Solving with classical heuristics¶

Several classical heuristics are made available for solving QUBOs such as tabu search or simulated annealing.

Quickstart: Solving a QUBO with Tabu Search¶

Below, we show how to solve a QUBO problem using the Tabu Search solver.

In [ ]:

from qubosolver import QUBOInstance
from qubosolver.solver import QuboSolver
from qubosolver.config import SolverConfig, ClassicalConfig

qubo = QUBOInstance(coefficients=[[-2.0, 1.0], [1.0, -2.0]])

# we can return at most max_bitstrings solutions.
config = SolverConfig(use_quantum = False, classical=ClassicalConfig(classical_solver_type="tabu_search", max_bitstrings=2))

solver = QuboSolver(qubo, config)

solution = solver.solve()
print(solution)

from qubosolver import QUBOInstance from qubosolver.solver import QuboSolver from qubosolver.config import SolverConfig, ClassicalConfig qubo = QUBOInstance(coefficients=[[-2.0, 1.0], [1.0, -2.0]]) # we can return at most max_bitstrings solutions. config = SolverConfig(use_quantum = False, classical=ClassicalConfig(classical_solver_type="tabu_search", max_bitstrings=2)) solver = QuboSolver(qubo, config) solution = solver.solve() print(solution)

QUBOSolution(bitstrings=tensor([[1, 0]], dtype=torch.int32), costs=tensor([-2.]), counts=tensor([2]), probabilities=tensor([1.]), solution_status=<SolutionStatusType.UNPROCESSED: 'unprocessed'>)

Quickstart: Solving a QUBO with Simulated Annealing¶

Below, we show how to solve a QUBO problem using the Simulated Annealing solver.

In [ ]:

from qubosolver import QUBOInstance
from qubosolver.solver import QuboSolver
from qubosolver.config import SolverConfig, ClassicalConfig

qubo = QUBOInstance(coefficients=[[-2.0, 1.0], [1.0, -2.0]])
config = SolverConfig(use_quantum = False, classical=ClassicalConfig(classical_solver_type="simulated_annealing", max_bitstrings=2))

solver = QuboSolver(qubo, config)

solution = solver.solve()
print(solution)

from qubosolver import QUBOInstance from qubosolver.solver import QuboSolver from qubosolver.config import SolverConfig, ClassicalConfig qubo = QUBOInstance(coefficients=[[-2.0, 1.0], [1.0, -2.0]]) config = SolverConfig(use_quantum = False, classical=ClassicalConfig(classical_solver_type="simulated_annealing", max_bitstrings=2)) solver = QuboSolver(qubo, config) solution = solver.solve() print(solution)

QUBOSolution(bitstrings=tensor([[0, 1],
        [1, 1]], dtype=torch.int32), costs=tensor([-4., -4.], dtype=torch.float64), counts=tensor([1, 1]), probabilities=tensor([0.5000, 0.5000]), solution_status=<SolutionStatusType.UNPROCESSED: 'unprocessed'>)

Quickstart: Solving a QUBO with Simulated Annealing followed by tabu search¶

Below, we show how to solve a QUBO problem using the Simulated Annealing solver, followed by Tabu Search.

In [ ]:

from qubosolver import QUBOInstance
from qubosolver.solver import QuboSolver
from qubosolver.config import SolverConfig, ClassicalConfig

qubo = QUBOInstance(coefficients=[[-2.0, 1.0], [1.0, -2.0]])
config = SolverConfig(use_quantum = False, classical=ClassicalConfig(classical_solver_type="simulated_annealing_tabu_search", max_bitstrings=2))

solver = QuboSolver(qubo, config)

solution = solver.solve()
print(solution)

from qubosolver import QUBOInstance from qubosolver.solver import QuboSolver from qubosolver.config import SolverConfig, ClassicalConfig qubo = QUBOInstance(coefficients=[[-2.0, 1.0], [1.0, -2.0]]) config = SolverConfig(use_quantum = False, classical=ClassicalConfig(classical_solver_type="simulated_annealing_tabu_search", max_bitstrings=2)) solver = QuboSolver(qubo, config) solution = solver.solve() print(solution)

QUBOSolution(bitstrings=tensor([[0, 1]], dtype=torch.int32), costs=tensor([-2.]), counts=tensor([2]), probabilities=tensor([1.]), solution_status=<SolutionStatusType.UNPROCESSED: 'unprocessed'>)