def setUp(self):
filename = os.path.join(BASE_DATA_PATH,'completeness_test_cat.csv')
parser0 = CsvCatalogueParser(filename)
self.catalogue = parser0.read_file()
self.config = {'maximum_iterations': 1000,
'number_earthquakes': 100,
'number_samples': 51,
'tolerance': 0.05}
self.model = KijkoNonParametricGaussian()
def setUp(self):
parser0 = CsvCatalogueParser(TEST_CAT_1)
self.catalogue = parser0.read_file()
self.config = {'maximum_iterations': 1000,
'number_earthquakes': 100,
'number_samples': 51,
'tolerance': 0.05}
self.model = KijkoNonParametricGaussian()
def setUp(self):
filename = os.path.join(BASE_DATA_PATH,'completeness_test_cat.csv')
parser0 = CsvCatalogueParser(filename)
self.catalogue = parser0.read_file()
self.config = {'maximum_iterations': 1000,
'number_earthquakes': 100,
'number_samples': 51,
'tolerance': 0.05}
self.model = KijkoNonParametricGaussian()
'input_mmin': 4.5,
'input_mmax': None,
'input_mmax_uncertainty': 0.5}
mmax_ksb = KijkoSellevolBayes()
mmax, mmax_sigma = mmax_ksb.get_mmax(catalogue, mmax_config)
print 'Mmax = %8.3f +/- %8.3f' %(mmax, mmax_sigma)
# In[ ]:
mmax_config = {'number_earthquakes': 100, # Selects the N largest earthquakes in the catalogue for analysis
'input_mmax': None,
'input_mmax_uncertainty': 0.5}
mmax_knpg = KijkoNonParametricGaussian()
mmax, mmax_sigma = mmax_knpg.get_mmax(catalogue, mmax_config)
print 'Mmax = %8.3f +/- %8.3f' %(mmax, mmax_sigma)
# In[ ]:
mmax_config = {'number_bootstraps': 1000} # Number of samples for the uncertainty analyis
mmax_cum_mo = CumulativeMoment()
mmax, mmax_sigma = mmax_cum_mo.get_mmax(catalogue, mmax_config)
print 'Mmax = %8.3f +/- %8.3f' %(mmax, mmax_sigma)
class TestKijkoNPG(unittest.TestCase):
'''
Class to test the Kijko Nonparametric Gaussian function
'''
def setUp(self):
parser0 = CsvCatalogueParser(TEST_CAT_1)
self.catalogue = parser0.read_file()
self.config = {'maximum_iterations': 1000,
'number_earthquakes': 100,
'number_samples': 51,
'tolerance': 0.05}
self.model = KijkoNonParametricGaussian()
def test_get_exponential_values(self):
'''
Tests the function to derive an exponentially spaced set of values.
Tested against Kijko implementation
'''
min_mag = 5.8
max_mag = 7.4
expected_output = np.array(
[5.8, 5.87609089, 5.94679912, 6.01283617, 6.07478116, 6.13311177,
6.18822664, 6.24046187, 6.29010351, 6.33739696, 6.38255438,
6.42576041, 6.46717674, 6.50694576, 6.5451935 , 6.58203209,
6.61756168, 6.65187211, 6.68504428, 6.71715129, 6.74825943,
6.77842898, 6.80771492, 6.83616754, 6.86383296, 6.89075356,
6.9169684 , 6.94251354, 6.96742235, 6.99172576, 7.0154525,
7.0386293 , 7.06128109, 7.08343111, 7.10510113, 7.1263115,
7.14708132, 7.16742851, 7.18736994, 7.20692147, 7.22609806,
7.24491382, 7.26338207, 7.28151542, 7.2993258, 7.31682451,
7.33402228, 7.35092928, 7.36755517, 7.38390916, 7.4])
self.assertTrue(np.allclose(expected_output,
_get_exponential_spaced_values(min_mag, max_mag, 51)))
def test_h_smooth(self):
'''
Function to test the smoothing factor functiob h_smooth
'''
# Test 1: Good magnitude range (4.0 - 8.0)
mag = np.arange(4.5, 8.1, 0.1)
self.assertAlmostEqual(self.model.h_smooth(mag), 0.46)
# Test 2: Bad magnitude
mag = np.array([6.5])
self.assertAlmostEqual(self.model.h_smooth(mag), 0.0)
def test_gauss_cdf(self):
'''
Tests the Gaussian cumulative distribution function
'''
# Simple case where x = -3 to 3 with a step of 1.
xvals = np.arange(-7., 8., 1.)
yvals_expected = np.array(
[0.00000000e+00, 0.00000000e+00, 3.01100756e-05, 8.22638484e-04,
2.36281702e-02, 2.06039365e-01, 3.39612064e-01, 5.00000000e-01,
6.60387936e-01, 7.93960635e-01, 9.76371830e-01, 9.99177362e-01,
9.99969890e-01, 1.00000000e+00, 1.00000000e+00])
self.assertTrue(np.allclose(yvals_expected,
self.model._gauss_cdf_hastings(xvals)))
def test_kijko_npg_intfunc_simps(self):
'''
Tests the integration function using Simpson's rule
'''
# Simple test using test catalogue data - verified against
# implementation in Kijko's own code
# Get the largest 100 events from the catalogue
idx = np.flipud(np.argsort(self.catalogue.data['magnitude']))
test_mag = self.catalogue.data['magnitude'][idx[:100]]
h_fact = self.model.h_smooth(test_mag)
mvals = _get_exponential_spaced_values(np.min(test_mag),
np.max(test_mag),
51)
self.assertAlmostEqual(0.11026752,
self.model._kijko_npg_intfunc_simps(mvals, test_mag,
np.max(test_mag), h_fact, 100.))
def test_get_mmax(self):
'''
Tests the main get_mmax function. These test results are derived by
applying Kijko's implementation to the top 100 events in the test
catalogue
'''
mmax, mmax_sig = self.model.get_mmax(self.catalogue, self.config)
self.assertAlmostEqual(mmax, 7.5434318)
self.assertAlmostEqual(mmax_sig, 0.17485045)
class TestKijkoNPG(unittest.TestCase):
'''
Class to test the Kijko Nonparametric Gaussian function
'''
def setUp(self):
filename = os.path.join(BASE_DATA_PATH, 'completeness_test_cat.csv')
parser0 = CsvCatalogueParser(filename)
self.catalogue = parser0.read_file()
self.config = {
'maximum_iterations': 1000,
'number_earthquakes': 100,
'number_samples': 51,
'tolerance': 0.05
}
self.model = KijkoNonParametricGaussian()
def test_get_exponential_values(self):
# Tests the function to derive an exponentially spaced set of values.
# Tested against Kijko implementation
min_mag = 5.8
max_mag = 7.4
expected_output = np.array([
5.8, 5.87609089, 5.94679912, 6.01283617, 6.07478116, 6.13311177,
6.18822664, 6.24046187, 6.29010351, 6.33739696, 6.38255438,
6.42576041, 6.46717674, 6.50694576, 6.5451935, 6.58203209,
6.61756168, 6.65187211, 6.68504428, 6.71715129, 6.74825943,
6.77842898, 6.80771492, 6.83616754, 6.86383296, 6.89075356,
6.9169684, 6.94251354, 6.96742235, 6.99172576, 7.0154525,
7.0386293, 7.06128109, 7.08343111, 7.10510113, 7.1263115,
7.14708132, 7.16742851, 7.18736994, 7.20692147, 7.22609806,
7.24491382, 7.26338207, 7.28151542, 7.2993258, 7.31682451,
7.33402228, 7.35092928, 7.36755517, 7.38390916, 7.4
])
np.testing.assert_almost_equal(
expected_output,
_get_exponential_spaced_values(min_mag, max_mag, 51))
def test_h_smooth(self):
# Function to test the smoothing factor functiob h_smooth
# Test 1: Good magnitude range (4.0 - 8.0)
mag = np.arange(4.5, 8.1, 0.1)
self.assertAlmostEqual(self.model.h_smooth(mag), 0.46)
# Test 2: Bad magnitude
mag = np.array([6.5])
self.assertAlmostEqual(self.model.h_smooth(mag), 0.0)
def test_gauss_cdf(self):
# Tests the Gaussian cumulative distribution function
# Simple case where x = -3 to 3 with a step of 1.
xvals = np.arange(-7., 8., 1.)
yvals_expected = np.array([
0.00000000e+00, 0.00000000e+00, 3.01100756e-05, 8.22638484e-04,
2.36281702e-02, 2.06039365e-01, 3.39612064e-01, 5.00000000e-01,
6.60387936e-01, 7.93960635e-01, 9.76371830e-01, 9.99177362e-01,
9.99969890e-01, 1.00000000e+00, 1.00000000e+00
])
self.assertTrue(
np.allclose(yvals_expected, self.model._gauss_cdf_hastings(xvals)))
def test_kijko_npg_intfunc_simps(self):
'''
Tests the integration function using Simpson's rule
'''
# Simple test using test catalogue data - verified against
# implementation in Kijko's own code
# Get the largest 100 events from the catalogue
idx = np.flipud(np.argsort(self.catalogue.data['magnitude']))
test_mag = self.catalogue.data['magnitude'][idx[:100]]
h_fact = self.model.h_smooth(test_mag)
mvals = _get_exponential_spaced_values(np.min(test_mag),
np.max(test_mag), 51)
self.assertAlmostEqual(
0.11026752,
self.model._kijko_npg_intfunc_simps(mvals, test_mag,
np.max(test_mag), h_fact,
100.))
def test_get_mmax(self):
# Tests the main get_mmax function. These test results are derived by
# applying Kijko's implementation to the top 100 events in the test
# catalogue
mmax, mmax_sig = self.model.get_mmax(self.catalogue, self.config)
self.assertAlmostEqual(mmax, 7.5434318)
self.assertAlmostEqual(mmax_sig, 0.17485045)
}
mmax_ksb = KijkoSellevolBayes()
mmax, mmax_sigma = mmax_ksb.get_mmax(catalogue, mmax_config)
print 'Mmax = %8.3f +/- %8.3f' % (mmax, mmax_sigma)
# In[ ]:
mmax_config = {
'number_earthquakes':
100, # Selects the N largest earthquakes in the catalogue for analysis
'input_mmax': None,
'input_mmax_uncertainty': 0.5
}
mmax_knpg = KijkoNonParametricGaussian()
mmax, mmax_sigma = mmax_knpg.get_mmax(catalogue, mmax_config)
print 'Mmax = %8.3f +/- %8.3f' % (mmax, mmax_sigma)
# In[ ]:
mmax_config = {
'number_bootstraps': 1000
} # Number of samples for the uncertainty analyis
mmax_cum_mo = CumulativeMoment()
mmax, mmax_sigma = mmax_cum_mo.get_mmax(catalogue, mmax_config)
print 'Mmax = %8.3f +/- %8.3f' % (mmax, mmax_sigma)
