qaoa_max_cut_border_one_state.py

import json
import networkx as nx
from qaoa_utils import (
    exprectum_hamiltonian,
    create_gibbs_state,
    expected_value,
    build_maxcut_hamiltonian,
    build_mixer_hamiltonian,
    max_cut_brute_force,
    create_local_state,
)
from qibo import models


test = 30
pb_conexion = 0.5
temperature = 4
n_layers = 9



for n_nodes in range(12, 4, -1):
    
    i = 0
    
    while i < test:

        print("case: ", i)

        G = nx.erdos_renyi_graph(n_nodes, pb_conexion)

        dict_values = exprectum_hamiltonian(G)


        gibbs_state = create_gibbs_state(dict_values, t=temperature)

        expected_value_gibbs_state = expected_value(
            state=gibbs_state, dict_values=dict_values
        )


        one_state = create_local_state(
            solutions_dict=dict_values, reference_energy=expected_value_gibbs_state
        )

        expected_value_one_state = expected_value(state=one_state, dict_values=dict_values)

        exact_value = - max_cut_brute_force(G)

        if exact_value != expected_value_one_state:

            hamiltonian = build_maxcut_hamiltonian(graph=G)
            mixer_hamiltonian = build_mixer_hamiltonian(graph=G)

            qaoa = models.QAOA(hamiltonian=hamiltonian, mixer=mixer_hamiltonian)

            initial_parameters = [0] * 2 * n_layers

            best_energy_gibbs_state, _, _ = qaoa.minimize(
                initial_p=initial_parameters,
                method="cma",
                options={"maxfevals": 17000},
                initial_state=gibbs_state,
            )

            best_energy_one_state, _, _ = qaoa.minimize(
                initial_p=initial_parameters,
                method="cma",
                options={"maxfevals": 17000},
                initial_state=one_state,
            )

            path_data = (
                f"logger_data/logger_data_qaoa_performance_gibbs_states/logger_data_qaoa_performance_gibbs_states_borders_one_state/logger_data_max_cut_qaoa_common_instances_n_{n_nodes}_borders/"
                f"cmaes_qaoa_max_cut_n_{n_nodes}_t_{temperature}_"
                f"layers_{n_layers}_case_{i}.json"
            )
            
        
            data_test = {
                "expected_value_gibbs_state": expected_value_gibbs_state,
                "expected_value_one_state": expected_value_one_state,
                "energy_qaoa_gibbs_state": best_energy_gibbs_state,
                "energy_qaoa_one_state": best_energy_one_state,
                "energy_value_exact": exact_value,
            }


            json_datos = json.dumps(data_test, indent=2)

            with open(path_data, 'w') as archivo: 
                archivo.write(json_datos)
                
            i += 1