def test_scale_motion(self):
data = np.array([1, 3, 7, -2, -10, 0])
gm = GM(data, unit='g', dt=0.1)
gm_scaled_1 = gm.scale_motion(factor=2.0) # scale by 2.0
gm_scaled_2 = gm.scale_motion(target_PGA_in_g=5.0) # scale by 0.5
self.assertTrue(np.allclose(gm.accel[:, 1] * 2, gm_scaled_1.accel[:, 1]))
self.assertTrue(np.allclose(gm.accel[:, 1] * 0.5, gm_scaled_2.accel[:, 1]))
def test_amplify_by_tf(self):
#---------- Case 1: An artificial transfer function -------------------
gm = GM('./files/sample_accel.txt', unit='gal')
ratio_benchmark = 2.76
freq = np.arange(0.01, 50, step=0.01)
tf = ratio_benchmark * np.ones_like(freq)
transfer_function = Frequency_Spectrum(np.column_stack((freq, tf)))
new_gm = gm.amplify_by_tf(transfer_function, show_fig=False)
ratio = new_gm.accel[:, 1] / gm.accel[:, 1]
self.assertTrue(np.allclose(ratio, ratio_benchmark))
#---------- Case 2: A transfer function from a Vs profile -------------
vs_prof = Vs_Profile('./files/profile_FKSH14.txt')
tf_RO, tf_BH, _ = vs_prof.get_transfer_function()
gm_with_tf_RO = gm.amplify_by_tf(tf_RO)
gm_with_tf_BH = gm.amplify_by_tf(tf_BH)
gm_with_tf_RO_ = gm.amplify(vs_prof, boundary='elastic')
gm_with_tf_BH_ = gm.amplify(vs_prof, boundary='rigid')
# Assert that `amplify_by_tf()` and `amplify()` can generate
# nearly identical results
self.assertTrue(
self.nearly_identical(gm_with_tf_RO.accel, gm_with_tf_RO_.accel))
self.assertTrue(
self.nearly_identical(gm_with_tf_BH.accel, gm_with_tf_BH_.accel))
def test_integration__artificial_example(self):
gm = GM(_join(f_dir, 'two_column_data_example.txt'), unit='m/s/s')
v_bench = np.array([[0.1000, 0.1000], # from MATLAB
[0.2000, 0.3000],
[0.3000, 0.6000],
[0.4000, 1.0000],
[0.5000, 1.5000],
[0.6000, 1.7000],
[0.7000, 2.0000],
[0.8000, 2.4000],
[0.9000, 2.9000],
[1.0000, 3.5000],
[1.1000, 3.8000],
[1.2000, 4.2000],
[1.3000, 4.7000],
[1.4000, 5.3000],
[1.5000, 6.0000]])
u_bench = np.array([[0.1000, 0.0100], # from MATLAB
[0.2000, 0.0400],
[0.3000, 0.1000],
[0.4000, 0.2000],
[0.5000, 0.3500],
[0.6000, 0.5200],
[0.7000, 0.7200],
[0.8000, 0.9600],
[0.9000, 1.2500],
[1.0000, 1.6000],
[1.1000, 1.9800],
[1.2000, 2.4000],
[1.3000, 2.8700],
[1.4000, 3.4000],
[1.5000, 4.0000]])
self.assertTrue(np.allclose(gm.veloc, v_bench))
self.assertTrue(np.allclose(gm.displ, u_bench))
def test_amplify_by_tf__case_1_an_artificial_transfer_function(self):
gm = GM(_join(f_dir, 'sample_accel.txt'), unit='gal')
ratio_benchmark = 2.76
freq = np.arange(0.01, 50, step=0.01)
tf = ratio_benchmark * np.ones_like(freq)
transfer_function = Frequency_Spectrum(np.column_stack((freq, tf)))
new_gm = gm.amplify_by_tf(transfer_function, show_fig=False)
ratio = new_gm.accel[:, 1] / gm.accel[:, 1]
self.assertTrue(np.allclose(ratio, ratio_benchmark))
def test_unit_convert(self):
data = np.array([1, 3, 7, -2, -10, 0])
gm = GM(data, unit='m', dt=0.1)
accel = gm.accel[:, 1]
accel_in_m = gm._unit_convert(unit='m/s/s')[:, 1]
accel_in_gal = gm._unit_convert(unit='gal')[:, 1]
accel_in_g = gm._unit_convert(unit='g')[:, 1]
self.assertTrue(np.allclose(accel_in_m, accel))
self.assertTrue(np.allclose(accel_in_gal, accel * 100))
self.assertTrue(np.allclose(accel_in_g, accel / 9.81))
def test_filters(self):
gm = GM('./files/sample_accel.txt', unit='m')
hp = gm.highpass(cutoff_freq=1.0, show_fig=True)
lp = gm.lowpass(cutoff_freq=1.0, show_fig=True)
bp = gm.bandpass(cutoff_freq=[0.5, 8], show_fig=True)
bs = gm.bandstop(cutoff_freq=[0.5, 8], show_fig=True)
self.assertTrue(isinstance(hp, GM))
self.assertTrue(isinstance(lp, GM))
self.assertTrue(isinstance(bp, GM))
self.assertTrue(isinstance(bs, GM))
def test_fourier_transform(self):
gm = GM(_join(f_dir, 'two_column_data_example.txt'), unit='m/s/s')
freq, spec = gm.get_Fourier_spectrum(real_val=False).raw_data.T
freq_bench = [0.6667, 1.3333, 2.0000, 2.6667, 3.3333, 4.0000, 4.6667,
5.3333]
FS_bench = [60.0000 + 0.0000j, -1.5000 + 7.0569j, -1.5000 + 3.3691j,
-7.5000 +10.3229j, -1.5000 + 1.3506j, -1.5000 + 0.8660j,
-7.5000 + 2.4369j, -1.5000 + 0.1577j]
self.assertTrue(np.allclose(freq, freq_bench, atol=0.0001, rtol=0.0))
self.assertTrue(np.allclose(spec, FS_bench, atol=0.0001, rtol=0.0))
def test_integration__real_world_example(self):
# Note: In this test, the result by cumulative trapezoidal numerical
# integration is used as the benchmark. Since it is infeasible to
# achieve perfect "alignment" between the two time histories,
# we check the correlation coefficient instead of element-wise
# check.
veloc_ = np.genfromtxt(_join(f_dir, 'sample_accel.txt'))
gm = GM(veloc_, unit='m/s', motion_type='veloc')
displ = gm.displ[:, 1]
displ_cumtrapz = np.append(0, sp.integrate.cumtrapz(veloc_[:, 1], dx=gm.dt))
r = np.corrcoef(displ_cumtrapz, displ)[1, 1] # cross-correlation
self.assertTrue(r >= 0.999)
def test_plot(self):
filename = _join(f_dir, 'sample_accel.txt')
gm = GM(filename, unit='m')
fig, axes = gm.plot() # automatically generate fig/ax objects
self.assertTrue(isinstance(axes, tuple))
self.assertEqual(len(axes), 3)
self.assertEqual(axes[0].title.get_text(), os.path.split(filename)[1])
fig2 = plt.figure(figsize=(8, 8))
fig2_, axes = gm.plot(fig=fig2) # feed an external figure object
self.assertTrue(np.allclose(fig2_.get_size_inches(), (8, 8)))
def test_loading_data__two_columns_from_file(self):
# Two columns from file
gm = GM(_join(f_dir, 'sample_accel.txt'), unit='gal')
PGA_benchmark = 294.30 # unit: cm/s/s
PGV_benchmark = 31.46 # unit: cm/s
PGD_benchmark = 38.77 # unit: cm
tol = 1e-2
self.assertAlmostEqual(gm.pga_in_gal, PGA_benchmark, delta=tol)
self.assertAlmostEqual(gm.pgv_in_cm_s, PGV_benchmark, delta=tol)
self.assertAlmostEqual(gm.pgd_in_cm, PGD_benchmark, delta=tol)
self.assertAlmostEqual(gm.peak_Arias_Intensity, 1.524, delta=tol)
self.assertAlmostEqual(gm.rms_accel, 0.4645, delta=tol)
def test_amplify_by_tf__case_2_a_transfer_function_from_a_Vs_profile(self):
gm = GM(_join(f_dir, 'sample_accel.txt'), unit='gal')
vs_prof = Vs_Profile(_join(f_dir, 'profile_FKSH14.txt'))
tf_RO, tf_BH, _ = vs_prof.get_transfer_function()
gm_with_tf_RO = gm.amplify_by_tf(tf_RO)
gm_with_tf_BH = gm.amplify_by_tf(tf_BH)
gm_with_tf_RO_ = gm.amplify(vs_prof, boundary='elastic')
gm_with_tf_BH_ = gm.amplify(vs_prof, boundary='rigid')
# Assert that `amplify_by_tf()` and `amplify()` can generate
# nearly identical results
self.assertTrue(self.nearly_identical(gm_with_tf_RO.accel,
gm_with_tf_RO_.accel))
self.assertTrue(self.nearly_identical(gm_with_tf_BH.accel,
gm_with_tf_BH_.accel))
def test_deconvolution(self):
'''
Test that `deconvolve()` and `amplify()` are reverse
operations to each other.
'''
gm = GM(_join(f_dir, 'sample_accel.txt'), unit='m')
vs_prof = Vs_Profile(_join(f_dir, 'profile_FKSH14.txt'))
for boundary in ['elastic', 'rigid']:
deconv_motion = gm.deconvolve(vs_prof, boundary=boundary)
output_motion = deconv_motion.amplify(vs_prof, boundary=boundary)
self.assertTrue(self.nearly_identical(gm.accel, output_motion.accel))
amplified_motion = gm.amplify(vs_prof, boundary=boundary)
output_motion = amplified_motion.deconvolve(vs_prof, boundary=boundary)
self.assertTrue(self.nearly_identical(gm.accel, output_motion.accel))
def test_loading_data(self):
# Two columns from file
gm = GM('./files/sample_accel.txt', unit='gal')
PGA_benchmark = 294.30 # unit: cm/s/s
PGV_benchmark = 31.46 # unit: cm/s
PGD_benchmark = 38.77 # unit: cm
tol = 1e-2
self.assertAlmostEqual(gm.pga_in_gal, PGA_benchmark, delta=tol)
self.assertAlmostEqual(gm.pgv_in_cm_s, PGV_benchmark, delta=tol)
self.assertAlmostEqual(gm.pgd_in_cm, PGD_benchmark, delta=tol)
self.assertAlmostEqual(gm.peak_Arias_Intensity, 1.524, delta=tol)
self.assertAlmostEqual(gm.rms_accel, 0.4645, delta=tol)
# Two columns from numpy array
gm = GM(np.array([[0.1, 0.2, 0.3, 0.4], [1, 2, 3, 4]]).T, unit='m/s/s')
self.assertAlmostEqual(gm.pga, 4)
# One column from file
gm = GM('./files/one_column_data_example.txt', unit='g', dt=0.2)
self.assertAlmostEqual(gm.pga_in_g, 12.0)
# One column from numpy array
gm = GM(np.array([1, 2, 3, 4, 5]), unit='gal', dt=0.1)
self.assertAlmostEqual(gm.pga_in_gal, 5.0)
# One column without specifying dt
error_msg = 'is needed for one-column `data`.'
with self.assertRaises(ValueError, msg=error_msg):
gm = GM(np.array([1, 2, 3, 4, 5]), unit='gal')
# Test invalid unit names
with self.assertRaises(ValueError, msg='Invalid `unit` name.'):
GM(np.array([1, 2, 3, 4, 5]), unit='test', dt=0.1)
with self.assertRaises(ValueError, msg="use '/s/s' instead of 's^2'"):
GM(np.array([1, 2, 3, 4, 5]), unit='m/s^2', dt=0.1)
def test_baseline_correction(self):
gm = GM('./files/sample_accel.txt', unit='m/s/s')
corrected = gm.baseline_correct(show_fig=True)
self.assertTrue(isinstance(corrected, GM))
def test_differentiation(self):
veloc = np.array([[.1, .2, .3, .4, .5, .6], [1, 3, 7, -1, -3, 5]]).T
gm = GM(veloc, unit='m', motion_type='veloc')
accel_benchmark = np.array([[.1, .2, .3, .4, .5, .6],
[0, 20, 40, -80, -20, 80]]).T
self.assertTrue(np.allclose(gm.accel, accel_benchmark))
def test_loading_data__test_invalid_unit_names(self):
# Test invalid unit names
with self.assertRaisesRegex(ValueError, 'Invalid `unit` name.'):
GM(np.array([1, 2, 3, 4, 5]), unit='test', dt=0.1)
with self.assertRaisesRegex(ValueError, r"use '/s/s' instead of 's\^2'"):
GM(np.array([1, 2, 3, 4, 5]), unit='m/s^2', dt=0.1)
def test_loading_data__one_column_without_specifying_dt(self):
# One column without specifying dt
error_msg = 'is needed for one-column `data`.'
with self.assertRaisesRegex(ValueError, error_msg):
gm = GM(np.array([1, 2, 3, 4, 5]), unit='gal')
def test_loading_data__one_column_from_numpy_array(self):
# One column from numpy array
gm = GM(np.array([1, 2, 3, 4, 5]), unit='gal', dt=0.1)
self.assertAlmostEqual(gm.pga_in_gal, 5.0)
def test_loading_data__one_column_from_file(self):
# One column from file
gm = GM(_join(f_dir, 'one_column_data_example.txt'), unit='g', dt=0.2)
self.assertAlmostEqual(gm.pga_in_g, 12.0)
def test_loading_data__two_columns_from_numpy_array(self):
# Two columns from numpy array
gm = GM(np.array([[0.1, 0.2, 0.3, 0.4], [1, 2, 3, 4]]).T, unit='m/s/s')
self.assertAlmostEqual(gm.pga, 4)
def test_integration_and_differentiation(self):
'''
Test that converting among acceleration/velocity/displacement is
correct.
'''
#------- Test differentiation -------
veloc = np.array([[.1, .2, .3, .4, .5, .6], [1, 3, 7, -1, -3, 5]]).T
gm = GM(veloc, unit='m', motion_type='veloc')
accel_benchmark = np.array([[.1, .2, .3, .4, .5, .6],
[0, 20, 40, -80, -20, 80]]).T
self.assertTrue(np.allclose(gm.accel, accel_benchmark))
#------- Test integartion (artificial example) ----------
gm = GM('./files/two_column_data_example.txt', unit='m/s/s')
v_bench = np.array([
[0.1000, 0.1000], # from MATLAB
[0.2000, 0.3000],
[0.3000, 0.6000],
[0.4000, 1.0000],
[0.5000, 1.5000],
[0.6000, 1.7000],
[0.7000, 2.0000],
[0.8000, 2.4000],
[0.9000, 2.9000],
[1.0000, 3.5000],
[1.1000, 3.8000],
[1.2000, 4.2000],
[1.3000, 4.7000],
[1.4000, 5.3000],
[1.5000, 6.0000]
])
u_bench = np.array([
[0.1000, 0.0100], # from MATLAB
[0.2000, 0.0400],
[0.3000, 0.1000],
[0.4000, 0.2000],
[0.5000, 0.3500],
[0.6000, 0.5200],
[0.7000, 0.7200],
[0.8000, 0.9600],
[0.9000, 1.2500],
[1.0000, 1.6000],
[1.1000, 1.9800],
[1.2000, 2.4000],
[1.3000, 2.8700],
[1.4000, 3.4000],
[1.5000, 4.0000]
])
self.assertTrue(np.allclose(gm.veloc, v_bench))
self.assertTrue(np.allclose(gm.displ, u_bench))
#------- Test integration (real-world example) -------
# Note: In this test, the result by cumulative trapezoidal numerical
# integration is used as the benchmark. Since it is infeasible to
# achieve perfect "alignment" between the two time histories,
# we check the correlation coefficient instead of element-wise
# check.
veloc_ = np.genfromtxt('./files/sample_accel.txt')
gm = GM(veloc_, unit='m/s', motion_type='veloc')
displ = gm.displ[:, 1]
displ_cumtrapz = np.append(
0, sp.integrate.cumtrapz(veloc_[:, 1], dx=gm.dt))
r = np.corrcoef(displ_cumtrapz, displ)[1, 1] # cross-correlation
self.assertTrue(r >= 0.999)
def test_baseline_correction(self):
gm = GM(_join(f_dir, 'sample_accel.txt'), unit='m/s/s')
corrected = gm.baseline_correct(show_fig=True)
self.assertTrue(isinstance(corrected, GM))
def test_low_pass_filter(self):
gm = GM(_join(f_dir, 'sample_accel.txt'), unit='m')
lp = gm.lowpass(cutoff_freq=1.0, show_fig=True)
self.assertTrue(isinstance(lp, GM))
def test_amplify_via_profile(self):
gm = GM(_join(f_dir, 'sample_accel.txt'), unit='m')
vs_prof = Vs_Profile(_join(f_dir, 'profile_FKSH14.txt'))
output_motion = gm.amplify(vs_prof, boundary='elastic')
self.assertTrue(isinstance(output_motion, GM))
def test_band_stop_filter(self):
gm = GM(_join(f_dir, 'sample_accel.txt'), unit='m')
bs = gm.bandstop(cutoff_freq=[0.5, 8], show_fig=True)
self.assertTrue(isinstance(bs, GM))
