def library_difference(params,
prop,
suffix='_sm',
ptlabels=False,
rescale=True,
plt_kw={'color': 'blue'}):
"""Plot the residuals (library-derived) for each star in the library.
Args:
params (pd.DataFrame): Parameter table with library and derived values
prop (str): Parameter to plot on the x-axis.
suffix (str): Suffix on columns in parameter table with derived values
"""
if prop == 'radius':
resid = (params[prop + suffix] - params[prop]) / params[prop]
plt.semilogx(params[prop], resid, 'o', **plt_kw)
plt.xlim(0.1, 20)
elif prop == 'mass':
resid = (params[prop + suffix] - params[prop]) / params[prop]
plt.plot(params[prop], resid, 'o', **plt_kw)
else:
resid = params[prop + suffix] - params[prop]
plt.plot(params[prop], resid, 'o', **plt_kw)
if ptlabels is not False and ptlabels in params.columns:
params['resid'] = resid
params.apply(lambda row: plots.annotate_point(row[prop], row['resid'],
row[ptlabels]),
axis=1)
mean = np.mean(resid)
rms = np.sqrt(np.mean(resid**2))
ax = plt.gca()
bbox = dict(facecolor='white', edgecolor='none', alpha=0.8)
plt.text(0.05,
0.1,
"Mean Diff: {0:.3g}\nRMS Diff: {1:.3g}".format(mean, rms),
transform=ax.transAxes,
bbox=bbox)
plt.axhline(y=0, color='k', linestyle='dashed')
plots.label_axes(param_x=prop, rescale=rescale)
def plot(self,
paramx,
paramy,
grouped=False,
marker='.',
ptlabels=False,
plt_kw={}):
"""Create a plot of the library in parameter space
Args:
paramx (str): Parameter to plot on the x-axis
paramy (str): Parameter to plot on the y-axis
grouped (bool, optional): Whether to group the stars by source
ptlabels (str, optional): Library column to annotate points with
plt_kw (dict, optional): Additional keyword arguments to pass to
pyplot.plot
"""
source_labels = {
'Gaidos': 'K-dwarfs',
'Von Braun': 'Interferometric',
'Brewer': 'Brewer (2016)',
'Mann': 'Mann (2015)',
'Bruntt': 'Bruntt (2012)'
}
if grouped:
g = self.library_params.groupby('source')
for source in g.groups:
cut = self.library_params.ix[g.groups[source]]
plt.plot(cut[paramx],
cut[paramy],
marker,
label=source_labels[source])
else:
plt.plot(self.library_params[paramx], self.library_params[paramy],
marker, **plt_kw)
if ptlabels is not False:
self.library_params.apply(lambda x: plots.annotate_point(
x[paramx], x[paramy], x[ptlabels]),
axis=1)
plots.label_axes(paramx, paramy)
def library_comparison(params,
param_x,
param_y,
suffix='_sm',
ptlabels=False,
legend=True,
rescale=True):
"""Create a library comparison plot showing the library values of the
parameters as points, with lines point to derived parameters.
Args:
params (pd.DataFrame): Parameter table with library and derived values
param_x (str): Parameter to plot on the x-axis.
param_y (str): Parameter to plot on the y-axis.
suffix (str): Suffix on columns in parameter table with derived values
ptlabels (str, optional): Column to annotate points with.
Defaults to False, denoting no labels.
"""
if param_x not in Library.STAR_PROPS:
raise ValueError("{0} is not a valid parameter.".format(param_x))
if param_y not in Library.STAR_PROPS:
raise ValueError("{0} is not a valid parameter.".format(param_y))
plt.plot(params[param_x], params[param_y], 'ko', label='Library value')
x = params[[param_x + suffix, param_x]]
y = params[[param_y + suffix, param_y]]
plt.plot(x.T, y.T, 'r')
plt.plot(x.iloc[0], y.iloc[0], 'r', label='SpecMatch-Emp value')
plots.label_axes(param_x, param_y, rescale)
if legend:
plt.legend(loc='best')
if ptlabels is not False and ptlabels in params.columns:
params.apply(lambda row: plots.annotate_point(row[param_x], row[
param_y], row[ptlabels]),
axis=1)
def plot_chi_squared_surface(self, region=0, num_best=None):
"""Plot the chi-squared surface from the pairwise matching procedure.
Creates a three-column plot of the best chi-squared obtained with
each library spectrum as a function of the library parameters.
Args:
region (int or tuple): The wavelength region to plot.
If an int is given, refers to the index of the region in
`self.regions`.
If a tuple is given, should be present in `self.regions`.
num_best (optional [int]): Number of best spectra to highlight.
(default: `self.num_best`)
"""
if self.match_results.empty:
return
if isinstance(region, int):
region = self.regions[region]
elif isinstance(region, tuple) and region not in self.regions:
raise ValueError("Region {0} was not a region.".format(region))
if len(self.regions) == 1:
cs_col = 'chi_squared'
else:
cs_col = 'chi_squared_{0:.0f}'.format(region[0])
self.match_results.sort_values(by=cs_col, inplace=True)
if num_best is None:
num_best = self.num_best
ax1 = plt.subplot(131)
plt.semilogy()
plt.plot(self.match_results['Teff'], self.match_results[cs_col], '.',
color='blue')
plt.plot(self.match_results['Teff'].head(num_best),
self.match_results[cs_col].head(num_best), 'r.')
plt.ylabel(r'$\chi^2$')
plt.xlabel(r'$T_{eff}$ (K)')
plots.label_axes(param_x='Teff')
plt.subplot(132, sharey=ax1)
plt.semilogy()
plt.plot(self.match_results['radius'], self.match_results[cs_col], '.',
color='blue')
plt.plot(self.match_results['radius'].head(num_best),
self.match_results[cs_col].head(num_best), 'r.')
plots.label_axes(param_x='radius')
plt.subplot(133, sharey=ax1)
plt.semilogy()
plt.plot(self.match_results['feh'], self.match_results[cs_col], '.',
color='blue')
plt.plot(self.match_results['feh'].head(num_best),
self.match_results[cs_col].head(num_best), 'r.')
plots.label_axes(param_x='feh')
def plot_references(self, region=0, num_best=None, verbose=True):
"""Plots the location of the best references used in the linear
combination step.
Args:
region (int or tuple): The region used in matching.
If an int is given, refers to the index of the region in
`self.regions`.
If a tuple is given, should be present in `self.regions`.
num_best (optional [int]): Number of best spectra to highlight.
(default: `self.num_best`)
verbose (optional [bool]): Whether to annotate the points with
the lincomb coefficients
"""
if self.match_results is None:
return
# get region
if isinstance(region, int):
region_num = region
region = self.regions[region_num]
elif isinstance(region, tuple):
if region in self.regions:
region_num = self.regions.index(region)
else:
# Raise exception of region was not found.
raise ValueError("Region {0} was not a region.".format(region))
if len(self.regions) == 1:
cs_col = 'chi_squared'
else:
cs_col = 'chi_squared_{0:.0f}'.format(region[0])
self.match_results.sort_values(by=cs_col, inplace=True)
if num_best is None:
num_best = self.num_best
def _plot_ref_params(paramx, paramy, zoom=False):
plt.plot(self.match_results[paramx], self.match_results[paramy],
'b.', alpha=0.6, label='_nolegend_')
plt.plot(self.match_results.head(num_best)[paramx],
self.match_results.head(num_best)[paramy], '^',
color='forestgreen', label='Best Matches')
if zoom:
plots.set_tight_lims(self.match_results.head(num_best)[paramx],
self.match_results.head(num_best)[paramy])
if verbose and self.coeffs is not None:
for i in range(self.num_best):
p = self.match_results.iloc[i]
plots.annotate_point(p[paramx], p[paramy],
'{0:.3f}'.format(
self.coeffs[region_num][i]))
gs = gridspec.GridSpec(2, 2)
plt.subplot(gs[0])
_plot_ref_params('Teff', 'radius')
if len(self.results) > 0:
plt.plot(self.lincomb_results[region_num]['Teff'],
self.lincomb_results[region_num]['radius'], 's',
color='purple', label='Derived Parameters')
plots.label_axes('Teff', 'radius')
plt.subplot(gs[1])
_plot_ref_params('Teff', 'radius', zoom=True)
if len(self.results) > 0:
plt.plot(self.lincomb_results[region_num]['Teff'],
self.lincomb_results[region_num]['radius'], 's',
color='purple', label='Derived Parameters')
plots.label_axes('Teff', 'radius', rescale=False)
plt.subplot(gs[2])
_plot_ref_params('feh', 'radius')
if len(self.results) > 0:
plt.plot(self.lincomb_results[region_num]['feh'],
self.lincomb_results[region_num]['radius'], 's',
color='purple', label='Derived Parameters')
plots.label_axes('feh', 'radius')
plt.subplot(gs[3])
_plot_ref_params('feh', 'radius', zoom=True)
if len(self.results) > 0:
plt.plot(self.lincomb_results[region_num]['feh'],
self.lincomb_results[region_num]['radius'], 's',
color='purple', label='Derived Parameters')
plots.label_axes('feh', 'radius', rescale=False)
def plot_chi_squared_surface(self, region=0, num_best=None):
"""Plot the chi-squared surface from the pairwise matching procedure.
Creates a three-column plot of the best chi-squared obtained with
each library spectrum as a function of the library parameters.
Args:
region (int or tuple): The wavelength region to plot.
If an int is given, refers to the index of the region in
`self.regions`.
If a tuple is given, should be present in `self.regions`.
num_best (optional [int]): Number of best spectra to highlight.
(default: `self.num_best`)
"""
if self.match_results.empty:
return
if isinstance(region, int):
region = self.regions[region]
elif isinstance(region, tuple) and region not in self.regions:
raise ValueError("Region {0} was not a region.".format(region))
if len(self.regions) == 1:
cs_col = 'chi_squared'
else:
cs_col = 'chi_squared_{0:.0f}'.format(region[0])
self.match_results.sort_values(by=cs_col, inplace=True)
if num_best is None:
num_best = self.num_best
ax1 = plt.subplot(131)
plt.semilogy()
plt.plot(self.match_results['Teff'],
self.match_results[cs_col],
'.',
color='blue')
plt.plot(self.match_results['Teff'].head(num_best),
self.match_results[cs_col].head(num_best), 'r.')
plt.ylabel(r'$\chi^2$')
plt.xlabel(r'$T_{eff}$ (K)')
plots.label_axes(param_x='Teff')
plt.subplot(132, sharey=ax1)
plt.semilogy()
plt.plot(self.match_results['radius'],
self.match_results[cs_col],
'.',
color='blue')
plt.plot(self.match_results['radius'].head(num_best),
self.match_results[cs_col].head(num_best), 'r.')
plots.label_axes(param_x='radius')
plt.subplot(133, sharey=ax1)
plt.semilogy()
plt.plot(self.match_results['feh'],
self.match_results[cs_col],
'.',
color='blue')
plt.plot(self.match_results['feh'].head(num_best),
self.match_results[cs_col].head(num_best), 'r.')
plots.label_axes(param_x='feh')
def plot_references(self, region=0, num_best=None, verbose=True):
"""Plots the location of the best references used in the linear
combination step.
Args:
region (int or tuple): The region used in matching.
If an int is given, refers to the index of the region in
`self.regions`.
If a tuple is given, should be present in `self.regions`.
num_best (optional [int]): Number of best spectra to highlight.
(default: `self.num_best`)
verbose (optional [bool]): Whether to annotate the points with
the lincomb coefficients
"""
if self.match_results is None:
return
# get region
if isinstance(region, int):
region_num = region
region = self.regions[region_num]
elif isinstance(region, tuple):
if region in self.regions:
region_num = self.regions.index(region)
else:
# Raise exception of region was not found.
raise ValueError("Region {0} was not a region.".format(region))
if len(self.regions) == 1:
cs_col = 'chi_squared'
else:
cs_col = 'chi_squared_{0:.0f}'.format(region[0])
self.match_results.sort_values(by=cs_col, inplace=True)
if num_best is None:
num_best = self.num_best
def _plot_ref_params(paramx, paramy, zoom=False):
plt.plot(self.match_results[paramx],
self.match_results[paramy],
'b.',
alpha=0.6,
label='_nolegend_')
plt.plot(self.match_results.head(num_best)[paramx],
self.match_results.head(num_best)[paramy],
'^',
color='forestgreen',
label='Best Matches')
if zoom:
plots.set_tight_lims(
self.match_results.head(num_best)[paramx],
self.match_results.head(num_best)[paramy])
if verbose and self.coeffs is not None:
for i in range(self.num_best):
p = self.match_results.iloc[i]
plots.annotate_point(
p[paramx], p[paramy],
'{0:.3f}'.format(self.coeffs[region_num][i]))
gs = gridspec.GridSpec(2, 2)
plt.subplot(gs[0])
_plot_ref_params('Teff', 'radius')
if len(self.results) > 0:
plt.plot(self.lincomb_results[region_num]['Teff'],
self.lincomb_results[region_num]['radius'],
's',
color='purple',
label='Derived Parameters')
plots.label_axes('Teff', 'radius')
plt.subplot(gs[1])
_plot_ref_params('Teff', 'radius', zoom=True)
if len(self.results) > 0:
plt.plot(self.lincomb_results[region_num]['Teff'],
self.lincomb_results[region_num]['radius'],
's',
color='purple',
label='Derived Parameters')
plots.label_axes('Teff', 'radius', rescale=False)
plt.subplot(gs[2])
_plot_ref_params('feh', 'radius')
if len(self.results) > 0:
plt.plot(self.lincomb_results[region_num]['feh'],
self.lincomb_results[region_num]['radius'],
's',
color='purple',
label='Derived Parameters')
plots.label_axes('feh', 'radius')
plt.subplot(gs[3])
_plot_ref_params('feh', 'radius', zoom=True)
if len(self.results) > 0:
plt.plot(self.lincomb_results[region_num]['feh'],
self.lincomb_results[region_num]['radius'],
's',
color='purple',
label='Derived Parameters')
plots.label_axes('feh', 'radius', rescale=False)
# code-stop-imports
rc('savefig',dpi=160)
# code-start-loadlibrary: load in the library around the Mgb triplet
lib = specmatchemp.library.read_hdf(wavlim=[5140,5200])
# code-stop-loadlibrary
# code-start-library: Here's how the library spans the HR diagram.
fig = figure()
plot(lib.library_params.Teff, lib.library_params.radius,'k.',)
smplot.label_axes('Teff','radius')
# code-stop-library
fig.savefig('quickstart-library.png')
# code-start-library-labeled
fig = figure()
g = lib.library_params.groupby('source')
colors = ['Red','Orange','LimeGreen','Cyan','RoyalBlue','Magenta','ForestGreen']
i = 0
for source, idx in g.groups.items():
cut = lib.library_params.ix[idx]
color = colors[i]
plot(