def persistent_compute(mi_mats, max_dim, bg, time_stamp, basename, tmpdir,
                       process_id, tlabel):
    # Compute persistent entropy and save to temporary file
    N = len(mi_mats)
    if os.path.isdir(tmpdir) == False:
        os.mkdir(tmpdir)
    outfile = os.path.join(tmpdir, basename)

    outstr = defaultdict(list)
    stats = defaultdict(list)

    for i in range(N):
        mimat = mi_mats[i]
        idx = bg + i

        # Calculate distance matrix
        #print(process_id, i, bg, idx, mimat[3][26])
        if tlabel == 'MI':
            netmeasures = nm.pearson(mimat)
            dist = np.sqrt(1.0 - netmeasures[0]**2)
            for j in range(dist.shape[0]):
                dist[j, j] = 0
        else:
            # trace dist, bures_dist, bures_angle
            dist = dm.compute_distance(mimat, tlabel)

        dgms = ph.compute_ph_unit(dist, max_dim=max_dim)

        for j, dgm in enumerate(dgms):
            for pt in dgm:
                outstr[j].append('{} {} {} {}'.format(pt[0], pt[1], 1, idx))
            npent, pent, pnorm, maxnorm = ph.measure_diagram(dgm, p=2)
            stats[j].append('{} {} {} {} {}'.format(idx, npent, pent, pnorm,
                                                    maxnorm))

    # Save diagrams to file txt
    for j, ostr in outstr.items():
        with open('{}_ph_dim_{}_{}.txt'.format(outfile, j, process_id),
                  'w') as file_hdl:
            file_hdl.write('\n'.join(ostr))

    # Save stats to file txt
    for j, ostr in stats.items():
        with open('{}_stats_dim_{}_{}.txt'.format(outfile, j, process_id),
                  'w') as file_hdl:
            file_hdl.write('\n'.join(ostr))
    print('Finished process {} with bg={}, ed={}, outfile={}'.format(
        process_id, bg, bg + N, outfile))
示例#2
0
def main(PostProcess=False, ShowPlots=True, Persistent=False, pdim=0, L=30):
    """
    Introductory example for openMPS to simulate the finite size statics of
    the transverse Ising model. Two modes are available when running the
    example from command line:

    * ``python IsingStatics.py --PostProcess=F`` : runs the MPSFortLib to
      determine the ground state statics of the Ising model. (default if
      ``--PostProcess`` not present.)
    * ``python IsingStatics.py --PostProcess=T`` : plotting the results of the
      simulation run before.
    * The option ``--ShowPlots=T`` (``--ShowPlots=F``) lets you turn on (off)
      the GUI windows with the plots. Default to on if not passed to the
      command line.
    """

    # Build spin operators for spin-1/2 system
    # Transform to perserve Z2 symmetry
    Operators = mps.BuildSpinOperators(0.5)
    Operators['sigmax'] = 2 * Operators['sz']
    Operators['sigmaz'] = (Operators['splus'] + Operators['sminus'])
    Operators['gen'] = np.array([[0, 0], [0, 1.]])
    # Define Hamiltonian of transverse Ising model
    H = mps.MPO(Operators)
    # Note the J parameter in the transverse Ising Hamiltonian
    # has been set to 1, we are modelling a ferromagnetic chain.
    H.AddMPOTerm('bond', ['sigmaz', 'sigmaz'], hparam='J', weight=-1.0)
    H.AddMPOTerm('site', 'sigmax', hparam='g', weight=-1.0)

    # Observables and convergence parameters
    myObservables = mps.Observables(Operators)
    #myObservables.AddObservable('corr', ['sigmaz', 'sigmaz'], 'zz')
    #myObservables.AddObservable('DensityMatrix_i', [])
    #myObservables.AddObservable('DensityMatrix_ij', [])
    myObservables.AddObservable('MI', True)
    myConv = mps.MPSConvParam(max_bond_dimension=200,
                              variance_tol=1E-10,
                              local_tol=1E-10,
                              max_num_sweeps=6)
    #modparam = ['max_bond_dimension', 'max_num_sweeps']
    #myConv.AddModifiedConvergenceParameters(0, modparam, [80, 4])

    # Specify constants and parameter lists
    J = 1.0
    glist = np.linspace(0.1, 2.1, 41)
    parameters = []

    for g in glist:
        parameters.append({
            'simtype':
            'Finite',
            # Directories
            'job_ID':
            'Ising_Statics',
            'unique_ID':
            'g_' + str(g),
            'Write_Directory':
            'TMP_ISING_01_L_{}/'.format(L),
            'Output_Directory':
            'OUTPUTS_ISING_01_L_{}/'.format(L),
            # System size and Hamiltonian parameters
            'L':
            L,
            'J':
            J,
            'g':
            g,
            # ObservablesConvergence parameters
            'verbose':
            1,
            'MPSObservables':
            myObservables,
            'MPSConvergenceParameters':
            myConv,
            'logfile':
            True,
            # Z2 symmetry
            'Discrete_generators': ['gen'],
            'Discrete_quantum_numbers': [0]
        })

    # Write Fortran-readable main files
    MainFiles = mps.WriteFiles(parameters,
                               Operators,
                               H,
                               PostProcess=PostProcess)

    # Run the simulations and quit if we are not just Post
    if (not PostProcess):
        if os.path.isfile('./Execute_MPSMain'):
            RunDir = './'
        else:
            RunDir = None
        mps.runMPS(MainFiles, RunDir=RunDir)
        return

    # PostProcess
    # -----------
    dmat_list = []
    MI_list = []
    Outputs = mps.ReadStaticObservables(parameters)
    for Output in Outputs:
        print(Output['converged'])
        if Persistent:
            netmeasures = nm.pearson(Output['MI'])
            dist = 1.0 - np.abs(netmeasures[0])
            for i in range(L):
                dist[i, i] = 0.0
            dmat_list.append(dist)
        MI_list.append(Output['MI'][3][26])
    plt.style.use('seaborn-colorblind')
    #plt.style.use('seaborn-whitegrid')
    plt.rc('font', family='serif')
    plt.rc('mathtext', fontset='cm')

    timestamp = int(time.time() * 1000.0)

    if (not Persistent):
        plt.scatter(glist, MI_list)
        plt.xlabel(r"transverse field coupling  " r"$g$", fontsize=16)
        plt.ylabel(r"Mutual information (4,27)", fontsize=16)
        plt.savefig('MI_L_{}_{}.pdf'.format(L, timestamp), bbox_inches='tight')
        plt.show()
        return

    npent_list, pent_list, pnorm_list = ph.compute_ph(
        out_path='./diagrams_{}_{}'.format(L, timestamp),
        fig_path='./figs_{}_{}'.format(L, timestamp),
        basename='ising',
        dmat_list=dmat_list,
        plot=True,
        pdim=pdim)

    # Save txt
    np.savetxt('stats_L_{}_{}.txt'.format(L, timestamp),
               list(zip(glist, npent_list, pent_list, pnorm_list)),
               fmt='%.18g')

    # Plot
    fig, ax1 = plt.subplots()
    ax1.scatter(glist, pnorm_list, c='red')
    ax1.set_xlabel(r"transverse field coupling  " r"$g$", fontsize=16)
    ax1.set_ylabel(r"p-norm persistence", fontsize=16)
    ax2 = ax1.twinx()
    ax2.set_ylabel('normalize persistent entropy', fontsize=16)
    ax2.scatter(glist, npent_list, c='blue')
    #plt.xlim((0, 2))
    #plt.ylim((0, 1))
    if (ShowPlots):
        plt.savefig('IsingStatic_L_{}_ph_dim_{}_{}.pdf'.format(
            L, pdim, timestamp),
                    bbox_inches='tight')
        plt.show()

    return