from constrain_parameters import paraH2COmodel
from masked_cubes import (cube303m,cube321m,cube303msm,cube321msm,
cube303,cube321,cube303sm,cube321sm,
sncube, sncubesm)
from masked_cubes import mask as cube_signal_mask
from co_cubes import cube13co, cube18co, cube13cosm, cube18cosm
from noise import noise, noise_cube, sm_noise_cube
from higal_gridded import column_regridded, dusttem_regridded
from common_constants import logabundance,elogabundance
import heating
from gaussian_correction import gaussian_correction
warnings.simplefilter('once')
if 'mf' not in locals():
mf = paraH2COmodel()
# For debugging, to make it faster
# biggest_tree = dend[89]
def get_root(structure):
""" Identify the root of the tree """
if structure.parent is None:
return structure
else:
return get_root(structure.parent)
def measure_dendrogram_properties(dend=None, cube303=cube303,
cube321=cube321, cube13co=cube13co,
cube18co=cube18co, noise_cube=noise_cube,
sncube=sncube,
from astropy.utils.console import ProgressBar
from paths import hpath
from constrain_parameters import paraH2COmodel
from masked_cubes import cube303m, cube321m, cube303, cube321, mask
from noise import noise, noise_cube
from common_constants import logabundance, elogabundance
from higal_gridded import column_regridded
from ratio_cubes import ratio303321, eratio303321, noise_flat
log.warn("Extremely slow for a very small fraction of overall data."
"Try a different method: make_piecewise_temcube.py")
nsigma = 5 # start big to minimize # of failures
mf = paraH2COmodel()
indices = np.where(mask)
usable = ((eratio303321 * nsigma < ratio303321) & (eratio303321 > 0) &
(ratio303321 > 0) & (noise_flat > 1e-10) & (noise_flat < 10) &
(ratio303321 < 100))
assert len(usable) == len(indices)
ngood = np.count_nonzero(usable)
usable_indices = [ind[usable] for ind in indices]
uz, uy, ux = usable_indices
column_flat = column_regridded.data[uy, ux]
uratio303321 = ratio303321[usable]
ueratio303321 = eratio303321[usable]
utline303 = cube303.flattened()[usable]
utline321 = cube321.flattened()[usable]
error distribution
EDIT 5/3/2015: After some further thinking, this approach doesn't really make sense.
We already have the likelihood of the model given the data. This computes the
distribution of the maximum-likelihood model, which simply isn't the same
thing.
However, I think we *can* take the modeled parconstraints and sample from
within those to see what ratios we get out and whether that distribution matches
the input distribution.
"""
import numpy as np
from constrain_parameters import paraH2COmodel
from astropy.utils.console import ProgressBar
model = paraH2COmodel()
ratio303321 = 0.3
eratio303321 = 0.01
logabundance = np.log10(1.2e-9)
elogabundance = 1.0
logh2column = 22.
elogh2column = 1.
mindens = 4.5
emindens = 0.2
linewidth=5.0
model.set_constraints(ratio303321=ratio303321,
eratio303321=eratio303321,
logabundance=logabundance,
elogabundance=elogabundance,
fig = pl.figure(1)
pl.clf()
ax = fig.gca()
ax.plot(data, ratio_to_temperature(data), alpha=0.5, linewidth=2,
label='LTE',
)
ax.set_ylim(0,150)
ax.set_xlabel("Ratio H$_2$CO $3_{2,1}-2_{2,0} / 3_{0,3}-2_{0,2}$")
ax.set_ylabel("Temperature (K)")
fig.savefig(paths.fpath('h2co_temperature_vs_ratio_lte.png'))
pm = paraH2COmodel()
densind = np.argmin(np.abs(pm.densityarr[0,:,0]-4.5))
colind = np.argmin(np.abs(pm.columnarr[0,:,0]-13.5))
ax.plot(pm.modelratio1[:,densind,colind],
pm.temparr[:,densind,colind],
'r',
alpha=0.5,
linewidth=2,
label='$n=10^{4.5}$ cm$^{-3}$, $N=10^{13.5}$ cm$^{-2}$',
)
pl.legend(loc='best')
fig.savefig(paths.fpath('h2co_temperature_vs_ratio_lte_and_radex.png'))
ax = fig.gca()
ax.plot(
data,
ratio_to_temperature(data),
alpha=0.5,
linewidth=2,
label='LTE',
)
ax.set_ylim(0, 150)
ax.set_xlabel("Ratio H$_2$CO $3_{2,1}-2_{2,0} / 3_{0,3}-2_{0,2}$")
ax.set_ylabel("Temperature (K)")
fig.savefig(paths.fpath('h2co_temperature_vs_ratio_lte.png'))
pm = paraH2COmodel()
densind = np.argmin(np.abs(pm.densityarr[0, :, 0] - 4.5))
colind = np.argmin(np.abs(pm.columnarr[0, :, 0] - 13.5))
ax.plot(
pm.modelratio1[:, densind, colind],
pm.temparr[:, densind, colind],
'r',
alpha=0.5,
linewidth=2,
label='$n=10^{4.5}$ cm$^{-3}$, $N=10^{13.5}$ cm$^{-2}$',
)
pl.legend(loc='best')
fig.savefig(paths.fpath('h2co_temperature_vs_ratio_lte_and_radex.png'))
