def plot_trans_vs_U(iZ, iNHI, grid, trans, ax):
atoms,nums = zip(*[split_trans_name(t) for t in trans])
count = dict((k, 0) for k in COLORS)
for atom, num in zip(atoms, nums):
col = COLORS[atom]
N = grid.N[atom][iNHI, :, iZ, roman_map[num]]
pl.plot(grid.nH, N, lw=3, ls=LS[count[atom] % len(LS)],
color=col,label=atom+num)
count[atom] += 1
ax.legend(frameon=0, ncol=2)
ax.set_ylabel('$\log_{10}\ N\ (\mathrm{cm}^{-2})$')
ax.set_xlabel('$\log_{10}\ n_\mathrm{H}\ (\mathrm{cm}^{-3})$')
puttext(0.05, 0.05, '$\log_{10}\ N_\mathrm{HI}=%.3g,\ Z=%.3g$' % (
grid.NHI[iNHI],grid.Z[iZ]),ax)
ax.set_xlim(grid.nH[0]+0.01, grid.nH[-1]-0.01)
ax1 = ax.twiny()
x0,x1 = ax.get_xlim()
const = (grid.U + grid.nH)[0]
assert np.allclose(const, grid.U + grid.nH)
ax1.set_xlim(const - x0, const - x1)
ax1.set_xlabel('$\log_{10}\ U$')
def plot_rho_QSO_prop(fig,rho,ab,iqso_from_id,qso_props=None):
"""For a given well defined rho dictionary, it plots different QSO
properties with the aim to test fo systematics.
"""
global rbin
if qso_props is None:
qso_props=['z','snr','nabs','RMAG','IMAG','ZMAG']
for i,qso_prop in enumerate(qso_props):
mean = []
err = []
median = []
n = []
for r in rho:
# QSOs in this bin
qids = np.array([iqso_from_id[ind] for ind in r['qid']], dtype=int)
prop = ab['qso'][qso_prop][qids]
mean += [np.mean(prop)]
err += [np.std(prop)]
median += [np.median(prop)]
n += [len(qids)]
ax = fig.add_subplot(1,len(qso_props),i)
ax.errorbar(rbin.cen,mean,yerr=err,capsize=0,fmt='-o',color='k')
ax.plot(rbin.cen,median,'-r',lw=3)
for j in range(len(n)):
puttext(rbin.cen[j], 0.1 , str(n[j]), ax, color='k',
fontsize=10, xcoord='data', ha='center')
ax.set_xlabel('rho')
ax.set_ylabel(qso_prop)
def plot_autocorr(chain):
""" Plot the autocorrelation of parameters in the chain. This can
be slow if there are many parameters.
"""
nwalkers, nsamples, npar = chain.shape
nrows, ncols = get_nrows_ncols(npar)
fig, axes = get_fig_axes(nrows, ncols, npar)
for i, ax in enumerate(axes):
acor = [autocorr(chain[j, :, i], maxlag=150) for j in xrange(nwalkers)]
distplot(np.transpose(acor), ax=ax)
ax.axhline(0, color='r', lw=0.5)
puttext(0.1, 0.1, P['names'][i], ax, fontsize=16)
return fig, axes
def plot_autocorr(chain):
""" Plot the autocorrelation of parameters in the chain. This can
be slow if there are many parameters.
"""
nwalkers, nsamples, npar = chain.shape
nrows, ncols = get_nrows_ncols(npar)
fig,axes = get_fig_axes(nrows, ncols, npar)
for i,ax in enumerate(axes):
acor = [autocorr(chain[j,:,i], maxlag=150) for j in xrange(nwalkers)]
distplot(np.transpose(acor), ax=ax)
ax.axhline(0, color='r', lw=0.5)
puttext(0.1, 0.1, P['names'][i], ax, fontsize=16)
return fig, axes
def plot_posteriors_burn(chain, P, npar='all'):
""" Plot the posteriors of a burn-in sample
"""
nwalkers, nsamples, npartot = chain.shape
c = chain.reshape(-1, npartot)
if npar == 'all':
npar = npartot
nrows, ncols = get_nrows_ncols(npar)
fig, axes = get_fig_axes(nrows, ncols, npar)
for i, ax in enumerate(axes):
j = i + 1
if j == npar:
j = 0
ax.plot(c[:, i], c[:, j], '.', ms=1, color='0.5')
# plot initial walker positions
ax.plot(chain[:, 0, i],
chain[:, 0, j],
'.r',
ms=4,
label='p$_{initial}$')
# and final positions
ax.plot(chain[:, -1, i],
chain[:, -1, j],
'.y',
ms=4,
label='p$_{final}$')
ax.plot(P['ml'][i], P['ml'][j], 'xk', ms=12, mew=4)
ax.plot(P['ml'][i], P['ml'][j], 'xr', ms=10, mew=2)
puttext(0.95, 0.05, P['names'][i], ax, fontsize=16, ha='right')
puttext(0.05, 0.95, P['names'][j], ax, fontsize=16, va='top')
x0, x1 = chain[:, 0, i].min(), chain[:, 0, i].max()
dx = x1 - x0
ax.set_xlim(x0 - 0.1 * dx, x1 + 0.1 * dx)
y0, y1 = chain[:, 0, j].min(), chain[:, 0, j].max()
dy = y1 - y0
ax.set_ylim(y0 - 0.1 * dy, y1 + 0.1 * dy)
axes[0].legend()
return fig, axes
def plot_trans_vs_NHI(iZ, inH, grid, trans, ax):
atoms,nums = zip(*[split_trans_name(t) for t in trans])
count = dict((k, 0) for k in COLORS)
for atom, num in zip(atoms, nums):
col = COLORS[atom]
N = grid.N[atom][:, inH, iZ, roman_map[num]]
pl.plot(grid.NHI, N, lw=2, ls=LS[count[atom] % len(LS)],
color=col,label=atom+num)
count[atom] += 1
pl.legend(frameon=0, ncol=3)
pl.ylabel('$\log_{10}\ N (\mathrm{cm}^{-2}$')
pl.xlabel('$\log_{10}\ N_\mathrm{HI} (\mathrm{cm}^{-2})$')
puttext(0.05, 0.95, '$n_{H}=%.3g\ Z=%.3g$' % (
grid.nH[inH], grid.Z[iZ]), ax)
def plot_trace(chain, nwplot=10):
""" Plot the sample trace for a subset of walkers for each parameter.
"""
nwalkers, nsample, npar = chain.shape
nrows, ncols = get_nrows_ncols(npar)
fig, axes = get_fig_axes(nrows, ncols, npar)
# number of walkers to plot
nwplot = min(nsample, nwplot)
for i in range(npar):
ax = axes[i]
for j in range(0, nwalkers, max(1, nwalkers // nwplot)):
ax.plot(chain[j, :, i], lw=0.5)
puttext(0.9, 0.1, P['names'][i], ax, ha='right', fontsize=14)
ax.set_xlim(0, nsample)
return fig, nwplot
def plot_trace(chain, nwplot=10):
""" Plot the sample trace for a subset of walkers for each parameter.
"""
nwalkers, nsample, npar = chain.shape
nrows, ncols = get_nrows_ncols(npar)
fig, axes = get_fig_axes(nrows, ncols, npar)
# number of walkers to plot
nwplot = min(nsample, nwplot)
for i in range(npar):
ax = axes[i]
for j in range(0, nwalkers, max(1, nwalkers // nwplot)):
ax.plot(chain[j,:,i], lw=0.5)
puttext(0.9, 0.1, P['names'][i], ax, ha='right', fontsize=14)
ax.set_xlim(0, nsample)
return fig, nwplot
def plot_posteriors_burn(chain, P, npar="all"):
""" Plot the posteriors of a burn-in sample
"""
nwalkers, nsamples, npartot = chain.shape
c = chain.reshape(-1, npartot)
if npar == "all":
npar = npartot
nrows, ncols = get_nrows_ncols(npar)
fig, axes = get_fig_axes(nrows, ncols, npar)
for i, ax in enumerate(axes):
j = i + 1
if j == npar:
j = 0
ax.plot(c[:, i], c[:, j], ".", ms=1, color="0.5")
# plot initial walker positions
ax.plot(chain[:, 0, i], chain[:, 0, j], ".r", ms=4, label="p$_{initial}$")
# and final positions
ax.plot(chain[:, -1, i], chain[:, -1, j], ".y", ms=4, label="p$_{final}$")
ax.plot(P["ml"][i], P["ml"][j], "xk", ms=12, mew=4)
ax.plot(P["ml"][i], P["ml"][j], "xr", ms=10, mew=2)
puttext(0.95, 0.05, P["names"][i], ax, fontsize=16, ha="right")
puttext(0.05, 0.95, P["names"][j], ax, fontsize=16, va="top")
x0, x1 = chain[:, 0, i].min(), chain[:, 0, i].max()
dx = x1 - x0
ax.set_xlim(x0 - 0.1 * dx, x1 + 0.1 * dx)
y0, y1 = chain[:, 0, j].min(), chain[:, 0, j].max()
dy = y1 - y0
ax.set_ylim(y0 - 0.1 * dy, y1 + 0.1 * dy)
axes[0].legend()
return fig, axes
def plot_posteriors_burn(chain, P, npar='all'):
""" Plot the posteriors of a burn-in sample
"""
nwalkers, nsamples, npartot = chain.shape
c = chain.reshape(-1, npartot)
if npar == 'all':
npar = npartot
nrows, ncols = get_nrows_ncols(npar)
fig, axes = get_fig_axes(nrows, ncols, npar)
for i,ax in enumerate(axes):
j = i+1
if j == npar:
j = 0
ax.plot(c[:, i], c[:, j], '.', ms=1, color='0.5')
# plot initial walker positions
ax.plot(chain[:,0,i], chain[:,0,j], '.r', ms=4, label='p$_{initial}$')
# and final positions
ax.plot(chain[:,-1,i], chain[:,-1,j], '.y', ms=4, label='p$_{final}$')
ax.plot(P['ml'][i], P['ml'][j], 'xk', ms=12, mew=4)
ax.plot(P['ml'][i], P['ml'][j], 'xr', ms=10, mew=2)
puttext(0.95, 0.05, P['names'][i], ax, fontsize=16, ha='right')
puttext(0.05, 0.95, P['names'][j], ax, fontsize=16, va='top')
x0, x1 = chain[:, 0, i].min(), chain[:, 0, i].max()
dx = x1 - x0
ax.set_xlim(x0 - 0.1*dx, x1 + 0.1*dx)
y0, y1 = chain[:, 0, j].min(), chain[:, 0, j].max()
dy = y1 - y0
ax.set_ylim(y0 - 0.1*dy, y1 + 0.1*dy)
axes[0].legend()
return fig, axes
def plot_rho_QSO_prop(fig, rho, ab, iqso_from_id, qso_props=None):
"""For a given well defined rho dictionary, it plots different QSO
properties with the aim to test fo systematics.
"""
global rbin
if qso_props is None:
qso_props = ['z', 'snr', 'nabs', 'RMAG', 'IMAG', 'ZMAG']
for i, qso_prop in enumerate(qso_props):
mean = []
err = []
median = []
n = []
for r in rho:
# QSOs in this bin
qids = np.array([iqso_from_id[ind] for ind in r['qid']], dtype=int)
prop = ab['qso'][qso_prop][qids]
mean += [np.mean(prop)]
err += [np.std(prop)]
median += [np.median(prop)]
n += [len(qids)]
ax = fig.add_subplot(1, len(qso_props), i)
ax.errorbar(rbin.cen, mean, yerr=err, capsize=0, fmt='-o', color='k')
ax.plot(rbin.cen, median, '-r', lw=3)
for j in range(len(n)):
puttext(rbin.cen[j],
0.1,
str(n[j]),
ax,
color='k',
fontsize=10,
xcoord='data',
ha='center')
ax.set_xlabel('rho')
ax.set_ylabel(qso_prop)
labels = ['0.6 < Wr$_{2796}$ < 5']
colors = 'g'
symbols = 'o'
offsets = [0]
for i in range(len(labels)):
dNdz, dNdz_er, n = find_dndz_vs_rho(
rho, ab['MgII'], iMgII_from_id,
ewbins.edges[i], ewbins.edges[i+1])
y = np.log10(dNdz)
ylo = np.log10(dNdz - dNdz_er)
yhi = np.log10(dNdz + dNdz_er)
errplot(rbin.cen + offsets[i], y, (ylo, yhi), ax=ax,
fmt=colors[i]+symbols[i], label=labels[i])
for j in range(len(n)):
puttext(rbin.cen[j], 0.03 + i*0.03, str(n[j]), ax, color=colors[i],
fontsize=10, xcoord='data', ha='center')
ax.legend(frameon=0)
ax.set_xlabel('Cluster-absorber impact par. (proper Mpc)')
ax.set_ylabel(r'$dN/dz\ (MgII)$')
# skip last bin, where not all pairs are measured.
ax.set_xlim(rbin.edges[0] - rbin.halfwidth[0],
rbin.edges[-2] + rbin.halfwidth[-1])
make_log_xlabels(ax)
make_log_ylabels(ax)
#fig3.savefig(run_id + '/dNdz_vs_rho.png')
plt.show()
if 0:
# check whether QSO properties are consistent across bins
ax1 = fig1.add_subplot(111)
ax2 = fig2.add_subplot(111)
for ax in (ax1,ax2):
N0 = -21
for i,n in enumerate(names):
T = Table(n)
ind = T.energy.searchsorted(0.99)
if i == len(names) - 1:
ref = T
ref.log10jnu = T.log10jnu - T.log10jnu[ind] + N0
color = '0.5' if i != 2 else 'k'
ax.semilogx(T.energy, T.log10jnu - T.log10jnu[ind] + N0, '-', color=color)
puttext(0.04, 0.07, '$\mathrm{Haardt\ & \ Madau}$\n$2012,\ z=2.5$',
ax=ax1, ha='left')
# plot a QSO spectrum
X = 10**np.linspace(-1, 4, 1000) # hv
# alpha = -1.7
# fnu = np.log10(X**alpha)
# ind = X.searchsorted(0.99)
# plt.semilogx(X, fnu - fnu[ind] + N0, '-' + 'c')
#alpha = -1.5
# Scott
alpha = -0.56
fnu = np.log10(X**alpha)
ind = X.searchsorted(0.99)
ax2.semilogx(X, fnu - fnu[ind] + N0, '--c', lw=2,
label=r'$J_\nu\propto\ \nu^{%.3g}$' % alpha)
dy2 = -dy2
valign = 'top'
color='b'
if tr_wrest[i] > 1215.67:
color = 'r'
elif tr_wrest[i] > 912.:
color = 'g'
ax.text(xvals[i] + dx, y[i]+ dy2, t, fontsize=9,
rotation=90, ha=halign, va=valign, color=color)
ax.plot([xvals[i]]*2, [y[i] + dy0, y[i] + dy1], lw=0.5, color='0.5')
c = 'k'
ax.set_xlabel('$\mathrm{Energy\ (Rydbergs)}$')
ax.set_ylabel(r'$\log_{10}\ [J_{\nu}\ \mathrm{(erg/s/cm^2/Hz/ster)}]$')
ax.set_xlim(-0.6, 1.6)
make_log_xlabels(ax)
ax.set_ylim(-24.3, -16.8)
puttext(0.95, 0.85, '$\mathrm{Haardt\ & \ Madau}$\n$2012,\ z=2.5$',
ax=ax, ha='right')
puttext(0.95, 0.75, '$\mathrm{Strongest\ }\lambda_0>\ 1216\ \AA$', ax=ax, ha='right',color='r',fontsize=10)
puttext(0.95, 0.7, '$\mathrm{Strongest\ }\lambda_0>\ 912\ \AA$', ax=ax, ha='right',color='g' ,fontsize=10)
puttext(0.95, 0.65, '$\mathrm{Strongest\ }\lambda_0>\ 600\ \AA$', ax=ax, ha='right',color='b' ,fontsize=10)
puttext(0.95, 0.55, '$\mathrm{Verner\ }P\ >\ %g$' % prob, ax=ax, ha='right',fontsize=10)
fig1.savefig('incident_field_trans.pdf')
def make_plot(sp, transitions, model, models, ticks, opt, width=None,
height=None, aspect=0.5, textloc='lower left',
unrelated=[], cols=None, flw=0.5, mlw=0.5, nrows=None, ncols=None,
unrelated_alpha=0.5, trans_fontsize='medium', zerovel_lw=0.5,
tlw=1):
"""
Parameters
----------
cols : dict
An optional dictionary of colours. The keys 'data', 'resid',
'ticks' and 'model' can be defined.
"""
if cols is None:
cols = {}
if 'data' not in cols:
cols['data'] = None
if 'resid' not in cols:
cols['resid'] = 'g'
if 'model' not in cols:
cols['model'] = 'r'
if 'ticks' not in cols:
cols['ticks'] = 'b'
wa = sp.wa
wextra = np.diff(wa)[0] * 2
nfl = sp.fl / sp.co
ner = sp.er / sp.co
if opt.f26name is not None:
resid = (nfl - model) / ner
# actual plotting
ntrans = len(transitions)
if None in (nrows, ncols):
nrows, ncols = get_nrows_ncols(ntrans)
assert ntrans <= nrows*ncols
fig, vplot = get_fig_axes(nrows, ncols, ntrans, width=width, height=height,
aspect=aspect)
colour = get_colours(transitions)
zp1 = opt.redshift + 1
vmax = opt.vmax
if opt.vmin is None:
vmin = -abs(vmax)
else:
vmin = opt.vmin
betamin = vmin / c_kms
betamax = vmax / c_kms
fig.subplots_adjust(left=0.05, right=0.99, bottom=0.07, top=0.99,
wspace=0.0001, hspace=0.00001)
# plot top down, so we need reversed()
for i,trans in enumerate(transitions):
ax = vplot.axes[i]
ion = trans[0].split()[0]
watrans = trans[1][0]
obswa = watrans * zp1
wmin = obswa * (1 - 3*abs(betamin))
wmax = obswa * (1 + 3*abs(betamax))
cond = between(wa, wmin, wmax)
wa1 = wa[cond]
fl = nfl[cond]
vel = (wa1 / obswa - 1) * c_kms
ax.axhline(0, color='gray', lw=0.5)
c = cols['data']
if c is None:
c = colour[ion]
ax.plot(vel, fl, color=c, lw=flw, ls='steps-mid')
for w0,w1 in unrelated:
c0 = between(wa1, w0, w1)
c1 = between(wa1, w0-wextra, w1+wextra)
if c0.any():
ax.plot(vel[c0], fl[c0], 'w', lw=flw+2, ls='-',
drawstyle='steps-mid')
#ax.plot(vel[c0], fl[c0], '--', lw=flw, color=c,
# drawstyle='steps-mid')
ax.plot(vel[c1], fl[c1], '-', lw=flw, color=c,
alpha=unrelated_alpha, drawstyle='steps-mid')
ax.axhline(1, color='gray', lw=0.5, ls='dashed', zorder=3)
for m in models:
ax.plot(vel, m[cond], cols['model'], lw=0.2)
if opt.f26name is not None:
if opt.modelfill:
ax.fill_between(vel, model[cond], 1, lw=0,
color=cols['model'], alpha=0.4, zorder=0)
else:
ax.plot(vel, model[cond], cols['model'], lw=mlw)
for w0,w1 in unrelated:
c0 = between(wa1, w0, w1)
if c0.any():
ax.plot(vel[c0], model[cond][c0], 'w', lw=flw+2, ls=':')
if opt.residuals and opt.f26name is not None:
mult = 0.05
offset = -0.2
ax.scatter(vel, offset + mult*resid[cond], marker='.', s=5,
faceted=False, c=cols['resid'])
ax.axhline(offset - mult, color='0.5', lw=0.3)
ax.axhline(offset + mult, color='0.5', lw=0.3)
if len(ticks) > 0:
tickwmin = obswa * (1 + betamin)
tickwmax = obswa * (1 + betamax)
wticks = ticks.wa
cond = between(wticks, tickwmin, tickwmax)
if cond.any():
# should really check ion name here too...
c0 = np.abs(trans[1][0] - ticks.wa0[cond]) < 1e-2
vel = (wticks[cond][c0] / obswa - 1) * c_kms
#import pdb; pdb.set_trace()
for j,t in enumerate(ticks[cond][c0]):
T = plot_tick_vel(ax, vel[j], 0, t, lw=tlw, col=cols['ticks'])
vel = (wticks[cond][~c0] / obswa - 1) * c_kms
for j,t in enumerate(ticks[cond][~c0]):
T = plot_tick_vel(ax, vel[j], 0, t, lw=tlw, col=cols['ticks'],
ls=':')
#bbox = dict(facecolor='w', edgecolor='None')
transf = mtransforms.blended_transform_factory(
ax.transAxes, ax.transData)
name = trans[0]
if opt.osc:
name = name + ' %.3g' % trans[1]['osc']
if textloc == 'lower left':
ax.text(0.03, 0.02, name, fontsize=trans_fontsize, transform=transf) # bbox=bbox,
elif textloc =='upper right':
puttext(0.97, 0.97, name, ax, fontsize=trans_fontsize, va='top',
ha='right') # bbox=bbox,
#ax.text(0.98, 0.02, name, ha='right', fontsize=13, transform=transf) # bbox=bbox,
for i,ax in enumerate(vplot.axes):
ax.axvline(0, color='0.7', lw=zerovel_lw, zorder=0)
ax.set_xlim(vmin, vmax)
ax.set_ylim(-0.49, 1.49)
return fig, vplot
dy0 = -dy0
dy1 = -dy1
dy2 = -dy2
valign = "top"
color = "b"
if tr_wrest[i] > 1215.67:
color = "r"
elif tr_wrest[i] > 912.0:
color = "g"
ax.text(xvals[i] + dx, y[i] + dy2, t, fontsize=9, rotation=90, ha=halign, va=valign, color=color)
ax.plot([xvals[i]] * 2, [y[i] + dy0, y[i] + dy1], lw=0.5, color="0.5")
c = "k"
ax.set_xlabel("$\mathrm{Energy\ (Rydbergs)}$")
ax.set_ylabel(r"$\log_{10}\ [J_{\nu}\ \mathrm{(erg/s/cm^2/Hz/ster)}]$")
ax.set_xlim(-0.6, 1.6)
make_log_xlabels(ax)
ax.set_ylim(-24.3, -16.8)
puttext(0.95, 0.85, "$\mathrm{Haardt\ & \ Madau}$\n$2012,\ z=2.5$", ax=ax, ha="right")
puttext(0.95, 0.75, "$\mathrm{Strongest\ }\lambda_0>\ 1216\ \AA$", ax=ax, ha="right", color="r", fontsize=10)
puttext(0.95, 0.7, "$\mathrm{Strongest\ }\lambda_0>\ 912\ \AA$", ax=ax, ha="right", color="g", fontsize=10)
puttext(0.95, 0.65, "$\mathrm{Strongest\ }\lambda_0>\ 600\ \AA$", ax=ax, ha="right", color="b", fontsize=10)
puttext(0.95, 0.55, "$\mathrm{Verner\ }P\ >\ %g$" % prob, ax=ax, ha="right", fontsize=10)
fig1.savefig("incident_field_trans.pdf")
def plot_posteriors(chain, P, nplot='all'):
""" Plot the posterior distributions for a series of parameter
samples. chain has shape (nsample, nparameters).
"""
if 'posterior_plots' in opt:
temp = opt['posterior_plots'].split(',')
pairs = []
for pair in temp:
pairs.append(pair.split())
nplot = len(pairs)
elif nplot == 'all':
nplot = chain.shape[-1]
nrows, ncols = get_nrows_ncols(nplot)
fig,axes = get_fig_axes(nrows, ncols, nplot)
for i,ax in enumerate(axes):
if 'posterior_plots' in opt:
i0 = P['names'].index(pairs[i][0])
i1 = P['names'].index(pairs[i][1])
else:
i0 = i
i1 = i + 1
if i1 == npar:
i1 = 0
#ax.plot(chain[:,0,i], chain[:,0,j], '.r', ms=4, label='p$_{initial}$')
dhist(chain[:, i0], chain[:, i1],
xbins=P['bins'][i0], ybins=P['bins'][i1],
fmt='.', ms=1.5, c='0.5', chist='b', ax=ax, loc='left, bottom')
#if contours:
# i1sig = get_levels(np.array([chain[:,i], chain[:,j]]).T,)
# #cont = ax.contour(*par, colors='k',linewidths=0.5)
# for ind in P.ijoint_sig:
# x,y = chain[:,i][ind], chain[:,j][ind]
# delaunay = Delaunay(np.array([x, y]).T)
# for i0,i1 in delaunay.convex_hull:
# ax.plot([x[i0], x[i1]], [y[i0], y[i1]], 'k', lw=0.5)
x,y = chain[:,i0][P['ijoint_sig'][1]], chain[:,i1][P['ijoint_sig'][1]]
ax.plot(x,y,'g.', ms=3, mew=0)
x,y = chain[:,i0][P['ijoint_sig'][0]], chain[:,i1][P['ijoint_sig'][0]]
ax.plot(x,y,'r.', ms=3, mew=0)
ax.plot(P['ml'][i0], P['ml'][i1], 'xk', ms=12, mew=4)
ax.plot(P['ml'][i0], P['ml'][i1], 'xr', ms=10, mew=2)
c = 'crimson'
ax.axvline(P['p1sig'][i0][0], ymax=0.2, color=c, lw=0.5)
ax.axvline(P['p1sig'][i0][1], ymax=0.2, color=c, lw=0.5)
ax.axhline(P['p1sig'][i1][0], xmax=0.2, color=c, lw=0.5)
ax.axhline(P['p1sig'][i1][1], xmax=0.2, color=c, lw=0.5)
ax.axvline(P['median'][i0], ymax=0.2, color=c, lw=1.5)
ax.axhline(P['median'][i1], xmax=0.2, color=c, lw=1.5)
puttext(0.95, 0.05, P['names'][i0], ax, fontsize=16, ha='right')
puttext(0.05, 0.95, P['names'][i1], ax, fontsize=16, va='top')
x0, x1 = np.percentile(chain[:,i0], [5, 95])
dx = x1 - x0
ax.set_xlim(x0 - dx, x1 + dx)
y0, y1 = np.percentile(chain[:,i1], [5, 95])
dy = y1 - y0
ax.set_ylim(y0 - dy, y1 + dy)
return fig, axes
def triplot(names, vals, sigvals, fig, indirect={}, labels=None, fontsize=14):
from barak.plot import hist_yedge, hist_xedge, puttext
npar = len(names)
bins = {}
for n in names:
x0, x1 = vals[n].min(), vals[n].max()
dx = x1 - x0
lo = x0 - 0.1*dx
hi = x1 + 0.1*dx
bins[n] = np.linspace(lo, hi, 20)
axes = {}
for i0,n0 in enumerate(names):
for i1,n1 in enumerate(names):
if i0 == i1:# or i1 < i0: # uncomment to keep just one triangle.
continue
ax = fig.add_subplot(npar,npar, i0 * npar + i1 + 1)
ax.locator_params(tight=True, nbins=8)
ax.xaxis.set_minor_locator(AutoMinorLocator())
ax.yaxis.set_minor_locator(AutoMinorLocator())
axes[(n0 + ' ' + n1)] = ax
y,x = vals[n0], vals[n1]
if USE_HEXBIN:
ax.hexbin(x,y,cmap=CM, gridsize=40,linewidths=0.1)
else:
ax.plot(x,y,'r.', ms=0.5, mew=0)#, alpha=0.5)
color = 'k' if n0 not in indirect else 'g'
text = labels[n0] if labels is not None else n0
puttext(0.05, 0.95, text, ax, color=color ,fontsize=fontsize, va='top')
color = 'k' if n1 not in indirect else 'g'
text = labels[n1] if labels is not None else n1
puttext(0.95, 0.08, text, ax, color=color ,fontsize=fontsize, ha='right')
# set limits
y0, y1 = np.percentile(vals[n0], [5, 95])
dy = y1 - y0
ax.set_ylim(y0 - dy, y1 + dy)
x0, x1 = np.percentile(vals[n1], [5, 95])
dx = x1 - x0
ax.set_xlim(x0 - dx, x1 + dx)
c = 'k'
if i0 == 0:
ax.xaxis.set_tick_params(labeltop='on')
ax.xaxis.set_tick_params(labelbottom='off')
for t in ax.get_xticklabels():
t.set_rotation(60)
elif i0 == npar-1 or (i0 == npar-2 and i1 == npar-1):
hist_xedge(vals[n1], ax, color='forestgreen',
fill=dict(color='forestgreen',alpha=0.3),
bins=bins[n1], loc='bottom')
ax.axvline(sigvals[n1][0], ymax=0.2, color=c, lw=0.5)
ax.axvline(sigvals[n1][1], ymax=0.2, color=c, lw=0.5)
cen = sum(sigvals[n1]) / 2.
ax.axvline(cen, ymax=0.2, color=c, lw=1.5)
for t in ax.get_xticklabels():
t.set_rotation(60)
else:
ax.set_xticklabels('')
if not (i1 == 0 or (i0 == 0 and i1 == 1) or i1 == npar-1):
ax.set_yticklabels('')
if (i0 == 0 and i1 == 1) or i1 == 0:
hist_yedge(vals[n0], ax, color='forestgreen',
fill=dict(color='forestgreen',alpha=0.3),
bins=bins[n0], loc='left')
ax.axhline(sigvals[n0][0], xmax=0.2, color=c, lw=0.5)
ax.axhline(sigvals[n0][1], xmax=0.2, color=c, lw=0.5)
cen = sum(sigvals[n0]) / 2.
ax.axhline(cen, xmax=0.2, color=c, lw=1.5)
if i1 == npar - 1:
ax.yaxis.set_tick_params(labelright='on')
ax.yaxis.set_tick_params(labelleft='off')
#ax.minorticks_on()
return axes
dNdz, dNdz_er, n = find_dndz_vs_rho(rho, ab['MgII'], iMgII_from_id,
ewbins.edges[i],
ewbins.edges[i + 1])
y = np.log10(dNdz)
ylo = np.log10(dNdz - dNdz_er)
yhi = np.log10(dNdz + dNdz_er)
errplot(rbin.cen + offsets[i],
y, (ylo, yhi),
ax=ax,
fmt=colors[i] + symbols[i],
label=labels[i])
for j in range(len(n)):
puttext(rbin.cen[j],
0.03 + i * 0.03,
str(n[j]),
ax,
color=colors[i],
fontsize=10,
xcoord='data',
ha='center')
ax.legend(frameon=0)
ax.set_xlabel('Cluster-absorber impact par. (proper Mpc)')
ax.set_ylabel(r'$dN/dz\ (MgII)$')
# skip last bin, where not all pairs are measured.
ax.set_xlim(rbin.edges[0] - rbin.halfwidth[0],
rbin.edges[-2] + rbin.halfwidth[-1])
make_log_xlabels(ax)
make_log_ylabels(ax)
#fig3.savefig(run_id + '/dNdz_vs_rho.png')
plt.show()
def plot_posteriors(chain, P, nplot='all'):
""" Plot the posterior distributions for a series of parameter
samples. chain has shape (nsample, nparameters).
"""
if nplot == 'all':
nplot = chain.shape[-1]
#nrows, ncols = get_nrows_ncols(nplot)
#fig,axes = get_fig_axes(nrows, ncols, nplot)
fig = pl.figure(figsize=(8.4, 8.4))
fig.subplots_adjust(left=0.05, bottom=0.05, hspace=0.001, wspace=0.001)
axes = []
pl.rc('xtick', labelsize=8)
pl.rc('ytick', labelsize=8)
for i0 in xrange(nplot):
for i1 in xrange(nplot):
if i0 == i1:# or i1 < i0:
continue
ax = fig.add_subplot(nplot,nplot, i0 * nplot + i1 + 1)
y,x = chain[:,i0], chain[:,i1]
ax.plot(x,y,'r.', ms=1, mew=0)#, alpha=0.5)
#y,x = chain[:,i0][P['ijoint_sig'][1]], chain[:,i1][P['ijoint_sig'][1]]
#ax.plot(x,y,'g.', ms=1.5, mew=0)
#y,x = chain[:,i0][P['ijoint_sig'][0]], chain[:,i1][P['ijoint_sig'][0]]
#ax.plot(x,y,'r.', ms=1.5, mew=0)
ax.plot(P['ml'][i1], P['ml'][i0], 'xk', ms=8, mew=2)
ax.plot(P['ml'][i1], P['ml'][i0], 'xr', ms=6, mew=1)
puttext(0.05, 0.95, P['names'][i0], ax, fontsize=12, va='top')
puttext(0.95, 0.05, P['names'][i1], ax, fontsize=12, ha='right')
y0, y1 = np.percentile(chain[:,i0], [5, 95])
dy = y1 - y0
ax.set_ylim(y0 - dy, y1 + dy)
x0, x1 = np.percentile(chain[:,i1], [5, 95])
dx = x1 - x0
ax.set_xlim(x0 - dx, x1 + dx)
c = 'crimson'
if i0 == 0:
ax.xaxis.set_tick_params(labeltop='on')
ax.xaxis.set_tick_params(labelbottom='off')
for t in ax.get_xticklabels():
t.set_rotation(60)
elif i0 == nplot-1 or (i0 == nplot-2 and i1 == nplot-1):
hist_xedge(chain[:, i1], ax, fmt='b', bins=P['bins'][i1], loc='bottom')
ax.axvline(P['p1sig'][i1][0], ymax=0.2, color=c, lw=0.5)
ax.axvline(P['p1sig'][i1][1], ymax=0.2, color=c, lw=0.5)
ax.axvline(P['median'][i1], ymax=0.2, color=c, lw=1.5)
for t in ax.get_xticklabels():
t.set_rotation(60)
else:
ax.set_xticklabels('')
if not (i1 == 0 or (i0 == 0 and i1 == 1) or i1 == nplot-1):
ax.set_yticklabels('')
if (i0 == 0 and i1 == 1) or i1 == 0:
hist_yedge(chain[:, i0], ax, fmt='b', bins=P['bins'][i0], loc='left')
ax.axhline(P['p1sig'][i0][0], xmax=0.2, color=c, lw=0.5)
ax.axhline(P['p1sig'][i0][1], xmax=0.2, color=c, lw=0.5)
ax.axhline(P['median'][i0], xmax=0.2, color=c, lw=1.5)
if i1 == nplot - 1:
ax.yaxis.set_tick_params(labelright='on')
ax.yaxis.set_tick_params(labelleft='off')
axes.append(ax)
return fig, axes
def plot_posteriors(chain, P, nplot='all'):
""" Plot the posterior distributions for a series of parameter
samples. chain has shape (nsample, nparameters).
"""
if 'posterior_plots' in opt:
temp = opt['posterior_plots'].split(',')
pairs = []
for pair in temp:
pairs.append(pair.split())
nplot = len(pairs)
elif nplot == 'all':
nplot = chain.shape[-1]
nrows, ncols = get_nrows_ncols(nplot)
fig, axes = get_fig_axes(nrows, ncols, nplot)
for i, ax in enumerate(axes):
if 'posterior_plots' in opt:
i0 = P['names'].index(pairs[i][0])
i1 = P['names'].index(pairs[i][1])
else:
i0 = i
i1 = i + 1
if i1 == npar:
i1 = 0
#ax.plot(chain[:,0,i], chain[:,0,j], '.r', ms=4, label='p$_{initial}$')
dhist(chain[:, i0],
chain[:, i1],
xbins=P['bins'][i0],
ybins=P['bins'][i1],
fmt='.',
ms=1.5,
c='0.5',
chist='b',
ax=ax,
loc='left, bottom')
#if contours:
# i1sig = get_levels(np.array([chain[:,i], chain[:,j]]).T,)
# #cont = ax.contour(*par, colors='k',linewidths=0.5)
# for ind in P.ijoint_sig:
# x,y = chain[:,i][ind], chain[:,j][ind]
# delaunay = Delaunay(np.array([x, y]).T)
# for i0,i1 in delaunay.convex_hull:
# ax.plot([x[i0], x[i1]], [y[i0], y[i1]], 'k', lw=0.5)
x, y = chain[:, i0][P['ijoint_sig'][1]], chain[:,
i1][P['ijoint_sig'][1]]
ax.plot(x, y, 'g.', ms=3, mew=0)
x, y = chain[:, i0][P['ijoint_sig'][0]], chain[:,
i1][P['ijoint_sig'][0]]
ax.plot(x, y, 'r.', ms=3, mew=0)
ax.plot(P['ml'][i0], P['ml'][i1], 'xk', ms=12, mew=4)
ax.plot(P['ml'][i0], P['ml'][i1], 'xr', ms=10, mew=2)
c = 'crimson'
ax.axvline(P['p1sig'][i0][0], ymax=0.2, color=c, lw=0.5)
ax.axvline(P['p1sig'][i0][1], ymax=0.2, color=c, lw=0.5)
ax.axhline(P['p1sig'][i1][0], xmax=0.2, color=c, lw=0.5)
ax.axhline(P['p1sig'][i1][1], xmax=0.2, color=c, lw=0.5)
ax.axvline(P['median'][i0], ymax=0.2, color=c, lw=1.5)
ax.axhline(P['median'][i1], xmax=0.2, color=c, lw=1.5)
puttext(0.95, 0.05, P['names'][i0], ax, fontsize=16, ha='right')
puttext(0.05, 0.95, P['names'][i1], ax, fontsize=16, va='top')
x0, x1 = np.percentile(chain[:, i0], [5, 95])
dx = x1 - x0
ax.set_xlim(x0 - dx, x1 + dx)
y0, y1 = np.percentile(chain[:, i1], [5, 95])
dy = y1 - y0
ax.set_ylim(y0 - dy, y1 + dy)
return fig, axes
def plot_posteriors(chain, P, nplot="all"):
""" Plot the posterior distributions for a series of parameter
samples. chain has shape (nsample, nparameters).
"""
if nplot == "all":
nplot = chain.shape[-1]
# nrows, ncols = get_nrows_ncols(nplot)
# fig,axes = get_fig_axes(nrows, ncols, nplot)
fig = pl.figure(figsize=(8.4, 8.4))
fig.subplots_adjust(left=0.05, bottom=0.06, hspace=0.001, wspace=0.001)
axes = []
pl.rc("xtick", labelsize=8)
pl.rc("ytick", labelsize=8)
for i0 in xrange(nplot):
for i1 in xrange(nplot):
if i0 == i1: # or i1 < i0: # uncomment to keep just one triangle.
continue
ax = fig.add_subplot(nplot, nplot, i0 * nplot + i1 + 1)
y, x = chain[:, i0], chain[:, i1]
ax.plot(x, y, "r.", ms=1, mew=0) # , alpha=0.5)
# y,x = chain[:,i0][P['ijoint_sig'][1]], chain[:,i1][P['ijoint_sig'][1]]
# ax.plot(x,y,'g.', ms=1.5, mew=0)
# y,x = chain[:,i0][P['ijoint_sig'][0]], chain[:,i1][P['ijoint_sig'][0]]
# ax.plot(x,y,'r.', ms=1.5, mew=0)
ax.plot(P["ml"][i1], P["ml"][i0], "xk", ms=8, mew=2)
ax.plot(P["ml"][i1], P["ml"][i0], "xr", ms=6, mew=1)
puttext(0.05, 0.95, P["names"][i0], ax, fontsize=12, va="top")
puttext(0.95, 0.05, P["names"][i1], ax, fontsize=12, ha="right")
y0, y1 = np.percentile(chain[:, i0], [5, 95])
dy = y1 - y0
ax.set_ylim(y0 - dy, y1 + dy)
x0, x1 = np.percentile(chain[:, i1], [5, 95])
dx = x1 - x0
ax.set_xlim(x0 - dx, x1 + dx)
c = "crimson"
if i0 == 0:
ax.xaxis.set_tick_params(labeltop="on")
ax.xaxis.set_tick_params(labelbottom="off")
for t in ax.get_xticklabels():
t.set_rotation(60)
elif i0 == nplot - 1 or (i0 == nplot - 2 and i1 == nplot - 1):
hist_xedge(chain[:, i1], ax, fmt="b", bins=P["bins"][i1], loc="bottom")
ax.axvline(P["p1sig"][i1][0], ymax=0.2, color=c, lw=0.5)
ax.axvline(P["p1sig"][i1][1], ymax=0.2, color=c, lw=0.5)
ax.axvline(P["median"][i1], ymax=0.2, color=c, lw=1.5)
for t in ax.get_xticklabels():
t.set_rotation(60)
else:
ax.set_xticklabels("")
if not (i1 == 0 or (i0 == 0 and i1 == 1) or i1 == nplot - 1):
ax.set_yticklabels("")
if (i0 == 0 and i1 == 1) or i1 == 0:
hist_yedge(chain[:, i0], ax, fmt="b", bins=P["bins"][i0], loc="left")
ax.axhline(P["p1sig"][i0][0], xmax=0.2, color=c, lw=0.5)
ax.axhline(P["p1sig"][i0][1], xmax=0.2, color=c, lw=0.5)
ax.axhline(P["median"][i0], xmax=0.2, color=c, lw=1.5)
if i1 == nplot - 1:
ax.yaxis.set_tick_params(labelright="on")
ax.yaxis.set_tick_params(labelleft="off")
axes.append(ax)
return fig, axes
N0 = -21
for i, n in enumerate(names):
T = Table(n)
ind = T.energy.searchsorted(0.99)
if i == len(names) - 1:
ref = T
ref.log10jnu = T.log10jnu - T.log10jnu[ind] + N0
color = '0.5' if i != 2 else 'k'
ax.semilogx(T.energy,
T.log10jnu - T.log10jnu[ind] + N0,
'-',
color=color)
puttext(0.04,
0.07,
'$\mathrm{Haardt\ & \ Madau}$\n$2012,\ z=2.5$',
ax=ax1,
ha='left')
# plot a QSO spectrum
X = 10**np.linspace(-1, 4, 1000) # hv
# alpha = -1.7
# fnu = np.log10(X**alpha)
# ind = X.searchsorted(0.99)
# plt.semilogx(X, fnu - fnu[ind] + N0, '-' + 'c')
#alpha = -1.5
# Scott
alpha = -0.56
fnu = np.log10(X**alpha)
ind = X.searchsorted(0.99)
if not tr[0].startswith('OVI') and not tr[0].startswith('NV'):
for j,v in enumerate(tvel):
#if j in (0,11,13,14,15):
# T = plot_tick_vel(ax, v, 0, '', lw=1, col='b')
if cfg.vmin < v < cfg.vmax:
T = plot_tick_vel(ax, v - vzero, 0.04,'',
height=0.14, lw=1, col='b')
else:
for j,v in enumerate(tvel_o6):
#if j in (0,11,13,14,15):
# T = plot_tick_vel(ax, v, 0, '', lw=1, col='b')
if cfg.vmin < v < cfg.vmax:
T = plot_tick_vel(ax, v - vzero, 0.04,'',
height=0.14, lw=1, col='r')
puttext(0.03, 0.12, label[i], ax, ha='left', va='bottom')
#ax.set_xticks([-50, -25, 0, 25, 50])
ax.set_yticks([0, 0.5, 1])
ax.minorticks_on()
ax.set_xlim(cfg.vmin - vzero, cfg.vmax - vzero)
ax.set_ylim(-0.1, 1.39)
if i > 9:
ax.set_yticklabels('')
if i not in (0,9,10, 19):
ax.set_xticklabels('')
if i == 4:
ax.set_ylabel('Transmission')
fig.text(0.52, 0.01, 'Velocity offset (km s$^{-1}$)', ha='center')
#fig.text(0.04, 0.4, 'Transmission', va='center', ha='center', rotation=90)
# rp must be in units of physical kpc
xi2h_rp = []
for rp in rp_vals:
def integrand(s):
"""rp and s must have the same units """
R = hypot(s, rp)
return np.interp(R, rvals, xi_real)
xi2h_rp.append(quad(integrand, -np.inf, np.inf)[0])
# make a plot of the 1 halo and two halo results. The
# normalisations for each term are arbitrary for now
# convert to proper (physical) coordinates.
rp_vals1 = rp_vals / (1 + REDSHIFT)
xi1h_rp = Sigma_1h(rp_vals1, M)
plt.figure()
ax = plt.gca()
ax.plot(np.log10(rp_vals1), np.log10(xi1h_rp.value) - 21.4)
ax.plot(np.log10(rp_vals1), np.log10(xi2h_rp))
from barak.plot import make_log_xlabels, puttext
puttext(0.9, 0.9, 'log10(mass/m_sun) {:.3g}'.format(np.log10(MASS.value)),
ax, ha='right')
make_log_xlabels(plt.gca())
ax.set_xlabel('log10 r_p (Mpc)')
ax.set_ylabel('log10 xi')
plt.show()
def plot_model(pars):
""" Plot the observed values and errors, along with the predicted
model values.
"""
import matplotlib.pyplot as pl
from barak.plot import draw_arrows, puttext
fig = pl.figure(figsize=(6.4, 3.4))
ax = fig.add_subplot(111)
ms = 6
ipot = [get_ionization_energy(t) for t in tr_plot]
for i,tr in enumerate(tr_plot):
if use_ipot:
ind = ipot[i]
else:
ind = i
if tr in obs:
colour = 'k' if tr in trans else 'w'
fmt = 'o' + colour
val, siglo, sighi = obs[tr]
if siglo == 0:
draw_arrows(ind, val, direction='up', ax=ax,lw=1)
ax.plot(ind, val, fmt,ms=ms)
elif sighi == 0:
draw_arrows(ind, val, direction='down', ax=ax,lw=1)
ax.plot(ind, val, fmt,ms=ms)
else:
ax.plot([ind, ind], [val-siglo, val+sighi], 'k',lw=1)
ax.plot(ind, val, fmt,ms=ms)
ax.text(ind, val + 0.8, tr,
fontsize=10, ha='center')
else:
puttext(ind, 0.02, tr, ax=ax, xcoord='data',
fontsize=10, ha='center')
puttext(0.9,0.1, 'Model', ax, color='r', ha='right')
xvals = list(range(len(tr_plot)))
#print np.asarray(pars).shape
for par in pars:
Nmodel = model(par, for_plot=True)
if use_ipot:
if len(pars) == 1:
ax.plot(ipot, Nmodel, 'r.-', lw=1, zorder=0)
else:
ax.plot(ipot, Nmodel, 'r-', lw=0.2, zorder=0)
else:
if len(pars) == 1:
ax.plot(xvals, Nmodel, 'r.-', lw=1, zorder=0)
else:
ax.plot(xvals, Nmodel, 'r-', lw=0.2, zorder=0)
if use_ipot:
ax.set_xlabel('Ionization potential (eV)')
ax.set_xlim(ipot[0]-1, ipot[-1] + 1)
else:
ax.set_xlim(-0.5, xvals[-1] + 0.5)
ax.set_xticks([])
ax.set_ylabel(r'$\log_{10}\ N$')
fig.tight_layout()
#return fig, ax
return fig
def plot_model(pars):
"""
`regions`, `x` must be defined in the model.py module namespace
"""
from run_emcee.plot_mcmc import get_fig_axes, get_nrows_ncols
linepars = pars
dz = 0
if options['wa_vary']:
linepars, regpars = pars[:-len(regions)], pars[-len(regions):]
lines = copy_par_to_lines(linepars, lines_vary)
nrows, ncols = get_nrows_ncols(len(regions))
fig, axes = get_fig_axes(nrows, ncols, len(regions), width=8.2)
fig.subplots_adjust(wspace=1e-6, hspace=1e-6, right=1, top=1,
left=0.07, bottom=0.06)
z0 = lines.z[1]
zp1 = lines.z.mean()
colours = dict(J0='purple',J1='g',J2='r',J3='b')
for ind, (wa, tau0, sp, (i, j), tname) in enumerate(regions):
tau = tau0.copy()
for l in lines:
if options['wa_vary']:
#import pdb; pdb.set_trace()
dz = regpars[ind] / wa[len(wa)//2] * (l.z + 1)
#print l.z, dz, l.logN
trans = atom[l.name.strip()]
tau += calc_iontau(wa, trans, l.z+1 + dz, l.logN, l.b)
dwpix = wa[1] - wa[0]
flmodel0 = convolve_with_COS_FOS(np.exp(-tau), wa, dwpix)
ymod = np.interp(sp.wa[i:j], wa, flmodel0)
ax = axes[ind]
i1 = max(i-50, 0)
j1 = min(j+50, len(sp.wa)-1)
nfl = sp.fl[i1:j1] / sp.co[i1:j1]
ner = sp.er[i1:j1] / sp.co[i1:j1]
tstr, t = findtrans(tname, atom)
obswa = t.wa * (1 + z0)
dv = c_kms * (sp.wa[i:j] / obswa - 1)
dv1 = c_kms * (sp.wa[i1:j1] / obswa - 1)
tstr = tstr[2:]
ax.axhline(0, color='0.5',lw=0.5)
ax.axhline(1, color='k', lw=0.5, ls='--')
ax.fill_between([-150, 150], -0.35, -0.15, color='0.9')
ax.axhline(-0.25, color='k', lw=0.25)
ax.plot(dv1, nfl, color='0.7', lw=0.5, drawstyle='steps-mid')
ax.plot(dv, sp.fl[i:j]/sp.co[i:j], 'k', lw=0.5, drawstyle='steps-mid')
ax.plot(dv, ymod, color='0.3')
#ax.axvline(0, color='0.5', lw=0.5)
ax.plot([-23]*2, [1.1, 1.4], '0.5')
ax.plot([0]*2, [1.1, 1.4], '0.5')
puttext(0.95, 0.95, tstr, ax, fontsize=9, va='top', ha='right',
color=colours[tstr[:2]])
resid = (sp.fl[i:j]/sp.co[i:j] - ymod) / sp.er[i:j] * sp.co[i:j]
ax.plot(dv, -0.25 + resid*0.1, '.', color='0.3', ms=2)
ax.set_ylim(-0.5, 1.9)
ax.set_xlim(-120, 120)
ax.set_yticks([0, 0.5, 1, 1.5])
ax.set_yticklabels(['0.0', '0.5', '1.0', '1.5'])
ax.set_xticks([-100, -50, 0, 50, 100])
ax.set_xticklabels(['', '-50', '0', '50', ''])
if ind+1 < (ncols*(nrows-1)):
ax.set_xticklabels([])
if ind % ncols:
ax.set_yticklabels([])
fig.text(0.5, 0.00, 'Velocity offset (km/s)', fontsize=12, ha='center',va='bottom')
fig.text(0.015, 0.5, 'Transmission', fontsize=12, rotation=90,va='center')
ndf = len(ydata) - len(pars)
print (((ydata - ymodel(x, pars))/ ysigma) **2).sum() / ndf
pl.savefig('fig/model.pdf')
pl.savefig('fig/model.png',dpi=300)
return fig
lw=1,
col='b')
else:
for j, v in enumerate(tvel_o6):
#if j in (0,11,13,14,15):
# T = plot_tick_vel(ax, v, 0, '', lw=1, col='b')
if cfg.vmin < v < cfg.vmax:
T = plot_tick_vel(ax,
v - vzero,
0.04,
'',
height=0.14,
lw=1,
col='r')
puttext(0.03, 0.12, label[i], ax, ha='left', va='bottom')
#ax.set_xticks([-50, -25, 0, 25, 50])
ax.set_yticks([0, 0.5, 1])
ax.minorticks_on()
ax.set_xlim(cfg.vmin - vzero, cfg.vmax - vzero)
ax.set_ylim(-0.1, 1.39)
if i > 9:
ax.set_yticklabels('')
if i not in (0, 9, 10, 19):
ax.set_xticklabels('')
if i == 4:
ax.set_ylabel('Transmission')
fig.text(0.52, 0.01, 'Velocity offset (km s$^{-1}$)', ha='center')
#fig.text(0.04, 0.4, 'Transmission', va='center', ha='center', rotation=90)
def plot_model(pars):
""" Plot the observed values and errors, along with the predicted
model values.
"""
import matplotlib.pyplot as pl
from barak.plot import draw_arrows, puttext
fig = pl.figure(figsize=(6.4, 3.4))
ax = fig.add_subplot(111)
ms = 6
ipot = [get_ionization_energy(t) for t in tr_plot]
for i, tr in enumerate(tr_plot):
if use_ipot:
ind = ipot[i]
else:
ind = i
if tr in obs:
colour = 'k' if tr in trans else 'w'
fmt = 'o' + colour
val, siglo, sighi = obs[tr]
if siglo == 0:
draw_arrows(ind, val, direction='up', ax=ax, lw=1)
ax.plot(ind, val, fmt, ms=ms)
elif sighi == 0:
draw_arrows(ind, val, direction='down', ax=ax, lw=1)
ax.plot(ind, val, fmt, ms=ms)
else:
ax.plot([ind, ind], [val - siglo, val + sighi], 'k', lw=1)
ax.plot(ind, val, fmt, ms=ms)
ax.text(ind, val + 0.8, tr, fontsize=10, ha='center')
else:
puttext(ind,
0.02,
tr,
ax=ax,
xcoord='data',
fontsize=10,
ha='center')
puttext(0.9, 0.1, 'Model', ax, color='r', ha='right')
xvals = list(range(len(tr_plot)))
#print np.asarray(pars).shape
for par in pars:
Nmodel = model(par, for_plot=True)
if use_ipot:
if len(pars) == 1:
ax.plot(ipot, Nmodel, 'r.-', lw=1, zorder=0)
else:
ax.plot(ipot, Nmodel, 'r-', lw=0.2, zorder=0)
else:
if len(pars) == 1:
ax.plot(xvals, Nmodel, 'r.-', lw=1, zorder=0)
else:
ax.plot(xvals, Nmodel, 'r-', lw=0.2, zorder=0)
if use_ipot:
ax.set_xlabel('Ionization potential (eV)')
ax.set_xlim(ipot[0] - 1, ipot[-1] + 1)
else:
ax.set_xlim(-0.5, xvals[-1] + 0.5)
ax.set_xticks([])
ax.set_ylabel(r'$\log_{10}\ N$')
fig.tight_layout()
#return fig, ax
return fig
Y = np.concatenate([[0], np.repeat(y,2), [0]])
Y = 0.2 * Y / Y.max()
import matplotlib.transforms as mtransforms
trans = mtransforms.blended_transform_factory(ax.transAxes, ax.transData)
ax.plot(Y, b, color='0.5', transform=trans)
if xaxis == 'NHI':
ax.set_xlabel('$N(\mathrm{HI})')
elif xaxis == 'Ntot':
ax.set_xlabel('$\log_{10}\ [N_\mathrm{HI}\ +\ 2N_{\mathrm{H}_2}]\ (\mathrm{cm}^{-2})$')
if xaxis == 'Ntot':
Nvals, fvals = detlim(np.linspace(17,23), NH2=15.)
im = shade_to_line(Nvals, fvals, blend=2., y0=-10, color='0.7')
puttext(0.05, -6.8, '$\mathrm{Detection\ limit}$', pl.gca(), rotation=-19, fontsize=14,
ycoord='data') #'$N$(H$_2)\ < \ 14.5$'
#pl.plot(Nvals, fvals, '-', label='Nh2=12.5')
x0,y0 = NHIlo, 0.7
# typical error bars for other points
# NHI
pl.plot([x0-0.21, x0+0.21], [y0+0.21, y0-0.21], 'k', lw=0.5)
pl.plot([x0, x0], [y0-0.3, y0+0.3], 'k', lw=0.5)
#pl.ylabel('$N$(H$_2$)/($N$(HI) $+\ 2N$(H$_2$)')
pl.ylabel('$\log_{10}\ f_{\mathrm{H}_2}$')
if xaxis == 'Ntot':
pl.xlim(18.7, 22.15)
ax.set_ylim(-8.2, 1.5)
#leg = ax.legend(loc='upper left', ncol=2, frameon=0, scatterpoints=1)
def plot_posteriors(chain, P, nplot='all'):
""" Plot the posterior distributions for a series of parameter
samples. chain has shape (nsample, nparameters).
"""
if nplot == 'all':
nplot = chain.shape[-1]
#nrows, ncols = get_nrows_ncols(nplot)
#fig,axes = get_fig_axes(nrows, ncols, nplot)
fig = pl.figure(figsize=(8.4, 8.4))
fig.subplots_adjust(left=0.05, bottom=0.06, hspace=0.001, wspace=0.001)
axes = []
pl.rc('xtick', labelsize=8)
pl.rc('ytick', labelsize=8)
for i0 in xrange(nplot):
for i1 in xrange(nplot):
if i0 == i1: # or i1 < i0: # uncomment to keep just one triangle.
continue
ax = fig.add_subplot(nplot, nplot, i0 * nplot + i1 + 1)
y, x = chain[:, i0], chain[:, i1]
ax.plot(x, y, 'r.', ms=1, mew=0) #, alpha=0.5)
#y,x = chain[:,i0][P['ijoint_sig'][1]], chain[:,i1][P['ijoint_sig'][1]]
#ax.plot(x,y,'g.', ms=1.5, mew=0)
#y,x = chain[:,i0][P['ijoint_sig'][0]], chain[:,i1][P['ijoint_sig'][0]]
#ax.plot(x,y,'r.', ms=1.5, mew=0)
ax.plot(P['ml'][i1], P['ml'][i0], 'xk', ms=8, mew=2)
ax.plot(P['ml'][i1], P['ml'][i0], 'xr', ms=6, mew=1)
puttext(0.05, 0.95, P['names'][i0], ax, fontsize=12, va='top')
puttext(0.95, 0.05, P['names'][i1], ax, fontsize=12, ha='right')
y0, y1 = np.percentile(chain[:, i0], [5, 95])
dy = y1 - y0
ax.set_ylim(y0 - dy, y1 + dy)
x0, x1 = np.percentile(chain[:, i1], [5, 95])
dx = x1 - x0
ax.set_xlim(x0 - dx, x1 + dx)
c = 'crimson'
if i0 == 0:
ax.xaxis.set_tick_params(labeltop='on')
ax.xaxis.set_tick_params(labelbottom='off')
for t in ax.get_xticklabels():
t.set_rotation(60)
elif i0 == nplot - 1 or (i0 == nplot - 2 and i1 == nplot - 1):
hist_xedge(chain[:, i1],
ax,
fmt='b',
bins=P['bins'][i1],
loc='bottom')
ax.axvline(P['p1sig'][i1][0], ymax=0.2, color=c, lw=0.5)
ax.axvline(P['p1sig'][i1][1], ymax=0.2, color=c, lw=0.5)
ax.axvline(P['median'][i1], ymax=0.2, color=c, lw=1.5)
for t in ax.get_xticklabels():
t.set_rotation(60)
else:
ax.set_xticklabels('')
if not (i1 == 0 or (i0 == 0 and i1 == 1) or i1 == nplot - 1):
ax.set_yticklabels('')
if (i0 == 0 and i1 == 1) or i1 == 0:
hist_yedge(chain[:, i0],
ax,
fmt='b',
bins=P['bins'][i0],
loc='left')
ax.axhline(P['p1sig'][i0][0], xmax=0.2, color=c, lw=0.5)
ax.axhline(P['p1sig'][i0][1], xmax=0.2, color=c, lw=0.5)
ax.axhline(P['median'][i0], xmax=0.2, color=c, lw=1.5)
if i1 == nplot - 1:
ax.yaxis.set_tick_params(labelright='on')
ax.yaxis.set_tick_params(labelleft='off')
axes.append(ax)
return fig, axes
