def test_template_match_with_resample():
"""
Test template_match when both observed and template spectra have different wavelength axis using resampling.
"""
np.random.seed(42)
# Create test spectra
spec_axis1 = np.linspace(0, 50, 50) * u.AA
spec_axis2 = np.linspace(0, 50, 50) * u.AA
spec = Spectrum1D(spectral_axis=spec_axis1,
flux=np.random.randn(50) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(50), unit='Jy'))
spec1 = Spectrum1D(spectral_axis=spec_axis2,
flux=np.random.randn(50) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(50), unit='Jy'))
# Get result from template_match
tm_result = template_comparison.template_match(spec, spec1)
# Create new spectrum for comparison
spec_result = Spectrum1D(spectral_axis=spec_axis1,
flux=spec1.flux * template_comparison._normalize_for_template_matching(spec, spec1))
assert quantity_allclose(tm_result[0].flux, spec_result.flux, atol=0.01*u.Jy)
np.testing.assert_almost_equal(tm_result[3], 40093.28353756253)
def test_create_collection_from_collections():
spec = Spectrum1D(spectral_axis=np.linspace(0, 50, 50) * u.AA,
flux=np.random.randn(50) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(50),
unit='Jy'))
spec1 = Spectrum1D(spectral_axis=np.linspace(20, 60, 50) * u.AA,
flux=np.random.randn(50) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(50),
unit='Jy'))
spec_coll1 = SpectrumCollection.from_spectra([spec, spec1])
spec2 = Spectrum1D(spectral_axis=np.linspace(40, 80, 50) * u.AA,
flux=np.random.randn(50) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(50),
unit='Jy'))
spec_coll2 = SpectrumCollection.from_spectra([spec, spec2])
spec_coll = SpectrumCollection.from_spectra(
[spec_coll1, spec_coll2, spec_coll1])
assert spec_coll.ndim == 2
assert spec_coll.shape == (3, 2)
assert spec_coll.nspectral == 50
assert isinstance(spec_coll.flux, u.Quantity)
assert isinstance(spec_coll.spectral_axis, u.Quantity)
assert spec.spectral_axis.unit == spec_coll.spectral_axis.unit
assert spec.flux.unit == spec_coll.flux.unit
def test_template_match_no_overlap():
"""
Test template_match when both observed and template spectra have no overlap on the wavelength axis.
"""
# Seed np.random so that results are consistent
np.random.seed(42)
# Create test spectra
spec_axis = np.linspace(0, 50, 50) * u.AA
spec_axis_no_overlap = np.linspace(51, 102, 50) * u.AA
spec = Spectrum1D(spectral_axis=spec_axis,
flux=np.random.randn(50) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(50), unit='Jy'))
spec1 = Spectrum1D(spectral_axis=spec_axis_no_overlap,
flux=np.random.randn(50) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(50), unit='Jy'))
# Get result from template_match
tm_result = template_comparison.template_match(spec, spec1)
# Create new spectrum for comparison
spec_result = Spectrum1D(spectral_axis=spec_axis,
flux=spec1.flux * template_comparison._normalize_for_template_matching(spec, spec1))
# assert quantity_allclose(tm_result[0].flux, spec_result.flux, atol=0.01*u.Jy)
assert np.isnan(tm_result[3])
def test_template_match_list():
"""
Test template_match when template spectra are in a list.
"""
np.random.seed(42)
# Create test spectra
spec_axis1 = np.linspace(0, 50, 50) * u.AA
spec_axis2 = np.linspace(0, 50, 50) * u.AA
spec = Spectrum1D(spectral_axis=spec_axis1,
flux=np.random.randn(50) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(50),
unit='Jy'))
spec1 = Spectrum1D(spectral_axis=spec_axis2,
flux=np.random.randn(50) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(50),
unit='Jy'))
spec2 = Spectrum1D(spectral_axis=spec_axis2,
flux=np.random.randn(50) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(50),
unit='Jy'))
# Combine spectra into list
template_list = [spec1, spec2]
# Get result from template_match
tm_result = template_comparison.template_match(spec, template_list)
np.testing.assert_almost_equal(tm_result[3], 40093.28353756253)
# make sure that multiple template spectra will create a list of
# chi2 values, one per template.
assert len(tm_result) == 5
def test_template_known_redshift():
"""
Test template match when the redshift is known.
"""
# Seed np.random so that results are consistent
np.random.seed(42)
# Create test spectra
spec_axis = np.linspace(0, 50, 50) * u.AA
perm_flux = np.random.randn(50) * u.Jy
redshift = 3
# Observed spectrum
spec = Spectrum1D(spectral_axis=spec_axis * (1 + redshift),
flux=perm_flux,
uncertainty=StdDevUncertainty(np.random.sample(50),
unit='Jy'))
# Template spectrum
spec1 = Spectrum1D(spectral_axis=spec_axis,
flux=perm_flux,
uncertainty=StdDevUncertainty(np.random.sample(50),
unit='Jy'))
tm_result = template_comparison.template_match(
observed_spectrum=spec,
spectral_templates=spec1,
resample_method="flux_conserving",
redshift=redshift)
assert len(tm_result) == 5
np.testing.assert_almost_equal(tm_result[1], redshift)
np.testing.assert_almost_equal(tm_result[3], 1.9062409482056814e-31)
def test_template_match_spectrum_collection():
"""
Test template_match when template spectra are in a SpectrumCollection object.
"""
np.random.seed(42)
# Create test spectra
spec_axis1 = np.linspace(0, 50, 50) * u.AA
spec_axis2 = np.linspace(0, 50, 50) * u.AA
spec = Spectrum1D(spectral_axis=spec_axis1,
flux=np.random.randn(50) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(50)))
spec1 = Spectrum1D(spectral_axis=spec_axis2,
flux=np.random.randn(50) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(50)))
spec2 = Spectrum1D(spectral_axis=spec_axis2,
flux=np.random.randn(50) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(50)))
# Combine spectra into SpectrumCollection object
spec_coll = SpectrumCollection.from_spectra([spec1, spec2])
# Get result from template_match
tm_result = template_comparison.template_match(spec, spec_coll)
np.testing.assert_almost_equal(tm_result[3], 40093.28353756253)
def test_snr_multiple_flux(simulated_spectra):
"""
Test the simple version of the spectral SNR, with multiple flux per single dispersion.
"""
np.random.seed(42)
#
# Set up the data and add the uncertainty and calculate the expected SNR
#
uncertainty = StdDevUncertainty(0.1 * np.random.random((5, 10)) * u.mJy)
spec = Spectrum1D(spectral_axis=np.arange(10) * u.AA,
flux=np.random.sample((5, 10)) * u.Jy,
uncertainty=uncertainty)
snr_spec = snr(spec)
assert np.allclose(
np.array(snr_spec),
[18.20863867, 31.89475309, 14.51598119, 22.24603204, 32.01461421])
uncertainty = StdDevUncertainty(0.1 * np.random.random(10) * u.mJy)
spec = Spectrum1D(spectral_axis=np.arange(10) * u.AA,
flux=np.random.sample(10) * u.Jy,
uncertainty=uncertainty)
snr_spec = snr(spec)
assert np.allclose(np.array(snr_spec), 31.325265361800415)
def __init__(self,
data,
wcs=None,
parent=None,
layer_mask=None,
uncertainty=None,
unit=None,
mask=None,
*args,
**kwargs):
# if not issubclass(data.__class__, Spectrum1DRefLayer):
if uncertainty is None:
uncertainty = StdDevUncertainty(np.zeros(data.shape))
elif isinstance(uncertainty, UnknownUncertainty):
uncertainty = StdDevUncertainty(uncertainty.array)
if mask is None:
mask = np.zeros(data.shape).astype(bool)
super(Spectrum1DRefLayer, self).__init__(data,
wcs=wcs,
unit=unit,
uncertainty=uncertainty,
mask=mask,
*args,
**kwargs)
self._parent = parent
self._layer_mask = layer_mask
def test_ccd_process_gain_corrected():
# test the through ccd_process with gain_corrected as False
ccd_data = CCDData(10.0 * np.ones((100, 100)), unit=u.adu)
ccd_data.data[:, -10:] = 2
ccd_data.meta['testkw'] = 100
mask = np.zeros((100, 90))
masterbias = CCDData(4.0 * np.ones((100, 90)), unit=u.adu)
masterbias.uncertainty = StdDevUncertainty(np.zeros((100, 90)))
dark_frame = CCDData(0.0 * np.ones((100, 90)), unit=u.adu)
dark_frame.uncertainty = StdDevUncertainty(np.zeros((100, 90)))
masterflat = CCDData(5.0 * np.ones((100, 90)), unit=u.adu)
masterflat.uncertainty = StdDevUncertainty(np.zeros((100, 90)))
occd = ccd_process(ccd_data, oscan=ccd_data[:, -10:], trim='[1:90,1:100]',
error=True, master_bias=masterbias,
master_flat=masterflat, dark_frame=dark_frame,
bad_pixel_mask=mask, gain=0.5 * u.electron/u.adu,
readnoise=5**0.5 * u.electron, oscan_median=True,
dark_scale=False, dark_exposure=1.*u.s,
data_exposure=1.*u.s, gain_corrected=False)
# final results should be (10 - 2) / 2.0 - 2 = 2
# error should be (4 + 5)**0.5 / 0.5 = 3.0
np.testing.assert_array_equal(2.0 * np.ones((100, 90)), occd.data)
np.testing.assert_almost_equal(3.0 * np.ones((100, 90)),
occd.uncertainty.array)
np.testing.assert_array_equal(mask, occd.mask)
assert(occd.unit == u.electron)
# Make sure the original keyword is still present. Regression test for #401
assert occd.meta['testkw'] == 100
def test_stddevuncertainty_compat_descriptor_no_weakref():
# TODO: Remove this test if astropy 1.0 isn't supported anymore
# This test might create a Memoryleak on purpose, so the last lines after
# the assert are IMPORTANT cleanup.
ccd = CCDData(np.ones((10, 10)), unit='')
uncert = StdDevUncertainty(np.ones((10, 10)))
uncert._parent_nddata = ccd
assert uncert.parent_nddata is ccd
uncert._parent_nddata = None
def test_correlation():
"""
Test correlation when both observed and template spectra have the same wavelength axis
"""
size = 4000
# Seed np.random so that results are consistent
np.random.seed(51)
# Create test spectra
spec_axis = np.linspace(4001., 8000., num=size) * u.AA
# Two narrow Gaussians are offset from each other so
# as to generate a correlation peak at a expected lag.
f1 = np.random.randn(size) * 0.5 * u.Jy
f2 = np.random.randn(size) * 0.5 * u.Jy
expected_lag = 10000. * u.km / u.s
mean2 = 5000. * u.AA
stdev2 = 30. * u.AA
mean1 = (1 + expected_lag / const.c.to('km/s')) * mean2
stdev1 = (1 + expected_lag / const.c.to('km/s')) * stdev2
rest_value = mean2
g1 = models.Gaussian1D(amplitude=30 * u.Jy, mean=mean1, stddev=stdev1)
g2 = models.Gaussian1D(amplitude=30 * u.Jy, mean=mean2, stddev=stdev2)
flux1 = f1 + g1(spec_axis)
flux2 = f2 + g2(spec_axis)
# Observed spectrum must have a rest wavelength value set in.
spec1 = Spectrum1D(spectral_axis=spec_axis,
flux=flux1,
uncertainty=StdDevUncertainty(np.random.sample(size),
unit='Jy'),
velocity_convention='optical',
rest_value=rest_value)
spec2 = Spectrum1D(spectral_axis=spec_axis,
flux=flux2,
uncertainty=StdDevUncertainty(np.random.sample(size),
unit='Jy'))
# Get result from correlation
corr, lag = correlation.template_correlate(spec1, spec2)
# Check units
assert corr.unit == u.dimensionless_unscaled
assert lag.unit == u.km / u.s
# Check position of correlation peak.
maximum = np.argmax(corr)
v_fit = _fit_peak(corr, lag, maximum)
# checks against 1.5 * 10**(-decimal)
np.testing.assert_almost_equal(v_fit.value, expected_lag.value, -1)
def test_is_continuum_below_threshold():
# No mask, no uncertainty
wavelengths = [300, 500, 1000] * u.nm
data = [0.001, -0.003, 0.003] * u.Jy
spectrum = Spectrum1D(spectral_axis=wavelengths, flux=data)
assert is_continuum_below_threshold(spectrum,
threshold=0.1 * u.Jy) == True # noqa
# # No mask, no uncertainty, threshold is float
# wavelengths = [300, 500, 1000] * u.nm
# data = [0.0081, 0.0043, 0.0072] * u.Jy
# spectrum = Spectrum1D(spectral_axis=wavelengths, flux=data)
# assert True == is_continuum_below_threshold(spectrum, threshold=0.1)
# No mask, with uncertainty
wavelengths = [300, 500, 1000] * u.nm
data = [0.03, 0.029, 0.031] * u.Jy
uncertainty = StdDevUncertainty([1.01, 1.03, 1.01] * u.Jy)
spectrum = Spectrum1D(spectral_axis=wavelengths,
flux=data,
uncertainty=uncertainty)
assert is_continuum_below_threshold(spectrum,
threshold=0.1 * u.Jy) == True # noqa
# With mask, with uncertainty
wavelengths = [300, 500, 1000] * u.nm
data = [0.01, 1.029, 0.013] * u.Jy
mask = np.array([False, True, False])
uncertainty = StdDevUncertainty([1.01, 1.13, 1.1] * u.Jy)
spectrum = Spectrum1D(spectral_axis=wavelengths,
flux=data,
uncertainty=uncertainty,
mask=mask)
assert is_continuum_below_threshold(spectrum,
threshold=0.1 * u.Jy) == True # noqa
# With mask, with uncertainty -- should throw an exception
wavelengths = [300, 500, 1000] * u.nm
data = [10.03, 10.029, 10.033] * u.Jy
mask = np.array([False, False, False])
uncertainty = StdDevUncertainty([1.01, 1.13, 1.1] * u.Jy)
spectrum = Spectrum1D(spectral_axis=wavelengths,
flux=data,
uncertainty=uncertainty,
mask=mask)
with pytest.warns(
AstropyUserWarning,
match=r'.*If you want to suppress this warning.*') as e_info:
find_lines_threshold(spectrum, noise_factor=1)
assert len(e_info) == 1
def test_template_redshift_with_multiple_template_spectra_in_match():
# Seed np.random so that results are consistent
np.random.seed(42)
# Create test spectra
spec_axis = np.linspace(0, 50, 50) * u.AA
perm_flux = np.random.randn(50) * u.Jy
# Test redshift
redshift = 3
# Observed spectrum
spec = Spectrum1D(spectral_axis=spec_axis * (1 + redshift),
flux=perm_flux,
uncertainty=StdDevUncertainty(np.random.sample(50),
unit='Jy'))
# Template spectrum
spec1 = Spectrum1D(spectral_axis=spec_axis,
flux=np.random.randn(50) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(50)))
spec2 = Spectrum1D(spectral_axis=spec_axis,
flux=np.random.randn(50) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(50)))
# Combine spectra into SpectrumCollection object
spec_coll = SpectrumCollection.from_spectra([spec1, spec2])
# Test redshift parameters
min_redshift = .5
max_redshift = 5.5
delta_redshift = .25
redshift_trial_values = np.arange(min_redshift,
max_redshift + delta_redshift,
delta_redshift)
tm_result = template_comparison.template_match(
observed_spectrum=spec,
spectral_templates=spec_coll,
resample_method="flux_conserving",
redshift=redshift_trial_values)
assert len(tm_result) == 5
# TODO: Determine cause of and fix failing assert
# np.testing.assert_almost_equal(tm_result[1], redshift)
np.testing.assert_almost_equal(tm_result[3], 6803.922741644725)
# When a spectrum collection is matched with a redshift
# grid, a list-of-lists is returned with the trial chi2
# values computed for every combination redshift-template.
# The external list spans the templates in the collection,
# while each internal list contains all chi2 values
# for a given template.
assert len(tm_result[4]) == 2
def test_create_with_uncertainty():
spec = Spectrum1D(spectral_axis=np.arange(1, 50) * u.nm,
flux=np.random.sample(49) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(49) * 0.1))
assert isinstance(spec.uncertainty, StdDevUncertainty)
spec = Spectrum1D(spectral_axis=np.arange(1, 50) * u.nm,
flux=np.random.sample(49) * u.Jy,
uncertainty=StdDevUncertainty(np.random.sample(49) * 0.1))
assert spec.flux.unit == spec.uncertainty.unit
def test_tabular_fits_writer(tmpdir, spectral_axis):
wlu = {
'wavelength': u.AA,
'frequency': u.GHz,
'energy': u.eV,
'wavenumber': u.cm**-1
}
# Create a small data set
disp = np.arange(1, 1.1, 0.01) * wlu[spectral_axis]
flux = np.ones(len(disp)) * 1.e-14 * u.Jy
unc = StdDevUncertainty(0.01 * flux)
if spectral_axis not in ('wavelength', ):
disp = np.flip(disp)
spectrum = Spectrum1D(flux=flux, spectral_axis=disp, uncertainty=unc)
tmpfile = str(tmpdir.join('_tst.fits'))
spectrum.write(tmpfile, format='tabular-fits')
# Read it in and check against the original
spec = Spectrum1D.read(tmpfile)
assert spec.flux.unit == spectrum.flux.unit
assert spec.spectral_axis.unit == spectrum.spectral_axis.unit
assert quantity_allclose(spec.spectral_axis, spectrum.spectral_axis)
assert quantity_allclose(spec.flux, spectrum.flux)
assert quantity_allclose(spec.uncertainty.quantity,
spectrum.uncertainty.quantity)
# Test spectrum with different flux unit
flux = np.random.normal(0., 1.e-9,
disp.shape[0]) * u.W * u.m**-2 * u.AA**-1
unc = StdDevUncertainty(0.1 * np.sqrt(np.abs(flux.value)) * flux.unit)
spectrum = Spectrum1D(flux=flux, spectral_axis=disp, uncertainty=unc)
# Try to overwrite the file
with pytest.raises(OSError, match=r'File exists:'):
spectrum.write(tmpfile, format='tabular-fits')
spectrum.write(tmpfile, format='tabular-fits', overwrite=True)
cmap = {
spectral_axis: ('spectral_axis', wlu[spectral_axis]),
'flux': ('flux', 'erg / (s cm**2 AA)'),
'uncertainty': ('uncertainty', None)
}
# Read it back again and check against the original
spec = Spectrum1D.read(tmpfile, format='tabular-fits', column_mapping=cmap)
assert spec.flux.unit == u.Unit('erg / (s cm**2 AA)')
assert spec.spectral_axis.unit == spectrum.spectral_axis.unit
assert quantity_allclose(spec.spectral_axis, spectrum.spectral_axis)
assert quantity_allclose(spec.flux, spectrum.flux)
assert quantity_allclose(spec.uncertainty.quantity,
spectrum.uncertainty.quantity)
def fromMarzJSON(json_spectrum):
from astropy.nddata import (
VarianceUncertainty,
StdDevUncertainty,
InverseVariance,
)
wavelength = Quantity(json_spectrum["wavelength"], u.Angstrom)
spectrum = SpectrumList()
uncertainty = None
if "variance" in json_spectrum.keys():
uncertainty = StdDevUncertainty(
Quantity(np.array(json_spectrum["variance"], dtype=np.float)))
for key in json_spectrum.keys():
if key != 'wavelength' and key != 'variance':
flux = Quantity(np.array(json_spectrum[key], dtype=np.float))
if key == 'intensity' and uncertainty:
aspectrum = Spectrum1D(flux=flux,
spectral_axis=wavelength,
uncertainty=uncertainty,
mask=np.isnan(flux),
meta={'purpose': 'reduced'})
else:
aspectrum = Spectrum1D(flux=flux,
spectral_axis=wavelength,
mask=np.isnan(flux),
meta={'purpose': key})
spectrum.append(aspectrum)
return spectrum
def test_line_flux_masked():
np.random.seed(42)
N = 100
wavelengths = np.linspace(0.4, 1.05, N) * u.um
g = models.Gaussian1D(amplitude=2000 * u.mJy,
mean=0.56 * u.um,
stddev=0.01 * u.um)
flux = g(wavelengths) + 1000 * u.mJy
noise = 400 * np.random.random(flux.shape) * u.mJy
flux += noise
spectrum = Spectrum1D(spectral_axis=wavelengths, flux=flux)
spectrum.uncertainty = StdDevUncertainty(noise)
spectrum_masked = snr_threshold(spectrum, 10.)
# Ensure we have at least 50% of the data being masked.
assert len(np.where(spectrum_masked.mask)[0]) > N / 2
result = line_flux(spectrum_masked)
assert result.unit.is_equivalent(u.Jy * u.um)
assert quantity_allclose(result.value, 720.52992, atol=0.001)
# With flux conserving resampler
result = line_flux(spectrum_masked,
mask_interpolation=FluxConservingResampler)
assert quantity_allclose(result.value, 720.61116, atol=0.001)
def snr_spec(flux, wl, n):
sample = len(wl)
noise = n * np.asarray(random.sample(range(0, len(wl)), sample)) / len(wl)
unc = StdDevUncertainty(noise)
fluxn = [[] for i in range(len(wl))]
i = 0
for inc in unc:
fluxn[i] = flux[i] + noise[i]
i = i + 1
spec1d = Spectrum1D(spectral_axis=wl * u.AA,
flux=fluxn * u.Jy,
uncertainty=unc)
#ax = plt.subplots()[1]
#ax.plot(spec1d.spectral_axis, spec1d.flux)
#ax.set_xlim([3520,3550])
sn1 = snr(spec1d, SpectralRegion(3070 * u.AA, 3090 * u.AA))
sn = snr_derived(spec1d, SpectralRegion(3070 * u.AA, 3090 * u.AA))
#print('SNR1: '+ str(snr(spec1d)), SpectralRegion(3500*u.AA, 3550*u.AA))
print('SNR: ' + str(sn1))
#print('SNR: '+ str(sn))
#print('FWHM:'+str(fwhm(spec1d)))
#0.042 = snr 50
#
try:
return fluxn
except:
raise Exception('Check S/N function')
def test_centroid_masked(simulated_spectra):
"""
Test centroid with masked spectrum.
"""
np.random.seed(42)
# Same as in test for unmasked spectrum, but using
# masked version of same spectrum.
spectrum = simulated_spectra.s1_um_mJy_e1_masked
uncertainty = StdDevUncertainty(
0.1 * np.random.random(len(spectrum.flux)) * u.mJy)
spectrum.uncertainty = uncertainty
# Use masked flux and dispersion arrays to compute
# the expected value for centroid.
wavelengths = spectrum.spectral_axis[~spectrum.mask]
flux = spectrum.flux[~spectrum.mask]
spec_centroid_expected = np.sum(flux * wavelengths) / np.sum(flux)
spec_centroid = centroid(spectrum, None)
assert isinstance(spec_centroid, u.Quantity)
assert np.allclose(spec_centroid.value, spec_centroid_expected.value)
def test_block_replicate():
ccd = CCDData(np.ones((4, 4)),
unit='adu',
meta={'testkw': 1},
mask=np.zeros((4, 4), dtype=bool),
uncertainty=StdDevUncertainty(np.ones((4, 4))),
wcs=np.zeros((4, 4)))
with catch_warnings(AstropyUserWarning) as w:
ccd_repl = block_replicate(ccd, (2, 2))
assert len(w) == 1
assert 'following attributes were set' in str(w[0].message)
assert isinstance(ccd_repl, CCDData)
assert np.all(ccd_repl.data == 0.25)
assert ccd_repl.data.shape == (8, 8)
assert ccd_repl.unit == u.adu
# Other attributes are set to None. In case the function is modified to
# work on these attributes correctly those tests need to be updated!
assert ccd_repl.meta == {'testkw': 1}
assert ccd_repl.mask is None
assert ccd_repl.wcs is None
assert ccd_repl.uncertainty is None
# Make sure meta is copied
ccd_repl.meta['testkw2'] = 10
assert 'testkw2' not in ccd.meta
def muscles_sed(file_obj, **kwargs):
"""
Load spectrum from a MUSCLES Treasury Survey panchromatic SED FITS file.
Parameters
----------
file_obj: str, file-like, or HDUList
FITS file name, object (provided from name by Astropy I/O Registry),
or HDUList (as resulting from astropy.io.fits.open()).
Returns
-------
data: Spectrum1D
The spectrum that is represented by the data in this table.
"""
# name is not used; what was it for?
# name = os.path.basename(file_name.rstrip(os.sep)).rsplit('.', 1)[0]
with read_fileobj_or_hdulist(file_obj, **kwargs) as hdulist:
header = hdulist[0].header
tab = Table.read(hdulist[1])
meta = {'header': header}
uncertainty = StdDevUncertainty(tab["ERROR"])
data = Quantity(tab["FLUX"])
wavelength = Quantity(tab["WAVELENGTH"])
return Spectrum1D(flux=data, spectral_axis=wavelength,
uncertainty=uncertainty, meta=meta)
def cos_spectrum_loader(file_name, **kwargs):
"""
Load COS spectral data from the MAST archive into a spectrum object.
Parameters
----------
file_name: str
The path to the FITS file
Returns
-------
data: Spectrum1D
The spectrum that is represented by the data in this table.
"""
with fits.open(file_name, **kwargs) as hdu:
header = hdu[0].header
name = header.get('FILENAME')
meta = {'header': header}
unit = Unit("erg/cm**2 Angstrom s")
disp_unit = Unit('Angstrom')
data = hdu[1].data['FLUX'].flatten() * unit
dispersion = hdu[1].data['wavelength'].flatten() * disp_unit
uncertainty = StdDevUncertainty(hdu[1].data["ERROR"].flatten() * unit)
sort_idx = dispersion.argsort()
dispersion = dispersion[sort_idx]
data = data[sort_idx]
uncertainty = uncertainty[sort_idx]
return Spectrum1D(flux=data,
spectral_axis=dispersion,
uncertainty=uncertainty,
meta=meta)
def test_snr(simulated_spectra):
"""
Test the simple version of the spectral SNR.
"""
np.random.seed(42)
#
# Set up the data and add the uncertainty and calculate the expected SNR
#
spectrum = simulated_spectra.s1_um_mJy_e1
uncertainty = StdDevUncertainty(
0.1 * np.random.random(len(spectrum.flux)) * u.mJy)
spectrum.uncertainty = uncertainty
wavelengths = spectrum.spectral_axis
flux = spectrum.flux
spec_snr_expected = np.mean(flux / (uncertainty.array * uncertainty.unit))
#
# SNR of the whole spectrum
#
spec_snr = snr(spectrum)
assert isinstance(spec_snr, u.Quantity)
assert np.allclose(spec_snr.value, spec_snr_expected.value)
def test_centroid(simulated_spectra):
"""
Test the simple version of the spectral centroid.
"""
np.random.seed(42)
#
# Set up the data and add the uncertainty and calculate the expected SNR
#
spectrum = simulated_spectra.s1_um_mJy_e1
uncertainty = StdDevUncertainty(
0.1 * np.random.random(len(spectrum.flux)) * u.mJy)
spectrum.uncertainty = uncertainty
wavelengths = spectrum.spectral_axis
flux = spectrum.flux
spec_centroid_expected = np.sum(flux * wavelengths) / np.sum(flux)
#
# SNR of the whole spectrum
#
spec_centroid = centroid(spectrum, None)
assert isinstance(spec_centroid, u.Quantity)
assert np.allclose(spec_centroid.value, spec_centroid_expected.value)
def cos_spectrum_loader(file_name, **kwargs):
""" Load file from COS spectral data into a spectrum object
Parameters
----------
file_name: str
The path to the FITS file
Returns
-------
data: Spectrum1DRef
The data.
"""
name = os.path.basename(file_name)
with fits.open(file_name, **kwargs) as hdu:
header = hdu[0].header
meta = {'header': header}
uncertainty = StdDevUncertainty(hdu[1].data["ERROR"].flatten())
data = hdu[1].data['FLUX'].flatten()
dispersion = hdu[1].data['wavelength'].flatten()
unit = Unit("erg/cm**2 Angstrom s")
return Spectrum1DRef.from_array(data=data,
dispersion=dispersion,
dispersion_unit=Unit('Angstrom'),
uncertainty=uncertainty,
unit=unit,
meta=meta)
def test_region_simple(simulated_spectra):
np.random.seed(42)
spectrum = simulated_spectra.s1_um_mJy_e1
uncertainty = StdDevUncertainty(
0.1 * np.random.random(len(spectrum.flux)) * u.mJy)
spectrum.uncertainty = uncertainty
region = SpectralRegion(0.6 * u.um, 0.8 * u.um)
sub_spectrum = extract_region(spectrum, region)
sub_spectrum_flux_expected = np.array([
1605.71612173, 1651.41650744, 2057.65798618, 2066.73502361,
1955.75832537, 1670.52711471, 1491.10034446, 1637.08084112,
1471.28982259, 1299.19484483, 1423.11195734, 1226.74494917,
1572.31888312, 1311.50503403, 1474.05051673, 1335.39944397,
1420.61880528, 1433.18623759, 1290.26966668, 1605.67341284,
1528.52281708, 1592.74392861, 1568.74162534, 1435.29407808,
1536.68040935, 1157.33825995, 1136.12679394, 999.92394692,
1038.61546167, 1011.60297294
])
assert quantity_allclose(sub_spectrum.flux.value,
sub_spectrum_flux_expected)
def test_snr_single_region(simulated_spectra):
"""
Test the simple version of the spectral SNR over a region of the spectrum.
"""
np.random.seed(42)
region = SpectralRegion(0.52 * u.um, 0.59 * u.um)
#
# Set up the data
#
spectrum = simulated_spectra.s1_um_mJy_e1
uncertainty = StdDevUncertainty(
0.1 * np.random.random(len(spectrum.flux)) * u.mJy)
spectrum.uncertainty = uncertainty
wavelengths = spectrum.spectral_axis
flux = spectrum.flux
# +1 because we want to include it in the calculation
l = np.nonzero(wavelengths > region.lower)[0][0]
r = np.nonzero(wavelengths < region.upper)[0][-1] + 1
spec_snr_expected = np.mean(flux[l:r] /
(uncertainty.array[l:r] * uncertainty.unit))
#
# SNR of the whole spectrum
#
spec_snr = snr(spectrum, region)
assert np.allclose(spec_snr.value, spec_snr_expected.value)
def test_collection_without_optional_arguments():
# Without uncertainties
flux = u.Quantity(np.random.sample((5, 10)), unit='Jy')
spectral_axis = u.Quantity(np.arange(50).reshape((5, 10)) + 1, unit='AA')
uncertainty = StdDevUncertainty(np.random.sample((5, 10)), unit='Jy')
wcs = np.array([gwcs_from_array(x) for x in spectral_axis])
mask = np.ones((5, 10)).astype(bool)
meta = [{'test': 5, 'info': [1, 2, 3]} for i in range(5)]
spec_coll = SpectrumCollection(flux=flux,
spectral_axis=spectral_axis,
wcs=wcs,
mask=mask,
meta=meta)
# Without mask
spec_coll = SpectrumCollection(flux=flux,
spectral_axis=spectral_axis,
uncertainty=uncertainty,
wcs=wcs,
meta=meta)
# Without meta
spec_coll = SpectrumCollection( # noqa
flux=flux,
spectral_axis=spectral_axis,
uncertainty=uncertainty,
wcs=wcs,
mask=mask)
def test_tabular_fits_2d(tmpdir, spectral_axis):
wlu = {
'wavelength': u.AA,
'frequency': u.GHz,
'energy': u.eV,
'wavenumber': u.cm**-1
}
# Create a small data set with 2D flux + uncertainty
disp = np.arange(1, 1.1, 0.01) * wlu[spectral_axis]
flux = np.ones(
(3, len(disp))) * np.arange(1,
len(disp) + 1)**2 * 1.e-14 * u.Jy
unc = StdDevUncertainty(0.01 * np.random.rand(3, len(disp)))
if spectral_axis not in ('wavelength', ):
disp = np.flip(disp)
spectrum = Spectrum1D(flux=flux, spectral_axis=disp, uncertainty=unc)
tmpfile = str(tmpdir.join('_tst.fits'))
spectrum.write(tmpfile, format='tabular-fits')
# Read it in and check against the original
spec = Spectrum1D.read(tmpfile)
assert spec.flux.unit == spectrum.flux.unit
assert spec.spectral_axis.unit == spectrum.spectral_axis.unit
assert spec.flux.shape == flux.shape
assert spec.uncertainty.array.shape == flux.shape
assert quantity_allclose(spec.spectral_axis, spectrum.spectral_axis)
assert quantity_allclose(spec.flux, spectrum.flux)
assert quantity_allclose(spec.uncertainty.quantity,
spectrum.uncertainty.quantity)
def test_arithmetic_overload_ccddata_operand(ccd_data):
ccd_data.uncertainty = StdDevUncertainty(np.ones_like(ccd_data))
operand = ccd_data.copy()
result = ccd_data.add(operand)
assert len(result.meta) == 0
np.testing.assert_array_equal(result.data, 2 * ccd_data.data)
np.testing.assert_array_equal(result.uncertainty.array,
np.sqrt(2) * ccd_data.uncertainty.array)
result = ccd_data.subtract(operand)
assert len(result.meta) == 0
np.testing.assert_array_equal(result.data, 0 * ccd_data.data)
np.testing.assert_array_equal(result.uncertainty.array,
np.sqrt(2) * ccd_data.uncertainty.array)
result = ccd_data.multiply(operand)
assert len(result.meta) == 0
np.testing.assert_array_equal(result.data, ccd_data.data**2)
expected_uncertainty = (np.sqrt(2) * np.abs(ccd_data.data) *
ccd_data.uncertainty.array)
np.testing.assert_allclose(result.uncertainty.array, expected_uncertainty)
result = ccd_data.divide(operand)
assert len(result.meta) == 0
np.testing.assert_array_equal(result.data, np.ones_like(ccd_data.data))
expected_uncertainty = (np.sqrt(2) / np.abs(ccd_data.data) *
ccd_data.uncertainty.array)
np.testing.assert_allclose(result.uncertainty.array, expected_uncertainty)
def new_variance_uncertainty_instance(array):
if array is None:
return
with warnings.catch_warnings():
if (array < 0.).any():
warnings.warn("Negative variance values found. Setting to zero.",
RuntimeWarning)
array = np.where(array >= 0., array, 0.)
warnings.simplefilter("ignore", RuntimeWarning)
obj = StdDevUncertainty(np.sqrt(array))
warnings.simplefilter("default", RuntimeWarning)
cls = obj.__class__
obj.__class__ = cls.__class__(cls.__name__ + "WithAsVariance", (cls, StdDevAsVariance), {})
return obj
