (int s : all_shifts) {
if (shift_requests[n][d][s] == 1) {
auto key = std::make_tuple(n, d, s);
objective_expr += shifts[key] * shift_requests[n][d][s];
}
}
}
}
cp_model.Maximize(objective_expr);
const CpSolverResponse response = Solve(cp_model.Build());
if (response.status() == CpSolverStatus::OPTIMAL) {
LOG(INFO) << "Solution:";
for (int d : all_days) {
LOG(INFO) << "Day " << std::to_string(d);
for (int n : all_nurses) {
for (int s : all_shifts) {
auto key = std::make_tuple(n, d, s);
if (SolutionIntegerValue(response, shifts[key]) == 1) {
if (shift_requests[n][d][s] == 1) {
LOG(INFO) << " Nurse " << std::to_string(n) << " works shift "
<< std::to_string(s) << " (requested).";
} else {
LOG(INFO) << " Nurse " << std::to_string(n) << " works shift "
<< std::to_string(s) << " (not requested).";
}
}
}
}
LOG(INFO) << "";
}
LOG(INFO) << "Number of shift requests met = " << response.objective_value()
<< " (out of " << num_nurses * min_shifts_per_nurse << ")";
} else {
LOG(INFO) << "No optimal solution found !";
}
// Statistics.
LOG(INFO) << "Statistics";
LOG(INFO) << CpSolverResponseStats(response);
}
} // namespace sat
} // namespace operations_research
int main() {
operations_research::sat::ScheduleRequestsSat();
return EXIT_SUCCESS;
}
Java
package com.google.ortools.sat.samples;
import com.google.ortools.Loader;
import com.google.ortools.sat.CpModel;
import com.google.ortools.sat.CpSolver;
import com.google.ortools.sat.CpSolverStatus;
import com.google.ortools.sat.LinearExpr;
import com.google.ortools.sat.LinearExprBuilder;
import com.google.ortools.sat.Literal;
import java.util.ArrayList;
import java.util.List;
import java.util.stream.IntStream;
/** Nurses problem with schedule requests. */
public class ScheduleRequestsSat {
public static void main(String[] args) {
Loader.loadNativeLibraries();
final int numNurses = 5;
final int numDays = 7;
final int numShifts = 3;
final int[] allNurses = IntStream.range(0, numNurses).toArray();
final int[] allDays = IntStream.range(0, numDays).toArray();
final int[] allShifts = IntStream.range(0, numShifts).toArray();
final int[][][] shiftRequests = new int[][][] {
{
{0, 0, 1},
{0, 0, 0},
{0, 0, 0},
{0, 0, 0},
{0, 0, 1},
{0, 1, 0},
{0, 0, 1},
},
{
{0, 0, 0},
{0, 0, 0},
{0, 1, 0},
{0, 1, 0},
{1, 0, 0},
{0, 0, 0},
{0, 0, 1},
},
{
{0, 1, 0},
{0, 1, 0},
{0, 0, 0},
{1, 0, 0},
{0, 0, 0},
{0, 1, 0},
{0, 0, 0},
},
{
{0, 0, 1},
{0, 0, 0},
{1, 0, 0},
{0, 1, 0},
{0, 0, 0},
{1, 0, 0},
{0, 0, 0},
},
{
{0, 0, 0},
{0, 0, 1},
{0, 1, 0},
{0, 0, 0},
{1, 0, 0},
{0, 1, 0},
{0, 0, 0},
},
};
// Creates the model.
CpModel model = new CpModel();
// Creates shift variables.
// shifts[(n, d, s)]: nurse 'n' works shift 's' on day 'd'.
Literal[][][] shifts = new Literal[numNurses][numDays][numShifts];
for (int n : allNurses) {
for (int d : allDays) {
for (int s : allShifts) {
shifts[n][d][s] = model.newBoolVar("shifts_n" + n + "d" + d + "s" + s);
}
}
}
// Each shift is assigned to exactly one nurse in the schedule period.
for (int d : allDays) {
for (int s : allShifts) {
List<Literal> nurses = new ArrayList<>();
for (int n : allNurses) {
nurses.add(shifts[n][d][s]);
}
model.addExactlyOne(nurses);
}
}
// Each nurse works at most one shift per day.
for (int n : allNurses) {
for (int d : allDays) {
List<Literal> work = new ArrayList<>();
for (int s : allShifts) {
work.add(shifts[n][d][s]);
}
model.addAtMostOne(work);
}
}
// Try to distribute the shifts evenly, so that each nurse works
// minShiftsPerNurse shifts. If this is not possible, because the total
// number of shifts is not divisible by the number of nurses, some nurses will
// be assigned one more shift.
int minShiftsPerNurse = (numShifts * numDays) / numNurses;
int maxShiftsPerNurse;
if ((numShifts * numDays) % numNurses == 0) {
maxShiftsPerNurse = minShiftsPerNurse;
} else {
maxShiftsPerNurse = minShiftsPerNurse + 1;
}
for (int n : allNurses) {
LinearExprBuilder numShiftsWorked = LinearExpr.newBuilder();
for (int d : allDays) {
for (int s : allShifts) {
numShiftsWorked.add(shifts[n][d][s]);
}
}
model.addLinearConstraint(numShiftsWorked, minShiftsPerNurse, maxShiftsPerNurse);
}
LinearExprBuilder obj = LinearExpr.newBuilder();
for (int n : allNurses) {
for (int d : allDays) {
for (int s : allShifts) {
obj.addTerm(shifts[n][d][s], shiftRequests[n][d][s]);
}
}
}
model.maximize(obj);
// Creates a solver and solves the model.
CpSolver solver = new CpSolver();
CpSolverStatus status = solver.solve(model);
if (status == CpSolverStatus.OPTIMAL || status == CpSolverStatus.FEASIBLE) {
System.out.printf("Solution:%n");
for (int d : allDays) {
System.out.printf("Day %d%n", d);
for (int n : allNurses) {
for (int s : allShifts) {
if (solver.booleanValue(shifts[n][d][s])) {
if (shiftRequests[n][d][s] == 1) {
System.out.printf(" Nurse %d works shift %d (requested).%n", n, s);
} else {
System.out.printf(" Nurse %d works shift %d (not requested).%n", n, s);
}
}
}
}
}
System.out.printf("Number of shift requests met = %f (out of %d)%n", solver.objectiveValue(),
numNurses * minShiftsPerNurse);
} else {
System.out.printf("No optimal solution found !");
}
// Statistics.
System.out.println("Statistics");
System.out.printf(" conflicts: %d%n", solver.numConflicts());
System.out.printf(" branches : %d%n", solver.numBranches());
System.out.printf(" wall time: %f s%n", solver.wallTime());
}
private ScheduleRequestsSat() {}
}
C#
using System;
using System.Collections.Generic;
using System.Linq;
using Google.OrTools.Sat;
public class ScheduleRequestsSat
{
public static void Main(String[] args)
{
const int numNurses = 5;
const int numDays = 7;
const int numShifts = 3;
int[] allNurses = Enumerable.Range(0, numNurses).ToArray();
int[] allDays = Enumerable.Range(0, numDays).ToArray();
int[] allShifts = Enumerable.Range(0, numShifts).ToArray();
int[,,] shiftRequests = new int[,,] {
{
{ 0, 0, 1 },
{ 0, 0, 0 },
{ 0, 0, 0 },
{ 0, 0, 0 },
{ 0, 0, 1 },
{ 0, 1, 0 },
{ 0, 0, 1 },
},
{
{ 0, 0, 0 },
{ 0, 0, 0 },
{ 0, 1, 0 },
{ 0, 1, 0 },
{ 1, 0, 0 },
{ 0, 0, 0 },
{ 0, 0, 1 },
},
{
{ 0, 1, 0 },
{ 0, 1, 0 },
{ 0, 0, 0 },
{ 1, 0, 0 },
{ 0, 0, 0 },
{ 0, 1, 0 },
{ 0, 0, 0 },
},
{
{ 0, 0, 1 },
{ 0, 0, 0 },
{ 1, 0, 0 },
{ 0, 1, 0 },
{ 0, 0, 0 },
{ 1, 0, 0 },
{ 0, 0, 0 },
},
{
{ 0, 0, 0 },
{ 0, 0, 1 },
{ 0, 1, 0 },
{ 0, 0, 0 },
{ 1, 0, 0 },
{ 0, 1, 0 },
{ 0, 0, 0 },
},
};
// Creates the model.
CpModel model = new CpModel();
// Creates shift variables.
// shifts[(n, d, s)]: nurse 'n' works shift 's' on day 'd'.
Dictionary<Tuple<int, int, int>, IntVar> shifts = new Dictionary<Tuple<int, int, int>, IntVar>();
foreach (int n in allNurses)
{
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
shifts.Add(Tuple.Create(n, d, s), model.NewBoolVar($"shifts_n{n}d{d}s{s}"));
}
}
}
// Each shift is assigned to exactly one nurse in the schedule period.
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
IntVar[] x = new IntVar[numNurses];
foreach (int n in allNurses)
{
var key = Tuple.Create(n, d, s);
x[n] = shifts[key];
}
model.Add(LinearExpr.Sum(x) == 1);
}
}
// Each nurse works at most one shift per day.
foreach (int n in allNurses)
{
foreach (int d in allDays)
{
IntVar[] x = new IntVar[numShifts];
foreach (int s in allShifts)
{
var key = Tuple.Create(n, d, s);
x[s] = shifts[key];
}
model.Add(LinearExpr.Sum(x) <= 1);
}
}
// Try to distribute the shifts evenly, so that each nurse works
// minShiftsPerNurse shifts. If this is not possible, because the total
// number of shifts is not divisible by the number of nurses, some nurses will
// be assigned one more shift.
int minShiftsPerNurse = (numShifts * numDays) / numNurses;
int maxShiftsPerNurse;
if ((numShifts * numDays) % numNurses == 0)
{
maxShiftsPerNurse = minShiftsPerNurse;
}
else
{
maxShiftsPerNurse = minShiftsPerNurse + 1;
}
foreach (int n in allNurses)
{
IntVar[] numShiftsWorked = new IntVar[numDays * numShifts];
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
var key = Tuple.Create(n, d, s);
numShiftsWorked[d * numShifts + s] = shifts[key];
}
}
model.AddLinearConstraint(LinearExpr.Sum(numShiftsWorked), minShiftsPerNurse, maxShiftsPerNurse);
}
IntVar[] flatShifts = new IntVar[numNurses * numDays * numShifts];
int[] flatShiftRequests = new int[numNurses * numDays * numShifts];
foreach (int n in allNurses)
{
foreach (int d in allDays)
{
foreach (int s in allShifts)
{
var key = Tuple.Create(n, d, s);
flatShifts[n * numDays * numShifts + d * numShifts + s] = shifts[key];
flatShiftRequests[n * numDays * numShifts + d * numShifts + s] = shiftRequests[n, d, s];
}
}
}
model.Maximize(LinearExpr.WeightedSum(flatShifts, flatShiftRequests));
// Solve
CpSolver solver = new CpSolver();
CpSolverStatus status = solver.Solve(model);
Console.WriteLine($"Solve status: {status}");
if (status == CpSolverStatus.Optimal || status == CpSolverStatus.Feasible)
{
Console.WriteLine("Solution:");
foreach (int d in allDays)
{
Console.WriteLine($"Day {d}");
foreach (int n in allNurses)
{
bool isWorking = false;
foreach (int s in allShifts)
{
var key = Tuple.Create(n, d, s);
if (solver.Value(shifts[key]) == 1L)
{
if (shiftRequests[n, d, s] == 1)
{
Console.WriteLine($" Nurse {n} work shift {s} (requested).");
}
else
{
Console.WriteLine($" Nurse {n} work shift {s} (not requested).");
}
}
}
}
}
Console.WriteLine(
$"Number of shift requests met = {solver.ObjectiveValue} (out of {numNurses * minShiftsPerNurse}).");
}
else
{
Console.WriteLine("No solution found.");
}
Console.WriteLine("Statistics");
Console.WriteLine($" conflicts: {solver.NumConflicts()}");
Console.WriteLine($" branches : {solver.NumBranches()}");
Console.WriteLine($" wall time: {solver.WallTime()}s");
}
}