def test_hyp_output_dicts(self):
self.hyp_file()
events = parse_hyp('hyptest.hyp')
event = events[0]
a, b, c = hyp_output_dicts(event, False, {'moment_tensor_space': np.matrix([[1., 2.], [2., 1.], [1., 2.], [2., 1.], [
1., 2.], [2., 1.]]), 'ln_pdf': np.matrix([0, 0.7]), 'probability': np.matrix([[1., 2.]]), 'total_number_samples': 400})
self.assertEqual(len(a.split('\n')), len(event['hyp_file'])+1)
self.assertTrue('MOMENTTENSOR' in a)
self.assertEqual(float(a.split()[a.split().index('MTNN')+1]), 2.0)
self.assertEqual(len(b), 169)
events = parse_hyp('hyptest.hyp')
event = events[0]
a, b, c = hyp_output_dicts(event, False, {'moment_tensor_space': np.matrix([[1., 2., 3.], [2., 1., 3.], [1., 2., 3.], [2., 1., 3.], [
1., 2., 3.], [2., 1., 3.]]), 'ln_pdf': np.matrix([0, 0.7, 0]), 'probability': np.matrix([[1., 2., 1.]]), 'total_number_samples': 400})
self.assertEqual(len(a.split('\n')), len(event['hyp_file'])+1)
self.assertTrue('MOMENTTENSOR' in a)
self.assertEqual(float(a.split()[a.split().index('MTNN')+1]), 2.0)
self.assertEqual(len(b), 233)
events = parse_hyp('hyptest.hyp')
event = events[0]
event['hyp_file'].pop(14)
a, b, c = hyp_output_dicts(event, False, {'moment_tensor_space': np.matrix([[1., 2., 3.], [2., 1., 3.], [1., 2., 3.], [2., 1., 3.], [
1., 2., 3.], [2., 1., 3.]]), 'ln_pdf': np.matrix([0, 0.7, 0]), 'probability': np.matrix([[1., 2., 1.]]), 'total_number_samples': 400})
self.assertEqual(len(a.split('\n')), len(event['hyp_file'])+2)
self.assertTrue('MOMENTTENSOR' in a)
self.assertEqual(float(a.split()[a.split().index('MTNN')+1]), 2.0)
self.assertEqual(len(b), 233)
event.pop('hyp_file')
a, b, c = hyp_output_dicts(event, False, {'moment_tensor_space': np.matrix([[1., 2., 3.], [2., 1., 3.], [1., 2., 3.], [2., 1., 3.], [
1., 2., 3.], [2., 1., 3.]]), 'ln_pdf': np.matrix([0, 0.7, 0]), 'probability': np.matrix([[1., 2., 1.]]), 'total_number_samples': 400})
self.assertEqual(len(a.split('\n')), 36)
self.assertTrue('MOMENTTENSOR' in a)
self.assertEqual(float(a.split()[a.split().index('MTNN')+1]), 2.0)
self.assertEqual(len(b), 233)
events = parse_hyp('hyptest.hyp')
event = events[0]
event['hyp_file'].pop(14)
a, b, c = hyp_output_dicts(event, False, {'moment_tensor_space': np.matrix([[1., -0.51969334, 3.], [2., 0.22610635, 3.], [1., 0.29358698, 3.], [2., 0.58532165, 3.], [
1., -0.27015115, 3.], [2., -0.42073549, 3.]]), 'ln_pdf': np.matrix([0, 0.7, 0]), 'probability': np.matrix([[1., 2., 1.]]), 'total_number_samples': 400})
self.assertEqual(len(a.split('\n')), len(event['hyp_file'])+2)
self.assertFalse('MOMENTTENSOR' in a)
self.assertTrue('FOCALMECH' in a)
self.assertAlmostEqual(
float(a.split()[a.split().index('Mech')+1]), 0.5*180/np.pi, 5)
self.assertEqual(len(b), 233)
a, b, c = hyp_output_dicts(event, False, {'moment_tensor_space': np.matrix([[1., -0.51969334, 3.], [2., 0.22610635, 3.], [1., 0.29358698, 3.], [2., 0.58532165, 3.], [1., -0.27015115, 3.], [2., -0.42073549, 3.]]),
'probability': np.matrix([[1., 2., 1.]]), 'ln_pdf': np.matrix([0, 0.7, 0]), 'total_number_samples': 400, 'g': np.array([0.2, 0.2, 0.2]), 'd': np.array([0.2, 0.2, 0.2]), 'k': np.array([0.2, 0.2, 0.2]),
'h': np.array([0.2, 0.2, 0.2]), 's': np.array([0.2, 0.2, 0.2]), 'S1': np.array([0.2, 0.2, 0.2]), 'D1': np.array([0.2, 0.2, 0.2]), 'R1': np.array([0.2, 0.2, 0.2]), 'u': np.array([0.2, 0.2, 0.2]),
'v': np.array([0.2, 0.2, 0.2]), 'S2': np.array([0.2, 0.2, 0.2]), 'D2': np.array([0.2, 0.2, 0.2]), 'R2': np.array([0.2, 0.2, 0.2]), 'ln_bayesian_evidence': (1, 10)})
self.assertEqual(len(a.split('\n')), len(event['hyp_file'])+2)
self.assertFalse('MOMENTTENSOR' in a)
self.assertTrue('FOCALMECH' in a)
self.assertAlmostEqual(
float(a.split()[a.split().index('Mech')+1]), 0.5*180/np.pi, 5)
self.assertEqual(len(b), 545)
def test_parse_hyp(self):
self.hyp_file()
events = parse_hyp('hyptest.hyp')
self.assertEqual(len(events), 1)
self.assertTrue('PPolarity' in events[0].keys())
self.assertTrue('UID' in events[0].keys())
self.assertTrue('hyp_file' in events[0].keys())
self.assertEqual(events[0]['UID'], "20150126222324275")
self.assertEqual(len(events[0].keys()), 3)
def test_hyp_output(self):
self.hyp_file()
events = parse_hyp('hyptest.hyp')
event = events[0]
a, b, c = hyp_output_dicts(event, False, {'moment_tensor_space': np.matrix([[1., 2.], [2., 1.], [1., 2.], [2., 1.], [
1., 2.], [2., 1.]]), 'ln_pdf': np.matrix([0, 0.7, 0]), 'probability': np.matrix([[1., 2.]]), 'total_number_samples': 400})
try:
os.remove('mtfitOUTPUTTEST.hyp')
except Exception:
pass
try:
os.remove('mtfitOUTPUTTEST.mt')
except Exception:
pass
self.assertFalse(os.path.exists('mtfitOUTPUTTEST.hyp'))
self.assertFalse(os.path.exists('mtfitOUTPUTTEST.mt'))
fid, out_str = hyp_output([a, b], fid='mtfitOUTPUTTEST.hyp')
self.assertTrue(os.path.exists('mtfitOUTPUTTEST.hyp'))
self.assertTrue(os.path.exists('mtfitOUTPUTTEST.mt'))
try:
os.remove('mtfitOUTPUTTEST.hyp')
except Exception:
pass
try:
os.remove('mtfitOUTPUTTEST.mt')
except Exception:
pass
event['hyp_file'].pop(14)
a, b, c = hyp_output_dicts(event, False, {'moment_tensor_space': np.matrix([[1., -0.51969334, 3.], [2., 0.22610635, 3.], [1., 0.29358698, 3.], [2., 0.58532165, 3.], [1., -0.27015115, 3.], [2., -0.42073549, 3.]]),
'probability': np.matrix([[1., 2., 1.]]), 'ln_pdf': np.matrix([0, 0.7, 0]), 'total_number_samples': 400, 'g': np.array([0.2, 0.2, 0.2]), 'd': np.array([0.2, 0.2, 0.2]), 'k': np.array([0.2, 0.2, 0.2]),
'h': np.array([0.2, 0.2, 0.2]), 's': np.array([0.2, 0.2, 0.2]), 'S1': np.array([0.2, 0.2, 0.2]), 'D1': np.array([0.2, 0.2, 0.2]), 'R1': np.array([0.2, 0.2, 0.2]), 'u': np.array([0.2, 0.2, 0.2]),
'v': np.array([0.2, 0.2, 0.2]), 'S2': np.array([0.2, 0.2, 0.2]), 'D2': np.array([0.2, 0.2, 0.2]), 'R2': np.array([0.2, 0.2, 0.2]), 'ln_bayesian_evidence': 1.+10})
try:
os.remove('mtfitOUTPUTTEST.hyp')
except Exception:
pass
try:
os.remove('mtfitOUTPUTTEST.mt')
except Exception:
pass
self.assertFalse(os.path.exists('mtfitOUTPUTTEST.hyp'))
self.assertFalse(os.path.exists('mtfitOUTPUTTEST.mt'))
fid, out_str = hyp_output([a, b], fid='mtfitOUTPUTTEST.hyp')
self.assertTrue(os.path.exists('mtfitOUTPUTTEST.hyp'))
self.assertTrue(os.path.exists('mtfitOUTPUTTEST.mt'))
try:
os.remove('mtfitOUTPUTTEST.hyp')
except Exception:
pass
try:
os.remove('mtfitOUTPUTTEST.mt')
except Exception:
pass
def test_read_matlab_output(self):
self.hyp_file()
events = parse_hyp('hyptest.hyp')
event = events[0]
mdict, sdict = full_pdf_output_dicts(event, False, {'moment_tensor_space': np.matrix([[1., 2.], [2., 1.], [1., 2.], [2., 1.], [
1., 2.], [2., 1.]]), 'ln_pdf': np.matrix([0, 0.7]), 'dV': 1, 'probability': np.matrix([[1., 2.]]), 'total_number_samples': 400})
MATLAB_output([mdict, sdict], 'hyptest.mat')
event, stations = read_matlab_output('hyptest.mat')
self.assertAlmostEqual(event, mdict['Events'], 10)
self.assertAlmostEqual(
stations['azimuth'], mdict['Stations'][:, 1], 10)
self.assertAlmostEqual(
stations['takeoff_angle'], mdict['Stations'][:, 2], 10)
self.assertAlmostEqual(
stations['polarity'], mdict['Stations'][:, 3], 10)
def test_read_binary_output(self):
self.hyp_file()
events = parse_hyp('hyptest.hyp')
event = events[0]
x = {'moment_tensor_space': np.matrix([[1., 2.], [2., 1.], [1., 2.], [2., 1.], [1., 2.], [2., 1.]]), 'dkl': 1.2, 'ln_pdf': np.matrix(
[0, 0.7]), 'probability': np.matrix([[1., 2.]]), 'total_number_samples': 400}
a, b, c = hyp_output_dicts(event, False, x)
try:
os.remove('mtfitOUTPUTTEST.hyp')
except Exception:
pass
try:
os.remove('mtfitOUTPUTTEST.mt')
except Exception:
pass
self.assertFalse(os.path.exists('mtfitOUTPUTTEST.hyp'))
self.assertFalse(os.path.exists('mtfitOUTPUTTEST.mt'))
fid, out_str = hyp_output([a, b], fid='mtfitOUTPUTTEST.hyp')
y = read_binary_output('mtfitOUTPUTTEST.mt')
self.assertTrue(os.path.exists('mtfitOUTPUTTEST.hyp'))
self.assertTrue(os.path.exists('mtfitOUTPUTTEST.mt'))
self.assertEqual(sorted(y[0].keys()), sorted(x.keys()))
self.assertTrue(
(y[0]['moment_tensor_space'] == x['moment_tensor_space']).all())
self.assertEqual(y[0]['dkl'], x['dkl'])
try:
os.remove('mtfitOUTPUTTEST.hyp')
except Exception:
pass
try:
os.remove('mtfitOUTPUTTEST.mt')
except Exception:
pass
x = {'moment_tensor_space': np.matrix([[1., -0.51969334, 3.], [2., 0.22610635, 3.], [1., 0.29358698, 3.], [2., 0.58532165, 3.], [1., -0.27015115, 3.], [2., -0.42073549, 3.]]),
'probability': np.matrix([[1., 2., 1.]]), 'dkl': 2.4, 'ln_pdf': np.matrix([0, 0.7, 0]), 'total_number_samples': 400, 'g': np.array([0.1, 0.2, 0.3]), 'd': np.array([0.2, 0.2, 0.2]), 'k': np.array([0.2, 0.2, 0.2]),
'h': np.array([0.2, 0.2, 0.2]), 's': np.array([0.2, 0.2, 0.2]), 'S1': np.array([0.2, 0.2, 0.2]), 'D1': np.array([0.2, 0.2, 0.2]), 'R1': np.array([0.2, 0.2, 0.2]), 'u': np.array([0.2, 0.2, 0.2]),
'v': np.array([0.2, 0.2, 0.2]), 'S2': np.array([0.2, 0.2, 0.2]), 'D2': np.array([0.2, 0.2, 0.2]), 'R2': np.array([0.2, 0.2, 0.2]), 'ln_bayesian_evidence': 1.+10.}
a, b, c = hyp_output_dicts(event, False, x)
try:
os.remove('mtfitOUTPUTTEST.hyp')
except Exception:
pass
try:
os.remove('mtfitOUTPUTTEST.mt')
except Exception:
pass
self.assertFalse(os.path.exists('mtfitOUTPUTTEST.hyp'))
self.assertFalse(os.path.exists('mtfitOUTPUTTEST.mt'))
fid, out_str = hyp_output([a, b], fid='mtfitOUTPUTTEST.hyp')
y = read_binary_output('mtfitOUTPUTTEST.mt')
self.assertTrue(os.path.exists('mtfitOUTPUTTEST.hyp'))
self.assertTrue(os.path.exists('mtfitOUTPUTTEST.mt'))
self.assertEqual(sorted(y[0].keys()), sorted(x.keys()))
self.assertTrue((y[0]['g'] == x['g']).all())
self.assertEqual(
y[0]['ln_bayesian_evidence'], x['ln_bayesian_evidence'])
self.assertEqual(y[0]['dkl'], x['dkl'])
try:
os.remove('mtfitOUTPUTTEST.hyp')
except Exception:
pass
try:
os.remove('mtfitOUTPUTTEST.mt')
except Exception:
pass
def test_full_pdf_output_dicts(self):
self.hyp_file()
events = parse_hyp('hyptest.hyp')
event = events[0]
mdict, sdict = full_pdf_output_dicts(event, False, {'moment_tensor_space': np.matrix([[1., 2.], [2., 1.], [1., 2.], [2., 1.], [
1., 2.], [2., 1.]]), 'ln_pdf': np.matrix([0, 0.7]), 'dV': 1, 'probability': np.matrix([[1., 2.]]), 'total_number_samples': 400})
self.assertTrue('Other' in mdict.keys())
self.assertTrue('Events' in mdict.keys())
self.assertTrue('Stations' in mdict.keys())
self.assertTrue('MTSpace' in mdict['Events'].keys())
self.assertTrue('Probability' in mdict['Events'].keys())
self.assertTrue('ln_pdf' in mdict['Events'].keys())
self.assertTrue('UID' in mdict['Events'].keys())
self.assertTrue('NSamples' in mdict['Events'].keys())
self.assertEqual(mdict['Stations'].shape, (20, 4))
event['PPolarityProbability'] = event['PPolarity'].copy()
event['PPolarityProbability']['Measured'] = np.matrix([[0.6, 0.4],
[0.7, 0.3],
[0.8, 0.2],
[0.67, 0.33],
[0.94, 0.06],
[0.32, 0.68],
[0.96, 0.04],
[0.76, 0.24],
[0.82, 0.18],
[0.12, 0.88],
[0.57, 0.43],
[0.68, 0.32],
[0.51, 0.49],
[0.68, 0.32],
[0.50, 0.50],
[0.02, 0.98],
[0.6, 0.4],
[0.7, 0.3],
[0.8, 0.2],
[0.67, 0.33]
])
event.pop('PPolarity')
mdict, sdict = full_pdf_output_dicts(event, ['PPolarityProbability'], {'moment_tensor_space': np.matrix([[1., 2.], [2., 1.], [1., 2.], [
2., 1.], [1., 2.], [2., 1.]]), 'ln_pdf': np.matrix([0, 0.7]), 'dV': 1, 'probability': np.matrix([[1., 2.]]), 'total_number_samples': 400})
self.assertTrue('Other' in mdict.keys())
self.assertTrue('Events' in mdict.keys())
self.assertTrue('Stations' in mdict.keys())
self.assertTrue('MTSpace' in mdict['Events'].keys())
self.assertTrue('Probability' in mdict['Events'].keys())
self.assertTrue('ln_pdf' in mdict['Events'].keys())
self.assertTrue('UID' in mdict['Events'].keys())
self.assertTrue('NSamples' in mdict['Events'].keys())
self.assertEqual(mdict['Stations'].shape, (20, 4))
self.assertTrue(all(mdict['Stations'][:, 3] == np.array(
[0.6, 0.7, 0.8, 0.67, 0.94, -0.68, 0.96, 0.76, 0.82, -0.88, 0.57, 0.68, 0.51, 0.68, 0.5, -0.98, 0.6, 0.7, 0.8, 0.67])))
mdict, sdict = full_pdf_output_dicts(event, False, {'moment_tensor_space': np.matrix([[1., 2.], [2., 1.], [1., 2.], [2., 1.], [
1., 2.], [2., 1.]]), 'ln_pdf': np.matrix([0, 0.7]), 'dV': 1, 'probability': np.matrix([[1., 2.]]), 'total_number_samples': 400})
self.assertTrue('Other' in mdict.keys())
self.assertTrue('Events' in mdict.keys())
self.assertTrue('Stations' in mdict.keys())
self.assertTrue('MTSpace' in mdict['Events'].keys())
self.assertTrue('Probability' in mdict['Events'].keys())
self.assertTrue('ln_pdf' in mdict['Events'].keys())
self.assertTrue('UID' in mdict['Events'].keys())
self.assertTrue('NSamples' in mdict['Events'].keys())
self.assertEqual(mdict['Stations'].shape, (20, 4))
def run_mtfit(catalog,
nlloc_dir,
parallel=True,
n=8,
algorithm='iterate',
phy_mem=1,
inversion_options='PPolarity',
number_location_samples=5000,
MT=True,
DC=True):
"""
Wrapper on mtfit to run over a catalog for which there are already
polarity picks and .scatangle nlloc files in the specified dir
:param catalog: Catalog of events
:param nlloc_dir: Directory with the necessary nlloc and .scatangle files
:param parallel: Run in parallel?
:param n: Number of cores
:param algorithm: MTfit inversion algorithm
:param phy_mem: A soft memory limit of 1Gb of RAM for estimating the
sample sizes. This is only a soft limit, so no errors are thrown
if the memory usage increases above this.
:param inversion_options: What data to include in the inversion
:param number_location_samples: How many random samples to draw from the
NLLoc location PDF
:return:
"""
for ev in catalog:
eid = str(ev.resource_id).split('/')[-1]
print('Running mtfit for {}'.format(eid))
nlloc_fs = glob('{}/{}*'.format(
nlloc_dir,
str(ev.resource_id).split('/')[-1].split('_')[0]))
# Find the hyp file with update pol information
print(nlloc_fs)
try:
hyp_path = [
path for path in nlloc_fs
if path.endswith('.hyp') and 'sum' not in path.split('.')
and path.split('_')[-1].startswith('pol')
][0]
except IndexError as msg:
print('No NLLoc location for this event. Probably low SNR?')
continue
print(hyp_path)
# Read in data dict
data = parse_hyp(hyp_path)
print(data)
print(type(data))
print(data['PPolarity'])
data['UID'] = '{}_ppolarity'.format(eid)
# Set the convert flag to convert the output to other source parameterisations
convert = True
# Set location uncertainty file path
location_pdf_file_path = [
path for path in nlloc_fs if path.endswith('.scatangle')
][0]
# Handle location uncertainty
# Set number of location samples to use (randomly sampled from PDF) as this
# reduces calculation time
# (each location sample is equivalent to running an additional event)
bin_scatangle = True
if DC:
### First run for DC contrained solution
max_samples = 100000
dc = True
print('Running DC for {}'.format(eid))
mtfit(data,
location_pdf_file_path=location_pdf_file_path,
algorithm=algorithm,
parallel=parallel,
inversion_options=inversion_options,
phy_mem=phy_mem,
dc=dc,
max_samples=max_samples,
convert=convert,
bin_scatangle=bin_scatangle,
number_location_samples=number_location_samples,
n=n)
if MT:
### Now for full MT
# Change max_samples for MT inversion
max_samples = 1000000
dc = False
print('Running full MT for {}'.format(eid))
# Create the inversion object with the set parameters.
mtfit(data,
location_pdf_file_path=location_pdf_file_path,
algorithm=algorithm,
parallel=parallel,
inversion_options=inversion_options,
phy_mem=phy_mem,
max_samples=max_samples,
convert=convert,
dc=dc,
bin_scatangle=bin_scatangle,
number_location_samples=number_location_samples,
n=n)
return