def plot_all_kSZ_curves_JKs_in_current_folder(show=True):
'''In the current folder, it looks for all files with .jk extension.
For all those files, plots the velocity curve with JK errorbars.
show: shows the plot interactively, False: saves the figure to disk.'''
fnames = glob.glob('*.pck')
for j, fname in enumerate(fnames):
jk = JK.load_JK(fname)
label = '%s (%i gals)' % (jk.query, jk.N_objects_in_this_run)
plt.errorbar(jk.rsep + j,
jk.kSZ_curveFullDataset,
yerr=jk.errorbars,
marker='o',
linestyle='',
label=label)
plt.xlabel('$r_{sep}$[Mpc]')
plt.ylabel('p [$\\mu$K]')
plt.axhline(0, color='black')
plt.legend(loc='upper right')
if show:
plt.show()
else:
plt.savefig('kSZ_curves.pdf')
plt.savefig('kSZ_curves.png', dpi=200)
plt.close()
def test_JK_container():
df = produceFakeCatalog()
params = produceFakeParams()
params.JK_NGROUPS = 3
jk = JK.JK_container(df, params, distributed=False)
diff_sq = np.sum((np.diag(jk.cov) - jk.errorbars**2)**2)
assert diff_sq < 1e10
def plot_JK_runtimes_in_current_folder(show=True):
'''In the current folder, plot the run times for all the .pck files.'''
fnames = glob.glob('*.pck')
runtimes = []
n_objects = []
for fname in fnames:
jk = JK.load_JK(fname)
runtimes.append(jk.runtime)
n_objects.append(jk.N_objects_in_this_run)
runtimes = np.array(runtimes)
n_objects = np.array(n_objects)
argsorted = np.argsort(n_objects)
n_objects = n_objects[argsorted]
runtimes = runtimes[argsorted]
plt.scatter(n_objects, runtimes)
ax = plt.gca()
ax.set_xscale('log')
ax.set_yscale('log')
plt.title('Distributed iskay 50JK performance\n'
'20 cores per machine, 10 machines')
plt.xlabel('N$_{galaxies}$')
plt.ylabel('Runtime [s]')
if show:
plt.show()
else:
if not os.path.exists('plots'):
os.makedirs('plots')
plt.savefig('plots/runtimes.pdf')
plt.close()
def plot_all_JK_errorbars_in_current_folder(show=True):
'''Idem to plot_all_kSZ_curves_JKs_in_current_folder but for error bar
only plots.'''
fnames = glob.glob('*.pck')
for fname in fnames:
jk = JK.load_JK(fname)
label = '%s (%i gals)' % (jk.query, jk.N_objects_in_this_run)
plt.scatter(jk.rsep, jk.errorbars, marker='o', label=label)
plt.xlabel('$r_{sep}$[Mpc]')
plt.ylabel('p [$\\mu$K]')
plt.legend(loc='upper right')
plt.title('JK (%i it.) errorbars' % jk.JK_Ngroups)
if show:
plt.show()
else:
plt.savefig('JK_errorbars.pdf')
plt.savefig('JK_errorbars.png', dpi=200)
plt.close()
def make_plots(fnames):
if len(fnames) == 0:
pass
else:
print("Making plots for:")
plt.figure(figsize=[8, 4.5])
for j in range(len(fnames)):
fname = fnames[j]
print(fname)
jk = JK.load_JK(fname)
m = re.match("(lum.*e10)", jk.query)
label = m.group(1)
label = label + ', N: %1.1fk' % (jk.N_objects_in_this_run / 1e3
) # noqa
plt.errorbar(jk.rsep + j * 1.5,
jk.kSZ_curveFullDataset,
yerr=jk.errorbars,
label=label,
ls='',
marker='.')
if 'jk' in fname:
folder = 'results_jk'
elif 'bs' in fname:
if 'bsdt' in fname:
folder = 'results_bsdt'
else:
folder = 'results_bs'
plt.axhline(0, color='black')
plt.legend(loc='lower right')
plt.title(jk.name[:jk.name.find('_lum_')])
plt.xlabel('$r_{sep} [Mpc]$')
plt.ylabel('p [$\\mu K$]')
if 'disjoint' in fname:
fname_out = '%s/kSZ_velocityCurves_disjoint_bins' % (folder)
else:
fname_out = '%s/kSZ_velocityCurves' % (folder)
plt.ylim([-0.3, 0.3])
plt.savefig(fname_out + '.png', dpi=120)
plt.savefig(fname_out + '.pdf')
#plt.show()
plt.close()
matplotlib.rc('font', **font)
fnames = glob.glob('./results_*/*.pck')
assert len(fnames) > 0
show = True
showTitle = False
def getPath(fname):
return '/'.join(fname.split('/')[:-1])
limitBins = 19 # show only these bins
for fname in fnames:
print("Plotting %s" % fname)
jk = JK.load_JK(fname)
labels = ["%i" % rsep for rsep in jk.rsep]
jk.corr.index = labels
jk.corr.columns = labels
#plt.figure(figsize=[6, 6])
sns.heatmap(jk.corr.iloc[:limitBins, :limitBins],
vmin=-1,
vmax=1,
cmap='seismic',
cbar_kws={"ticks": [-1, -0.5, 0.0, 0.5, 1.0]})
plt.xlabel('r [Mpc]')
plt.ylabel('r [Mpc]')
if showTitle:
plt.title(jk.name)
plt.tight_layout()
plt.subplots_adjust(top=0.95)
#! /nfs/user/pag227/miniconda/bin/python
''' pairwsise kSZ analysis script example for iskay.
Usage: iskay_analysis.py param.ini
Written by: P. Gallardo.
'''
from iskay import paramTools
from iskay import catalogTools
from iskay import JK
import sys
param_fname = sys.argv[1]
params = paramTools.params(param_fname)
df = catalogTools.preProcessedCat(howMany=None,
query=params.CAT_QUERY).df
jk = JK.JK(df, params, distributed=True)