# Custom redshift sampling
z = np.concatenate(( [0.,],
np.linspace(0.001, 1., 40),
np.linspace(1.05, 3., 34),
np.linspace(3.15, 10., 25) ))
a = 1./(1.+z)
pcts = [2.5, 16., 50., 84., 97.5]
try:
pct = np.load("%s.pctcache.npy" % fn)
print("%s: Loaded percentiles from cache" % fn)
except:
# Load samples
print("%s: Loading samples" % fn)
dat = load_chain(fn)
ode = []
print(dat['h'].shape)
#print("Walkers:", dat['walker'][1200000:1200000+10])
# Choose random subset of samples
#sample_idxs = range(1200000, dat['h'].size, 1)
np.random.seed(SAMPLE_SEED)
sample_idxs = np.random.randint(low=BURNIN, high=dat['h'].size, size=NSAMP)
print("Selecting %d samples out of %d available (burnin %d discarded)" \
% (sample_idxs.size, dat['h'].size - BURNIN, BURNIN))
# Calculate model for selected samples
for i in sample_idxs:
pp = {key: dat[key][i] for key in dat.keys()}
# f = open(fname, 'r')
# hdr = f.readline()
# f.close()
# names = hdr[2:-1].split(' ')
#
# # Load data
# dat = np.genfromtxt(fname).T
# data = {}
# for i, n in enumerate(names):
# data[n] = dat[i]
# return data
if MODE == 'scatter':
P.subplot(111)
for i, fname in enumerate(fnames):
dat = load_chain(fname)
print(dat.keys())
print(dat['w0'].shape)
# Select random subsample
sample_idxs = np.random.randint(low=BURNIN,
high=dat['h'].size,
size=NSAMP)
# Steppyness
#x = (dat['w0'][10000:] - dat['winf'][10000:]) / (1. + dat['zc'][10000:])
#y = dat['logl'][10000:] #dat['zc'][10000:] # logL
# w0, winf
x = dat['w0'][sample_idxs] # FIXME: 100k!
y = dat['winf'][sample_idxs]
P.hist(dat['logl'], bins=80, color=colours[i], histtype='step',
label=labels[i], density=True, range=ranges['logl'])
P.tight_layout()
P.show()
exit()
"""
fig = None
#axes = [P.subplot(321), P.subplot(322), P.subplot(323),
# P.subplot(324), P.subplot(325), P.subplot(326), ]
for i, fname in enumerate(fnames):
# Load data
dat = load_chain(fname, burnin=1500000)
print(dat.keys())
print(dat['w0'].shape)
#idxs = np.where(dat['logl'] > np.max(dat['logl']) - 5.)
lbls = [key for key in dat.keys()]
print(lbls)
lbls.remove('logl')
lbls.remove('walker')
data = np.array([dat[key] for key in lbls]).T
"""
for j, pname in enumerate(['omegaM', 'w0', 'h', 'omegaB', 'winf', 'zc',]):
#!/usr/bin/env python
"""
Plot superposed 1D marginal histograms for two experiments.
"""
import numpy as np
import pylab as plt
from load_data_files import load_chain
fname1 = "chains/final_wztanh_seed200_cmb_lss"
fname2 = "chains/final_wztanh_seed200_cmb_lss_cosvis_pbw"
# Load data
d1 = load_chain(fname1)
d2 = load_chain(fname2)
# Reshape into per-walker chains
nw = np.max(d1['walker']).astype(int) + 1
for k in d1.keys():
d1[k] = d1[k].reshape((-1, nw))
for k in d2.keys():
d2[k] = d2[k].reshape((-1, nw))
#d2 = d2.reshape((d2.shape[0], -1, nw))
print(d1['walker'].shape, d2['walker'].shape)
# Create Figure
fig = plt.figure(constrained_layout=True)
gs = fig.add_gridspec(2, 3)
ax11 = fig.add_subplot(gs[0, 0])
ax12 = fig.add_subplot(gs[0, 1])
ax13 = fig.add_subplot(gs[0, 2])