def test_nonan_nomask(self):
dmd = HODMD(d=3)
dmd.fit(X=sample_data)
rec = dmd.reconstructed_data
assert not isinstance(rec, np.ma.MaskedArray)
assert not np.nan in rec
def test_eigs_3(self):
dmd = HODMD(svd_rank=2)
dmd.fit(X=sample_data)
expected_eigs = np.array([
-8.09016994e-01 + 5.87785252e-01j, -4.73868662e-01 + 8.80595532e-01j
])
np.testing.assert_almost_equal(dmd.eigs, expected_eigs, decimal=6)
def test_Atilde_values(self):
dmd = HODMD(svd_rank=2)
dmd.fit(X=sample_data)
exact_atilde = np.array(
[[-0.70558526 + 0.67815084j, 0.22914898 + 0.20020143j],
[0.10459069 + 0.09137814j, -0.57730040 + 0.79022994j]])
np.testing.assert_allclose(exact_atilde, dmd.atilde)
def test_reconstruction_method_constructor(self):
assert HODMD(
reconstruction_method='mean')._reconstruction_method == 'mean'
assert HODMD(reconstruction_method=[3])._reconstruction_method == [3]
assert all(
HODMD(reconstruction_method=np.array([1, 2]),
d=2)._reconstruction_method == np.array([1, 2]))
def test_nonan_nomask(self):
dmd = HODMD(d=3)
dmd.fit(X=sample_data)
for timeindex in range(sample_data.shape[1]):
ds = dmd.reconstructions_of_timeindex(timeindex)
for mx in ds:
assert not np.ma.is_masked(mx)
assert np.nan not in mx
def test_reconstruction_method_constructor_error(self):
with self.assertRaises(ValueError):
HODMD(reconstruction_method=[1, 2, 3], d=4)
with self.assertRaises(ValueError):
HODMD(reconstruction_method=np.array([1, 2, 3]), d=4)
with self.assertRaises(ValueError):
HODMD(reconstruction_method=np.array([[1, 2, 3], [3, 4, 5]]), d=3)
def test_dynamics_opt_2(self):
dmd = HODMD(svd_rank=1, opt=True)
dmd.fit(X=sample_data)
expected_dynamics = np.array([[
-4.56004133 - 6.48054238j, 7.61228319 + 1.4801793j,
-6.37489962 + 4.11788355j, 1.70548899 - 7.22866146j,
3.69875496 + 6.25701574j, -6.85298745 - 1.90654427j,
6.12829151 - 3.30212967j, -2.08469012 + 6.48584004j,
-2.92745126 - 5.99004747j, 6.12772217 + 2.24123565j,
-5.84352626 + 2.57413711j, 2.37745273 - 5.77906544j,
2.24158249 + 5.68989493j, -5.44023459 - 2.49457492j,
5.53024740 - 1.92916437j
]])
np.testing.assert_allclose(dmd.dynamics, expected_dynamics)
def test_dynamics_2(self):
dmd = HODMD(svd_rank=1)
dmd.fit(X=sample_data)
expected_dynamics = np.array([[
-2.20639502 - 9.10168802e-16j, 1.55679980 - 1.49626864e+00j,
-0.08375915 + 2.11149018e+00j, -1.37280962 - 1.54663768e+00j,
2.01748787 + 1.60312745e-01j, -1.53222592 + 1.25504678e+00j,
0.23000498 - 1.92462280e+00j, 1.14289644 + 1.51396355e+00j,
-1.83310653 - 2.93174173e-01j, 1.49222925 - 1.03626336e+00j,
-0.35015209 + 1.74312867e+00j, -0.93504202 - 1.46738182e+00j,
1.65485808 + 4.01263449e-01j, -1.43976061 + 8.39117825e-01j,
0.44682540 - 1.56844403e+00j
]])
np.testing.assert_allclose(dmd.dynamics, expected_dynamics)
def test_dmd_time_3(self):
dmd = HODMD()
dmd.fit(X=sample_data)
dmd.dmd_time['t0'] = 8
dmd.dmd_time['tend'] = 11
expected_data = sample_data[:, 8:12]
np.testing.assert_allclose(dmd.reconstructed_data, expected_data)
def test_dmd_time_4(self):
dmd = HODMD(svd_rank=3)
dmd.fit(X=sample_data)
dmd.dmd_time['t0'] = 20
dmd.dmd_time['tend'] = 20
expected_data = np.array([[-7.29383297e+00 - 4.90248179e-14j],
[-5.69109796e+00 - 2.74068833e+00j],
[3.38410649e-83 + 3.75677740e-83j]])
np.testing.assert_almost_equal(dmd.dynamics, expected_data, decimal=6)
def test_rank(self):
dmd = HODMD(svd_rank=0.9)
dmd.fit(X=sample_data)
assert len(dmd.eigs) == 2
def test_truncation_shape(self):
dmd = HODMD(svd_rank=3)
dmd.fit(X=sample_data)
assert dmd.modes.shape[1] == 3
def test_shape(self):
dmd = HODMD(svd_rank=-1)
dmd.fit(X=sample_data)
assert dmd.modes.shape[1] == sample_data.shape[1] - 1
def test_rec_method_weighted(self):
dmd = HODMD(d=2, reconstruction_method=[10, 20])
dmd.fit(X=sample_data)
assert (dmd.reconstructed_data.T[4] == np.average(
dmd.reconstructions_of_timeindex(4), axis=0,
weights=[10, 20]).T).all()
def test_rec_method_mean(self):
dmd = HODMD(d=3, reconstruction_method='mean')
dmd.fit(X=sample_data)
assert (dmd.reconstructed_data.T[2] == np.mean(
dmd.reconstructions_of_timeindex(2), axis=0).T).all()
def test_tdmd_plot(self):
dmd = HODMD(tlsq_rank=3)
dmd.fit(X=sample_data)
dmd.plot_eigs(show_axes=False, show_unit_circle=False)
plt.close()
def test_plot_snapshots_2(self):
dmd = HODMD(svd_rank=-1)
dmd.fit(X=sample_data)
dmd.plot_snapshots_2D((1, 2, 5), x=np.arange(20), y=np.arange(20))
plt.close()
def test_dynamics_1(self):
dmd = HODMD(svd_rank=5)
dmd.fit(X=sample_data)
assert dmd.dynamics.shape == (5, sample_data.shape[1])
def test_reconstruction_method_default_constructor(self):
assert HODMD()._reconstruction_method == 'first'
def test_plot_snapshots_4(self):
dmd = HODMD()
snapshots = [snap.reshape(20, 20) for snap in sample_data.T]
dmd.fit(X=snapshots)
dmd.plot_snapshots_2D(index_snap=2)
plt.close()
def test_sorted_eigs_param(self):
dmd = HODMD(sorted_eigs='real')
assert dmd.operator._sorted_eigs == 'real'
def test_sorted_eigs_default(self):
dmd = HODMD()
assert dmd.operator._sorted_eigs == False
def test_Atilde_shape(self):
dmd = HODMD(svd_rank=3)
dmd.fit(X=sample_data)
assert dmd.atilde.shape == (dmd.svd_rank, dmd.svd_rank)
def test_plot_snapshots_1(self):
dmd = HODMD()
dmd.fit(X=sample_data)
with self.assertRaises(ValueError):
dmd.plot_snapshots_2D()
def test_d(self):
single_data = np.sin(np.linspace(0, 10, 100))
dmd = HODMD(svd_rank=-1, d=50, opt=True)
dmd.fit(single_data)
assert np.allclose(dmd.reconstructed_data.flatten(), single_data)
def test_rec_method_first(self):
dmd = HODMD(d=3, reconstruction_method='first')
dmd.fit(X=sample_data)
assert (dmd.reconstructed_data == dmd.reconstructions_of_timeindex()
[:, 0].T).all()
def test_eigs_2(self):
dmd = HODMD(svd_rank=5)
dmd.fit(X=sample_data)
assert len(dmd.eigs) == 5
def test_plot_snapshots_5(self):
dmd = HODMD()
snapshots = [snap.reshape(20, 20) for snap in sample_data.T]
dmd.fit(X=snapshots)
dmd.plot_snapshots_2D(index_snap=2, filename='tmp.png')
self.addCleanup(os.remove, 'tmp.2.png')
def test_extract_versions_nonan(self):
dmd = HODMD(d=3)
dmd.fit(X=sample_data)
for timeindex in range(sample_data.shape[1]):
assert not np.nan in dmd.reconstructions_of_timeindex(timeindex)
def test_plot_modes_3(self):
dmd = HODMD()
snapshots = [snap.reshape(20, 20) for snap in sample_data.T]
dmd.fit(X=snapshots)
dmd.plot_modes_2D()
plt.close()