def run_synth(self, planes=None, force=False):
full_synth_dir = os.path.join(STARFISH, self.synth_dir)
if not os.path.exists(full_synth_dir):
os.makedirs(full_synth_dir)
# Use PHAT AST from the outer field (field 0)
crowd_path = os.path.join(self.synth_dir, "crowding.dat")
full_crowd_path = os.path.join(STARFISH, crowd_path)
tbl = PhatAstTable()
tbl.write_crowdfile_for_field(full_crowd_path, 0,
bands=PHAT_BANDS)
crowd = ExtantCrowdingTable(crowd_path)
# No extinction, yet
young_av = ExtinctionDistribution()
old_av = ExtinctionDistribution()
rel_extinction = np.ones(len(PHAT_BANDS), dtype=float)
for av in (young_av, old_av):
av.set_uniform(0.)
if planes is None:
planes = self.planes.all_planes
self.synth = Synth(self.synth_dir, self.builder, self.lockfile, crowd,
rel_extinction,
young_extinction=young_av,
old_extinction=old_av,
planes=planes,
mass_span=(0.08, 150.),
nstars=10000000)
existing_synth = len(glob(
os.path.join(STARFISH, self.synth_dir, "z*"))) == 0
if existing_synth or force:
self.synth.run_synth(n_cpu=4, clean=False)
def run_synth(self, planes=None, force=False):
full_synth_dir = os.path.join(STARFISH, self.synth_dir)
if not os.path.exists(full_synth_dir):
os.makedirs(full_synth_dir)
# Use PHAT AST from the outer field (field 0)
crowd_path = os.path.join(self.synth_dir, "crowding.dat")
full_crowd_path = os.path.join(STARFISH, crowd_path)
tbl = PhatAstTable()
tbl.write_crowdfile_for_field(full_crowd_path, 0, bands=PHAT_BANDS)
crowd = ExtantCrowdingTable(crowd_path)
# No extinction, yet
young_av = ExtinctionDistribution()
old_av = ExtinctionDistribution()
rel_extinction = np.ones(len(PHAT_BANDS), dtype=float)
for av in (young_av, old_av):
av.set_uniform(0.)
if planes is None:
planes = self.planes.all_planes
self.synth = Synth(self.synth_dir,
self.builder,
self.lockfile,
crowd,
rel_extinction,
young_extinction=young_av,
old_extinction=old_av,
planes=planes,
mass_span=(0.08, 150.),
nstars=10000000)
existing_synth = len(glob(os.path.join(STARFISH, self.synth_dir,
"z*"))) == 0
if existing_synth or force:
self.synth.run_synth(n_cpu=4, clean=False)
def run_synth(self, n_cpu=1, config_only=False):
full_synth_dir = os.path.join(STARFISH, self.synth_dir)
self.synth = Synth(self.synth_dir,
self.builder,
self.lockfile,
self.crowd,
self.bands,
self.rel_extinction,
young_extinction=self.young_av,
old_extinction=self.old_av,
planes=self.all_planes,
mass_span=(0.08, 150.),
nstars=10000000)
existing_synth = len(glob(
os.path.join(full_synth_dir, "z*"))) > 0
if not config_only:
if not existing_synth:
self.synth.run_synth(n_cpu=n_cpu, clean=False)
class Pipeline(object):
"""Pipeline for Multi-CMD fitting and comparison"""
def __init__(self, brick, root_dir, isoc_args=None, phases=None):
super(Pipeline, self).__init__()
self.brick = brick
self.catalog = Catalog(brick)
self.root_dir = root_dir
self.isoc_dir = os.path.join(root_dir, 'isoc')
self.lib_dir = os.path.join(root_dir, 'lib')
self.synth_dir = os.path.join(root_dir, 'synth')
self.z_grid = [0.015, 0.019, 0.024]
self.get_isochrones(isoc_args=isoc_args, phases=phases)
self.build_lockfile()
self.planes = PhatPlanes()
self.run_synth()
self.fits = OrderedDict()
self._solution_tables = {}
def get_solution_table(self, key):
if key not in self._solution_tables:
tbl = self.fits[key].solution_table()
self._solution_tables[key] = tbl
return tbl
def get_isochrones(self, isoc_args=None, phases=None):
if isoc_args is None:
isoc_args = {}
if not os.path.exists(os.path.join(STARFISH, self.isoc_dir)):
for z in self.z_grid:
r_wfc3 = AgeGridRequest(z,
min_log_age=6.6,
max_log_age=10.13,
delta_log_age=0.02,
photsys='wfc3_wide',
**isoc_args)
r_acs = AgeGridRequest(z,
min_log_age=6.6,
max_log_age=10.13,
delta_log_age=0.02,
photsys='acs_wfc',
**isoc_args)
isoc_set = join_isochrone_sets(r_wfc3.isochrone_set,
r_acs.isochrone_set,
left_bands=WFC3_BANDS,
right_bands=ACS_BANDS)
for isoc in isoc_set:
isoc = Isochrone(isoc)
isoc.rename_column('F275W1', 'F275W')
if phases is not None:
sels = []
for p in phases:
sels.append(np.where(isoc['stage'] == p)[0])
s = np.concatenate(sels)
isoc = isoc[s]
isoc.export_for_starfish(os.path.join(
STARFISH, self.isoc_dir),
bands=PHAT_BANDS)
d = Distance(785 * u.kpc)
self.builder = LibraryBuilder(self.isoc_dir,
self.lib_dir,
nmag=len(PHAT_BANDS),
dmod=d.distmod.value,
iverb=3)
if not os.path.exists(self.builder.full_isofile_path):
self.builder.install()
def build_lockfile(self):
if not os.path.exists(os.path.join(STARFISH, self.synth_dir)):
os.makedirs(os.path.join(STARFISH, self.synth_dir))
self.lockfile = Lockfile(self.builder.read_isofile(),
self.synth_dir,
unbinned=False)
# Bin young isochrones
young_grid = np.linspace(6.5, 8.95, 10)
for i, logage0 in enumerate(young_grid[:-1]):
logage0 = logage0
logage1 = young_grid[i + 1]
z_str = "0019"
mean_age = (logage0 + logage1) / 0.2
name = "z{0}_{1:05.2f}".format(z_str, mean_age)
self.lockfile.lock_box(name, (logage0, logage1), (0.014, 0.025))
# Bin old isochrones
old_grid = np.arange(1e9, 14 * 1e9, 1e9)
for i, age0 in enumerate(old_grid[:-1]):
logage0 = np.log10(age0 - 0.05 * 1e9)
logage1 = np.log10(old_grid[i + 1])
z_str = "0019"
mean_age = (logage0 + logage1) / 0.2
name = "z{0}_{1:05.2f}".format(z_str, mean_age)
self.lockfile.lock_box(name, (logage0, logage1), (0.014, 0.025))
def run_synth(self, planes=None, force=False):
full_synth_dir = os.path.join(STARFISH, self.synth_dir)
if not os.path.exists(full_synth_dir):
os.makedirs(full_synth_dir)
# Use PHAT AST from the outer field (field 0)
crowd_path = os.path.join(self.synth_dir, "crowding.dat")
full_crowd_path = os.path.join(STARFISH, crowd_path)
tbl = PhatAstTable()
tbl.write_crowdfile_for_field(full_crowd_path, 0, bands=PHAT_BANDS)
crowd = ExtantCrowdingTable(crowd_path)
# No extinction, yet
young_av = ExtinctionDistribution()
old_av = ExtinctionDistribution()
rel_extinction = np.ones(len(PHAT_BANDS), dtype=float)
for av in (young_av, old_av):
av.set_uniform(0.)
if planes is None:
planes = self.planes.all_planes
self.synth = Synth(self.synth_dir,
self.builder,
self.lockfile,
crowd,
rel_extinction,
young_extinction=young_av,
old_extinction=old_av,
planes=planes,
mass_span=(0.08, 150.),
nstars=10000000)
existing_synth = len(glob(os.path.join(STARFISH, self.synth_dir,
"z*"))) == 0
if existing_synth or force:
self.synth.run_synth(n_cpu=4, clean=False)
def fit_planes(self, key, color_planes, phot_colors, redo=False):
fit_dir = os.path.join(self.root_dir, key)
data_root = os.path.join(fit_dir, "phot.")
for plane, (band1, band2) in zip(color_planes, phot_colors):
self.catalog.write(band1, band2, data_root, plane.suffix)
sfh = SFH(data_root, self.synth, fit_dir, planes=color_planes)
if (not os.path.exists(sfh.full_outfile_path)) or redo:
sfh.run_sfh()
self.fits[key] = sfh
def show_isoc_phase_sim_hess(self, fig):
opt_sim = self.planes.get_sim_hess(('f475w', 'f814w'), self.synth,
self.lockfile)
ir_sim = self.planes.get_sim_hess(('f110w', 'f160w'), self.synth,
self.lockfile)
opt_cmd = self.planes[('f475w', 'f814w')]
ir_cmd = self.planes[('f110w', 'f160w')]
gs = gridspec.GridSpec(2,
3,
wspace=0.4,
bottom=0.2,
width_ratios=[1., 1., 0.1])
ax_opt = fig.add_subplot(gs[0, 0])
ax_ir = fig.add_subplot(gs[0, 1])
ax_obs_opt = fig.add_subplot(gs[1, 0])
ax_obs_ir = fig.add_subplot(gs[1, 1])
cb_ax = fig.add_subplot(gs[1, 2])
plot_hess(ax_opt,
opt_sim.hess,
opt_cmd,
opt_sim.origin,
imshow_args=None)
plot_hess(ax_ir, ir_sim.hess, ir_cmd, ir_sim.origin, imshow_args=None)
c = self.catalog.data['f475w_vega'] - self.catalog.data['f814w_vega']
contour_hess(ax_obs_opt,
c,
self.catalog.data['f814w_vega'],
opt_cmd.x_span,
opt_cmd.y_span,
plot_args={'ms': 3})
plot_isochrone_phases(ax_obs_opt, 'F475W', 'F814W', show_cb=False)
# opt_cmd.plot_mask(ax_obs_opt)
ax_obs_opt.set_xlabel(opt_cmd.x_label)
ax_obs_opt.set_ylabel(opt_cmd.y_label)
ax_obs_opt.set_xlim(opt_cmd.xlim)
ax_obs_opt.set_ylim(opt_cmd.ylim)
c = self.catalog.data['f110w_vega'] - self.catalog.data['f160w_vega']
contour_hess(ax_obs_ir,
c,
self.catalog.data['f160w_vega'],
ir_cmd.x_span,
ir_cmd.y_span,
plot_args={'ms': 3})
plot_isochrone_phases(ax_obs_ir,
'F110W',
'F160W',
show_cb=True,
cb_ax=cb_ax)
# ir_cmd.plot_mask(ax_obs_ir)
ax_obs_ir.set_xlabel(ir_cmd.x_label)
ax_obs_ir.set_ylabel(ir_cmd.y_label)
ax_obs_ir.set_xlim(ir_cmd.xlim)
ax_obs_ir.set_ylim(ir_cmd.ylim)
fig.show()
def plot_contour_hess(self, ax, bands, plane_key):
plane = self.planes[plane_key]
c = self.catalog.data[bands[0]] - self.catalog.data[bands[-1]]
contour_hess(ax,
c,
self.catalog.data[bands[-1]],
plane.x_span,
plane.y_span,
plot_args={'ms': 3})
ax.set_xlabel(plane.x_label)
ax.set_ylabel(plane.y_label)
ax.set_xlim(*plane.xlim)
ax.set_ylim(*plane.ylim)
def plot_sim_hess(self, ax, plane_key):
plane = self.planes[plane_key]
sim = self.planes.get_sim_hess(plane_key, self.synth, self.lockfile)
plot_hess(ax, sim.hess, plane, sim.origin, imshow_args=None)
def plot_obs_hess(self, arg1):
pass
def plot_fit_hess(self, arg1):
pass
def plot_predicted_hess(self, arg1):
pass
def plot_triptyk(self,
fig,
ax_obs,
ax_model,
ax_chi,
fit_key,
plane_key,
xtick=1.,
xfmt="%.0f"):
cmapper = lambda: cubehelix.cmap(startHue=240,
endHue=-300,
minSat=1,
maxSat=2.5,
minLight=.3,
maxLight=.8,
gamma=.9)
fit = self.fits[fit_key]
plane = self.planes[plane_key]
ctp = ChiTriptykPlot(fit, plane)
ctp.setup_axes(fig,
ax_obs=ax_obs,
ax_mod=ax_model,
ax_chi=ax_chi,
major_x=xtick,
major_x_fmt=xfmt)
ctp.plot_obs_in_ax(ax_obs, cmap=cmapper())
ctp.plot_mod_in_ax(ax_model, cmap=cmapper())
ctp.plot_chi_in_ax(ax_chi, cmap=cubehelix.cmap())
ax_obs.text(0.0,
1.01,
"Observed",
transform=ax_obs.transAxes,
size=8,
ha='left')
ax_model.text(0.0,
1.01,
"Model",
transform=ax_model.transAxes,
size=8,
ha='left')
ax_chi.text(0.0,
1.01,
r"$\log \chi^2$",
transform=ax_chi.transAxes,
size=8,
ha='left')
def plot_isoc_grid_ages(self, ax, band1, band2, show_cb=False, cb_ax=None):
isoc_set = get_demo_age_grid(
**dict(isoc_kind='parsec_CAF09_v1.2S', photsys_version='yang'))
cmap = cubehelix.cmap(startHue=240,
endHue=-300,
minSat=1,
maxSat=2.5,
minLight=.3,
maxLight=.8,
gamma=.9)
norm = mpl.colors.Normalize(vmin=7., vmax=10.1)
scalar_map = mpl.cm.ScalarMappable(norm=norm, cmap=cmap)
scalar_map.set_array(np.array([isoc.age for isoc in isoc_set]))
d = Distance(785 * u.kpc)
for isoc in isoc_set:
ax.plot(isoc[band1] - isoc[band2],
isoc[band2] + d.distmod.value,
c=scalar_map.to_rgba(np.log10(isoc.age)),
lw=0.8)
if show_cb:
cb = plt.colorbar(mappable=scalar_map,
cax=cb_ax,
ax=ax,
ticks=np.arange(6., 10.2))
cb.set_label(r"log(age)")
def plot_isoc_grid_phases(self,
ax,
band1,
band2,
show_cb=False,
cb_ax=None):
plot_isochrone_phases(ax, band1, band2, show_cb=show_cb, cb_ax=cb_ax)
def show_lockfile(self,
fig,
logage_lim=(6.2, 10.2),
logzzsol_lim=(-0.2, 0.2)):
# fig = plt.figure(figsize=(6, 6))
ax = fig.add_subplot(111)
plot_isochrone_logage_logzsol(ax, self.builder, c='k', s=8)
plot_lock_polygons(ax, self.lockfile, facecolor='None', edgecolor='r')
ax.set_xlim(*logage_lim)
ax.set_ylim(*logzzsol_lim)
ax.set_xlabel(r"$\log(A)$")
ax.set_ylabel(r"$\log(Z/Z_\odot)$")
fig.show()
class Pipeline(object):
"""Pipeline for Multi-CMD fitting and comparison"""
def __init__(self, brick, root_dir, isoc_args=None, phases=None):
super(Pipeline, self).__init__()
self.brick = brick
self.catalog = Catalog(brick)
self.root_dir = root_dir
self.isoc_dir = os.path.join(root_dir, 'isoc')
self.lib_dir = os.path.join(root_dir, 'lib')
self.synth_dir = os.path.join(root_dir, 'synth')
self.z_grid = [0.015, 0.019, 0.024]
self.get_isochrones(isoc_args=isoc_args, phases=phases)
self.build_lockfile()
self.planes = PhatPlanes()
self.run_synth()
self.fits = OrderedDict()
self._solution_tables = {}
def get_solution_table(self, key):
if key not in self._solution_tables:
tbl = self.fits[key].solution_table()
self._solution_tables[key] = tbl
return tbl
def get_isochrones(self, isoc_args=None, phases=None):
if isoc_args is None:
isoc_args = {}
if not os.path.exists(os.path.join(STARFISH, self.isoc_dir)):
for z in self.z_grid:
r_wfc3 = AgeGridRequest(z,
min_log_age=6.6,
max_log_age=10.13,
delta_log_age=0.02,
photsys='wfc3_wide', **isoc_args)
r_acs = AgeGridRequest(z,
min_log_age=6.6,
max_log_age=10.13,
delta_log_age=0.02,
photsys='acs_wfc', **isoc_args)
isoc_set = join_isochrone_sets(r_wfc3.isochrone_set,
r_acs.isochrone_set,
left_bands=WFC3_BANDS,
right_bands=ACS_BANDS)
for isoc in isoc_set:
isoc = Isochrone(isoc)
isoc.rename_column('F275W1', 'F275W')
if phases is not None:
sels = []
for p in phases:
sels.append(np.where(isoc['stage'] == p)[0])
s = np.concatenate(sels)
isoc = isoc[s]
isoc.export_for_starfish(os.path.join(STARFISH,
self.isoc_dir),
bands=PHAT_BANDS)
d = Distance(785 * u.kpc)
self.builder = LibraryBuilder(self.isoc_dir, self.lib_dir,
nmag=len(PHAT_BANDS),
dmod=d.distmod.value,
iverb=3)
if not os.path.exists(self.builder.full_isofile_path):
self.builder.install()
def build_lockfile(self):
if not os.path.exists(os.path.join(STARFISH, self.synth_dir)):
os.makedirs(os.path.join(STARFISH, self.synth_dir))
self.lockfile = Lockfile(self.builder.read_isofile(), self.synth_dir,
unbinned=False)
# Bin young isochrones
young_grid = np.linspace(6.5, 8.95, 10)
for i, logage0 in enumerate(young_grid[:-1]):
logage0 = logage0
logage1 = young_grid[i + 1]
z_str = "0019"
mean_age = (logage0 + logage1) / 0.2
name = "z{0}_{1:05.2f}".format(z_str, mean_age)
self.lockfile.lock_box(name, (logage0, logage1), (0.014, 0.025))
# Bin old isochrones
old_grid = np.arange(1e9, 14 * 1e9, 1e9)
for i, age0 in enumerate(old_grid[:-1]):
logage0 = np.log10(age0 - 0.05 * 1e9)
logage1 = np.log10(old_grid[i + 1])
z_str = "0019"
mean_age = (logage0 + logage1) / 0.2
name = "z{0}_{1:05.2f}".format(z_str, mean_age)
self.lockfile.lock_box(name, (logage0, logage1), (0.014, 0.025))
def run_synth(self, planes=None, force=False):
full_synth_dir = os.path.join(STARFISH, self.synth_dir)
if not os.path.exists(full_synth_dir):
os.makedirs(full_synth_dir)
# Use PHAT AST from the outer field (field 0)
crowd_path = os.path.join(self.synth_dir, "crowding.dat")
full_crowd_path = os.path.join(STARFISH, crowd_path)
tbl = PhatAstTable()
tbl.write_crowdfile_for_field(full_crowd_path, 0,
bands=PHAT_BANDS)
crowd = ExtantCrowdingTable(crowd_path)
# No extinction, yet
young_av = ExtinctionDistribution()
old_av = ExtinctionDistribution()
rel_extinction = np.ones(len(PHAT_BANDS), dtype=float)
for av in (young_av, old_av):
av.set_uniform(0.)
if planes is None:
planes = self.planes.all_planes
self.synth = Synth(self.synth_dir, self.builder, self.lockfile, crowd,
rel_extinction,
young_extinction=young_av,
old_extinction=old_av,
planes=planes,
mass_span=(0.08, 150.),
nstars=10000000)
existing_synth = len(glob(
os.path.join(STARFISH, self.synth_dir, "z*"))) == 0
if existing_synth or force:
self.synth.run_synth(n_cpu=4, clean=False)
def fit_planes(self, key, color_planes, phot_colors, redo=False):
fit_dir = os.path.join(self.root_dir, key)
data_root = os.path.join(fit_dir, "phot.")
for plane, (band1, band2) in zip(color_planes, phot_colors):
self.catalog.write(band1, band2, data_root, plane.suffix)
sfh = SFH(data_root, self.synth, fit_dir, planes=color_planes)
if (not os.path.exists(sfh.full_outfile_path)) or redo:
sfh.run_sfh()
self.fits[key] = sfh
def show_isoc_phase_sim_hess(self, fig):
opt_sim = self.planes.get_sim_hess(('f475w', 'f814w'),
self.synth, self.lockfile)
ir_sim = self.planes.get_sim_hess(('f110w', 'f160w'),
self.synth, self.lockfile)
opt_cmd = self.planes[('f475w', 'f814w')]
ir_cmd = self.planes[('f110w', 'f160w')]
gs = gridspec.GridSpec(2, 3, wspace=0.4, bottom=0.2,
width_ratios=[1., 1., 0.1])
ax_opt = fig.add_subplot(gs[0, 0])
ax_ir = fig.add_subplot(gs[0, 1])
ax_obs_opt = fig.add_subplot(gs[1, 0])
ax_obs_ir = fig.add_subplot(gs[1, 1])
cb_ax = fig.add_subplot(gs[1, 2])
plot_hess(ax_opt, opt_sim.hess, opt_cmd, opt_sim.origin,
imshow_args=None)
plot_hess(ax_ir, ir_sim.hess, ir_cmd, ir_sim.origin,
imshow_args=None)
c = self.catalog.data['f475w_vega'] - self.catalog.data['f814w_vega']
contour_hess(ax_obs_opt, c, self.catalog.data['f814w_vega'],
opt_cmd.x_span, opt_cmd.y_span,
plot_args={'ms': 3})
plot_isochrone_phases(ax_obs_opt, 'F475W', 'F814W', show_cb=False)
# opt_cmd.plot_mask(ax_obs_opt)
ax_obs_opt.set_xlabel(opt_cmd.x_label)
ax_obs_opt.set_ylabel(opt_cmd.y_label)
ax_obs_opt.set_xlim(opt_cmd.xlim)
ax_obs_opt.set_ylim(opt_cmd.ylim)
c = self.catalog.data['f110w_vega'] - self.catalog.data['f160w_vega']
contour_hess(ax_obs_ir, c, self.catalog.data['f160w_vega'],
ir_cmd.x_span, ir_cmd.y_span,
plot_args={'ms': 3})
plot_isochrone_phases(ax_obs_ir, 'F110W', 'F160W', show_cb=True,
cb_ax=cb_ax)
# ir_cmd.plot_mask(ax_obs_ir)
ax_obs_ir.set_xlabel(ir_cmd.x_label)
ax_obs_ir.set_ylabel(ir_cmd.y_label)
ax_obs_ir.set_xlim(ir_cmd.xlim)
ax_obs_ir.set_ylim(ir_cmd.ylim)
fig.show()
def plot_contour_hess(self, ax, bands, plane_key):
plane = self.planes[plane_key]
c = self.catalog.data[bands[0]] - self.catalog.data[bands[-1]]
contour_hess(ax, c, self.catalog.data[bands[-1]],
plane.x_span, plane.y_span,
plot_args={'ms': 3})
ax.set_xlabel(plane.x_label)
ax.set_ylabel(plane.y_label)
ax.set_xlim(*plane.xlim)
ax.set_ylim(*plane.ylim)
def plot_sim_hess(self, ax, plane_key):
plane = self.planes[plane_key]
sim = self.planes.get_sim_hess(plane_key, self.synth, self.lockfile)
plot_hess(ax, sim.hess, plane, sim.origin,
imshow_args=None)
def plot_obs_hess(self, arg1):
pass
def plot_fit_hess(self, arg1):
pass
def plot_predicted_hess(self, arg1):
pass
def plot_triptyk(self, fig, ax_obs, ax_model, ax_chi, fit_key, plane_key,
xtick=1., xfmt="%.0f"):
cmapper = lambda: cubehelix.cmap(startHue=240, endHue=-300, minSat=1,
maxSat=2.5, minLight=.3,
maxLight=.8, gamma=.9)
fit = self.fits[fit_key]
plane = self.planes[plane_key]
ctp = ChiTriptykPlot(fit, plane)
ctp.setup_axes(fig, ax_obs=ax_obs, ax_mod=ax_model, ax_chi=ax_chi,
major_x=xtick, major_x_fmt=xfmt)
ctp.plot_obs_in_ax(ax_obs, cmap=cmapper())
ctp.plot_mod_in_ax(ax_model, cmap=cmapper())
ctp.plot_chi_in_ax(ax_chi, cmap=cubehelix.cmap())
ax_obs.text(0.0, 1.01, "Observed",
transform=ax_obs.transAxes, size=8, ha='left')
ax_model.text(0.0, 1.01, "Model",
transform=ax_model.transAxes, size=8, ha='left')
ax_chi.text(0.0, 1.01, r"$\log \chi^2$",
transform=ax_chi.transAxes, size=8, ha='left')
def plot_isoc_grid_ages(self, ax, band1, band2,
show_cb=False, cb_ax=None):
isoc_set = get_demo_age_grid(**dict(isoc_kind='parsec_CAF09_v1.2S',
photsys_version='yang'))
cmap = cubehelix.cmap(startHue=240, endHue=-300,
minSat=1, maxSat=2.5, minLight=.3,
maxLight=.8, gamma=.9)
norm = mpl.colors.Normalize(vmin=7., vmax=10.1)
scalar_map = mpl.cm.ScalarMappable(norm=norm, cmap=cmap)
scalar_map.set_array(np.array([isoc.age for isoc in isoc_set]))
d = Distance(785 * u.kpc)
for isoc in isoc_set:
ax.plot(isoc[band1] - isoc[band2],
isoc[band2] + d.distmod.value,
c=scalar_map.to_rgba(np.log10(isoc.age)),
lw=0.8)
if show_cb:
cb = plt.colorbar(mappable=scalar_map,
cax=cb_ax, ax=ax,
ticks=np.arange(6., 10.2))
cb.set_label(r"log(age)")
def plot_isoc_grid_phases(self, ax, band1, band2,
show_cb=False, cb_ax=None):
plot_isochrone_phases(ax, band1, band2,
show_cb=show_cb, cb_ax=cb_ax)
def show_lockfile(self, fig, logage_lim=(6.2, 10.2),
logzzsol_lim=(-0.2, 0.2)):
# fig = plt.figure(figsize=(6, 6))
ax = fig.add_subplot(111)
plot_isochrone_logage_logzsol(ax, self.builder, c='k', s=8)
plot_lock_polygons(ax, self.lockfile, facecolor='None', edgecolor='r')
ax.set_xlim(*logage_lim)
ax.set_ylim(*logzzsol_lim)
ax.set_xlabel(r"$\log(A)$")
ax.set_ylabel(r"$\log(Z/Z_\odot)$")
fig.show()
class PipelineBase(object):
"""Abstract baseclass for running StarFISH pipelines."""
__metaclass__ = abc.ABCMeta
def __init__(self, **kwargs):
self.root_dir = kwargs.pop('root_dir')
self.n_synth_cpu = kwargs.pop('n_synth_cpu', 1)
self.synth_config_only = kwargs.pop('synth_config_only', False)
# StarFISH product directories
self.isoc_dir = os.path.join(self.root_dir, 'isoc')
self.lib_dir = os.path.join(self.root_dir, 'lib')
self.synth_dir = os.path.join(self.root_dir, 'synth')
# result caches
self.fits = OrderedDict()
self._solution_tables = {}
print "PipelineBase", kwargs
if len(kwargs) > 0:
print "Uncaught arguments:", kwargs
super(PipelineBase, self).__init__()
dirs = (self.isoc_dir, self.lib_dir, self.synth_dir)
for d in dirs:
if not os.path.exists(os.path.join(STARFISH, d)):
os.makedirs(os.path.join(STARFISH, d))
print self.isoc_dir, self.lib_dir, self.synth_dir
# Now run the pipeline
self.setup_isochrones()
print "self.builder.full_isofile_path", self.builder.full_isofile_path
self.build_lockfile()
self.build_crowding()
self.build_extinction()
self.mask_planes() # mask planes based on completeness cuts
if self.synth_config_only:
self.run_synth(n_cpu=1,
config_only=self.synth_config_only)
else:
self.run_synth(n_cpu=self.n_synth_cpu,
config_only=self.synth_config_only)
def run_synth(self, n_cpu=1, config_only=False):
full_synth_dir = os.path.join(STARFISH, self.synth_dir)
self.synth = Synth(self.synth_dir,
self.builder,
self.lockfile,
self.crowd,
self.bands,
self.rel_extinction,
young_extinction=self.young_av,
old_extinction=self.old_av,
planes=self.all_planes,
mass_span=(0.08, 150.),
nstars=10000000)
existing_synth = len(glob(
os.path.join(full_synth_dir, "z*"))) > 0
if not config_only:
if not existing_synth:
self.synth.run_synth(n_cpu=n_cpu, clean=False)
@abc.abstractmethod
def mask_planes(self):
pass
@property
def hold_template(self):
return self.lockfile.empty_hold
def fit(self, fit_key, plane_keys, dataset, fit_dir=None,
redo=False, hold=None):
if fit_dir is None:
fit_dir = os.path.join(self.root_dir, fit_key)
data_root = os.path.join(fit_dir, "phot.")
planes = []
for plane_key in plane_keys:
plane = self.planes[plane_key]
planes.append(plane)
dataset.write_phot(plane.x_mag, plane.y_mag,
data_root, plane.suffix)
sfh = SFH(data_root, self.synth, fit_dir, planes=planes)
if (not os.path.exists(sfh.full_outfile_path)) or redo:
sfh.run_sfh(hold=hold)
self.fits[fit_key] = sfh
def make_fit_diff_hess(self, dataset, fit_key, plane_key):
obs_hess = self.make_obs_hess(dataset, plane_key)
fit_hess = self.make_fit_hess(fit_key, plane_key)
return Hess(obs_hess.hess - fit_hess.hess,
self.planes[plane_key])
def make_chisq_hess(self, dataset, fit_key, plane_key):
obs_hess = self.make_obs_hess(dataset, plane_key)
fit_hess = self.make_fit_hess(fit_key, plane_key)
sigma = np.sqrt(obs_hess.hess)
chi = ((obs_hess.hess - fit_hess.hess) / sigma) ** 2.
return Hess(chi, self.planes[plane_key])
def compute_fit_chi(self, dataset, fit_key, plane_key, chi_hess=None):
"""Compute the reduced chi-sq for the plane with the given fit.
Returns both the sum of chi-sq and the total number of pixels in
in the plane that were not masked.
"""
if chi_hess is None:
chi_hess = self.make_chisq_hess(dataset, fit_key, plane_key)
g = np.where(np.isfinite(chi_hess.masked_hess))
n_pix = len(g[0])
chi_sum = chi_hess.masked_hess[g].sum()
n_amp = len(self.lockfile.active_groups)
return chi_sum / (n_pix - n_amp)
def make_sim_hess(self, plane_key):
return self.get_sim_hess(plane_key)
def make_fit_hess(self, fit_key, plane_key):
plane = self.planes[plane_key]
return SimHess.from_sfh_solution(self.fits[fit_key], plane)
def make_obs_hess(self, dataset, plane_key):
plane = self.planes[plane_key]
x = dataset.get_phot(plane.x_mag)
y = dataset.get_phot(plane.y_mag)
return StarCatalogHess(x, y, plane)
def init_plane_axes(self, ax, plane_key):
plane = self.planes[plane_key]
setup_hess_axes(ax, plane, 'lower')
def plot_sim_hess(self, ax, plane_key, imshow=None):
plane = self.planes[plane_key]
sim = self.get_sim_hess(plane_key)
return plot_hess(ax, sim.hess, plane, sim.origin,
imshow_args=imshow)
def plot_fit_hess(self, ax, fit_key, plane_key, imshow=None):
plane = self.planes[plane_key]
fit_hess = SimHess.from_sfh_solution(self.fits[fit_key], plane)
return plot_hess(ax, fit_hess.hess, plane, fit_hess.origin,
imshow_args=imshow)
def plot_obs_hess(self, ax, dataset, plane_key, imshow=None):
plane = self.planes[plane_key]
x = dataset.get_phot(plane.x_mag)
y = dataset.get_phot(plane.y_mag)
obs_hess = StarCatalogHess(x, y, plane)
return plot_hess(ax, obs_hess.hess, plane, obs_hess.origin,
imshow_args=imshow)
def plot_hess_array(self, ax, hess, plane_key, imshow=None, log=True):
plane = self.planes[plane_key]
return plot_hess(ax, hess, plane, plane.origin,
imshow_args=imshow, log=log)
def plot_lockfile(self, ax,
logage_lim=(6.2, 10.2),
logzzsol_lim=(-0.2, 0.2)):
plot_isochrone_logage_logzsol(ax, self.builder, c='k', s=8)
plot_lock_polygons(ax, self.lockfile, facecolor='None', edgecolor='r')
ax.set_xlim(*logage_lim)
ax.set_ylim(*logzzsol_lim)
ax.set_xlabel(r"$\log(A)$")
ax.set_ylabel(r"$\log(Z/Z_\odot)$")
def plot_triptyk(self, fig, ax_obs, ax_model, ax_chi, fit_key, plane_key,
xtick=1., xfmt="%.0f"):
fit = self.fits[fit_key]
plane = self.planes[plane_key]
ctp = ChiTriptykPlot(fit, plane)
ctp.setup_axes(fig, ax_obs=ax_obs, ax_mod=ax_model, ax_chi=ax_chi,
major_x=xtick, major_x_fmt=xfmt)
ctp.plot_obs_in_ax(ax_obs, cmap=perceptual_rainbow_16.mpl_colormap)
ctp.plot_mod_in_ax(ax_model, cmap=perceptual_rainbow_16.mpl_colormap)
ctp.plot_chi_in_ax(ax_chi, cmap=perceptual_rainbow_16.mpl_colormap)
ax_obs.text(0.0, 1.01, "Observed",
transform=ax_obs.transAxes, size=8, ha='left')
ax_model.text(0.0, 1.01, "Model",
transform=ax_model.transAxes, size=8, ha='left')
ax_chi.text(0.0, 1.01, r"$\log \chi^2$",
transform=ax_chi.transAxes, size=8, ha='left')
def plot_linear_sfh_circles(self, ax, fit_key, ylim=(-0.2, 0.2),
amp_key='sfr'):
sfh = self.fits[fit_key]
cp = LinearSFHCirclePlot(sfh.solution_table())
cp.plot_in_ax(ax, max_area=800, amp_key=amp_key)
for tl in ax.get_ymajorticklabels():
tl.set_visible(False)
ax.set_ylim(*ylim)
def plot_log_sfh_circles(self, ax, fit_key, ylim=(-0.2, 0.2),
amp_key='sfr'):
sfh = self.fits[fit_key]
cp = SFHCirclePlot(sfh.solution_table())
cp.plot_in_ax(ax, max_area=800, amp_key=amp_key)
for logage in np.log10(np.arange(1, 13, 1) * 1e9):
ax.axvline(logage, c='0.8', zorder=-1)
ax.set_ylim(*ylim)