Exemplo n.º 1
0
def test_wavelet_fbm(fbm_file):

    for fname in fbm_file:

        with open(fname, 'rb') as f:
            X = np.load(f)

        j2 = 7
        wt_coefs, _, j2_eff, _ = wavelet_analysis(X, p_exp=None, j2=j2)
        hmin = estimate_hmin(wt_coefs, j1=1, j2_eff=j2_eff, weighted=True)[0]
        hmin = hmin.min()
        gamint = 0.0 if hmin >= 0 else -hmin + 0.1
        wavelet_analysis(X, p_exp=np.inf, j2=j2)
        wavelet_analysis(X, p_exp=2, j2=j2, gamint=gamint)
Exemplo n.º 2
0
def test_mfa_mrw(mrw_file):

    with open('tests/mrw_config.json', 'rb') as f:
        config_list = json.load(f)

    for i, fname in enumerate(mrw_file):

        with open(fname, 'rb') as f:
            X = np.load(f)

        if config_list[i]['H'] == 0.01:
            continue

        j2 = np.log2(X.shape[0]) - 3
        wt_coefs, _, j2_eff, _ = wavelet_analysis(X, p_exp=None, j2=j2)

        hmin = estimate_hmin(wt_coefs, j1=1, j2_eff=j2_eff, weighted=True)[0]
        hmin = hmin.min()
        gamint = 0.0 if hmin >= 0 else -hmin + 0.1

        q = np.array([-2, -1, 0, 1, 2])

        dwt, lwt = mf_analysis_full(X,
                                    j1=3,
                                    j2=j2_eff,
                                    gamint=gamint,
                                    p_exp=np.inf,
                                    n_cumul=3,
                                    q=q)
        assert abs(dwt.structure.H.mean() - config_list[i]['H']) < 0.1
        assert abs(lwt.cumulants.log_cumulants[1, :].mean() +
                   (config_list[i]['lam']**2)) < 0.025

        mf_analysis_full(X,
                         j1=3,
                         j2=j2_eff,
                         gamint=gamint,
                         p_exp=2,
                         n_cumul=3,
                         q=q)
        assert abs(lwt.cumulants.log_cumulants[1, :].mean() +
                   (config_list[i]['lam']**2)) < 0.025
Exemplo n.º 3
0
def test_mfa_fbm(fbm_file):

    with open('tests/fbm_config.json', 'rb') as f:
        config_list = json.load(f)

    for i, fname in enumerate(fbm_file):

        with open(fname, 'rb') as f:
            X = np.load(f)

        j2 = np.log2(config_list[i]['shape']) - 3
        wt_coefs, _, j2_eff, _ = wavelet_analysis(X, p_exp=None, j2=j2)

        hmin = estimate_hmin(wt_coefs, j1=1, j2_eff=j2_eff, weighted=True)[0]
        hmin = hmin.min()
        gamint = 0.0 if hmin >= 0 else -hmin + 0.1

        q = np.array([-2, -1, 0, 1, 2])

        dwt, lwt = mf_analysis_full(X,
                                    j1=3,
                                    j2=j2_eff,
                                    gamint=gamint,
                                    p_exp=np.inf,
                                    n_cumul=3,
                                    q=q)
        if config_list[i]['H'] != 0.01:
            assert abs(dwt.structure.H.mean() - config_list[i]['H']) < 0.1,\
                print(f'{dwt.structure.H.mean()=}, {config_list[i]["H"]=}, '
                      f'{gamint=}')
        assert abs(lwt.cumulants.log_cumulants[1, :].mean()) < 0.01

        _, lwt = mf_analysis_full(X,
                                  j1=3,
                                  j2=j2_eff,
                                  gamint=gamint,
                                  p_exp=2,
                                  n_cumul=3,
                                  q=q)
        assert abs(lwt.cumulants.log_cumulants[1, :].mean()) < 0.01