def custom_data_loader(path):
dataset = data_loader(path, floor=False)
percentile = np.ones_like(dataset["ra"]) * np.nan
dataset = append_fields(
dataset, "percentile", percentile, usemask=False, dtypes=[np.float]
)
return dataset
weighted_quantiles = np.cumsum(sample_weight) - 0.5 * sample_weight
weighted_quantiles /= np.sum(sample_weight)
return np.interp(quantiles, weighted_quantiles, values)
sin_dec_bins = IC86_1_dict["sinDec bins"]
n_log_e_bins = 20 + 1
log_e_bins = np.linspace(2.0, 6.0, n_log_e_bins)
quantiles = np.linspace(0.0, 1.0, n_log_e_bins)
# mc_path = "/lustre/fs22/group/icecube/data_mirror/misc
# /CombinedTracks_AllSky_Uncorrected_MC.npy"
mc_path = IC86_234567_dict["mc_path"]
mc = data_loader(mc_path, floor=False)
exp = data_loader(IC86_234567_dict["exp_path"], floor=False)
# sin_dec_bins = weighted_quantile(
# mc["sinDec"], quantiles, np.ones_like(mc["sinDec"]))
# for gamma in [1.0, 1.5, 2.0, 3.0, 3.7]:
for gamma in [2.0, 3.5]:
# X = [sin_dec_bins for _ in range(n_log_e_bins)]
# Y = []
Z_quantile = np.ones((len(sin_dec_bins), n_log_e_bins))
Z_uniform = np.ones((len(sin_dec_bins), n_log_e_bins))
Z_floor = np.ones((len(sin_dec_bins), n_log_e_bins)) * np.nan
Z_floor2 = np.ones((len(sin_dec_bins), n_log_e_bins)) * np.nan
x = []
y = []
def plot_ratio(seasons):
for season_dict in seasons:
exp = data_loader(season_dict["exp_path"])
mc = data_loader(season_dict["mc_path"])
cut = 4.5
exp_cut = exp[exp["logE"] > cut]
mc_cut = mc[mc["logE"] > cut]
print(len(exp), len(exp_cut))
print(len(mc), len(mc_cut))
print("Min energy", np.exp(min(exp["logE"])))
print(exp.dtype.names)
data = [
("exp", exp["sinDec"], azimuth_proxy(exp, season_dict)),
("mc", np.sin(mc["dec"]), azimuth_proxy(mc, season_dict)),
(("exp_cut"), exp_cut["sinDec"],
azimuth_proxy(exp_cut, season_dict)),
(("mc cut"), np.sin(mc_cut["dec"]),
azimuth_proxy(mc_cut, season_dict)),
]
root = plots_dir + "azimuth/" + season_dict["Data Sample"] + "/"
for (name, sin_dec, az) in data:
save_dir = root + name + "/"
try:
os.makedirs(save_dir)
except OSError:
pass
if name in ["exp", "exp_cut"]:
hist_2d = create_2d_hist(sin_dec,
az,
sin_dec_bins,
weights=np.ones_like(sin_dec)).T
else:
ow = mc["ow"]
trueE = mc["trueE"]
weights = numexpr.evaluate("ow * trueE **(-2)")
mem_use = str(
float(resource.getrusage(resource.RUSAGE_SELF).ru_maxrss) /
1.0e6)
print("")
print("Memory usage max: %s (Gb)" % mem_use)
if name == "mc":
hist_2d = create_2d_hist(
sin_dec,
az,
sin_dec_bins,
weights=energy_pdf.weight_mc(mc, gamma=3.7),
).T
else:
hist_2d = create_2d_hist(
sin_dec,
az,
sin_dec_bins,
weights=energy_pdf.weight_mc(mc_cut, gamma=3.7),
).T
hist_2d = np.array([x / np.mean(x) for x in hist_2d])
plt.figure()
ax = plt.subplot(111)
sin_bin_center = (sin_dec_bins[:-1] + sin_dec_bins[1:]) / 2.0
az_bin_center = (azimuth_bins[:-1] + azimuth_bins[1:]) / 2.0
X, Y = np.meshgrid(az_bin_center, sin_bin_center)
cbar = ax.pcolormesh(X,
Y,
hist_2d,
vmin=0.5,
vmax=1.5,
cmap=cm.get_cmap("seismic"))
# ax.set_aspect('equal')
plt.axis([
min(azimuth_bins),
max(azimuth_bins),
min(sin_dec_bins),
max(sin_dec_bins),
])
ax.set_ylabel("Sin(Declination)")
ax.set_xlabel("Azimuth")
plt.colorbar(cbar)
savename = save_dir + season_dict["Name"] + ".pdf"
plt.savefig(savename)
plt.close()
spline = scipy.interpolate.RectBivariateSpline(az_bin_center,
sin_bin_center,
hist_2d.T,
kx=3,
ky=1,
s=0)
plt.figure()
ax = plt.subplot(111)
sin_bin_center = (sin_dec_bins[:-1] + sin_dec_bins[1:]) / 2.0
az_bin_center = (azimuth_bins[:-1] + azimuth_bins[1:]) / 2.0
z = []
for sin_dec in sin_dec_bins:
z.append(list(spline(azimuth_bins, sin_dec).T.tolist()[0]))
z = np.array(z).T
X, Y = np.meshgrid(azimuth_bins, sin_dec_bins)
cbar = ax.pcolormesh(X,
Y,
z.T,
vmin=0.5,
vmax=1.5,
cmap=cm.get_cmap("seismic"))
# ax.set_aspect('equal')
plt.axis([
min(azimuth_bins),
max(azimuth_bins),
min(sin_dec_bins),
max(sin_dec_bins),
])
ax.set_ylabel("Sin(Declination)")
ax.set_xlabel("Azimuth")
plt.colorbar(cbar)
savename = save_dir + season_dict["Name"] + "_spline.pdf"
plt.savefig(savename)
plt.close()
upgoing = [
azimuth_proxy(exp[exp["sinDec"] > 0.0], season_dict),
azimuth_proxy(mc[mc["dec"] > 0.0], season_dict),
]
downgoing = [
azimuth_proxy(exp[exp["sinDec"] < 0.0], season_dict),
azimuth_proxy(mc[mc["dec"] < 0.0], season_dict),
]
upgoing_cut = [
azimuth_proxy(exp_cut[exp_cut["sinDec"] > 0.0], season_dict),
azimuth_proxy(mc_cut[mc_cut["dec"] > 0.0], season_dict),
]
# downgoing_cut = [azimuth_proxy(exp_cut[exp_cut["sinDec"] < 0.],
# season_dict),
# azimuth_proxy(mc_cut[mc_cut["dec"] < 0.], season_dict)]
for i, dataset in enumerate([upgoing, downgoing, upgoing_cut]):
plt.figure()
ax1 = plt.subplot2grid((4, 1), (0, 0), colspan=3, rowspan=3)
weights = [np.ones_like(x) / float(len(x)) for x in dataset]
n, edges, patches = plt.hist(
dataset,
bins=azimuth_bins,
histtype="step",
weights=weights,
label=["Data", "MC"],
color=["red", "b"],
)
med_bkg = np.median(n[0])
mask = n[0] > med_bkg
over = n[0][mask]
sum_over = np.sum(over - med_bkg)
title = (season_dict["Name"] + " " + [
"upgoing",
"downgoing",
"upgoing [Log(E) > " + str(cut) + "]",
"downgoing [Log(E) > " + str(cut) + "]",
][i])
plt.axhline(
med_bkg,
linestyle="--",
color="k",
linewidth=0.5,
)
message = "{0:.2f}".format(sum_over)
plt.annotate(message,
xy=(0.05, 0.9),
xycoords="axes fraction",
color="red")
mids = (edges[:-1] + edges[1:]) / 2.0
fills = np.array(n[0])
fills[~mask] = med_bkg
ax1.fill_between(mids, fills, med_bkg, facecolor="red", alpha=0.5)
ax1.set_ylim(ymin=0.0)
plt.title(title)
plt.legend()
plt.xlabel("Azimuth")
plt.ylabel("Fraction of Total")
ax2 = plt.subplot2grid((4, 1), (3, 0),
colspan=3,
rowspan=1,
sharex=ax1)
plt.plot(mids, n[1] / n[0], color="orange")
plt.axhline(1.0, linestyle="--", color="k")
plt.ylabel("Ratio (MC/Data)")
plt.xlabel("Azimuth (rad)")
xticklabels = ax1.get_xticklabels()
plt.setp(xticklabels, visible=False)
yticklabels = ax1.get_yticklabels()
plt.setp(yticklabels[0], visible=False)
plt.subplots_adjust(hspace=0.001)
# ax2.set_xlim(0.0, 2*np.pi)
plt.savefig(root + title + " 1D.pdf")
plt.close()
of IC86_1 into a format useable for science with flarestack. The data files
themselves are duplicates of those provided at:
https://icecube.wisc.edu/science/data/PS-IC86-2011
"""
import os
import numpy as np
import csv
import pickle
from flarestack.data.icecube.ps_tracks.ps_v002_p01 import IC86_1_dict
from flarestack.icecube_utils.dataset_loader import data_loader
from flarestack.shared import input_dir
ref_data = data_loader(IC86_1_dict["exp_path"])
src_dir = os.path.dirname(os.path.realpath(__file__)) + "/"
data_dir = src_dir + "raw_data/"
output_path = input_dir + "public_datasets/"
data_dtype = np.dtype([('ra', np.float), ('dec', np.float), ('logE', np.float),
('sigma', np.float), ('time', np.float),
('sinDec', np.float)])
def parse_numpy_dataset():
"""Function to parse the .txt file of events into a numpy format
readable by flarestack, which is the saved in the products/ subdirectory.
"""
def load_data(self, path, **kwargs):
return data_loader(path, **kwargs)
def get_mc(floor_dict):
return data_loader(floor_dict["season"]["mc_path"])
def get_data(season):
mc = data_loader(season["mc_path"], floor=True)
x = np.degrees(angular_distance(
mc["ra"], mc["dec"], mc["trueRa"], mc["trueDec"]))
y = np.degrees(mc["sigma"]) * 1.177
return mc, x, y
import os
import matplotlib.pyplot as plt
from flarestack.data.icecube.northern_tracks.nt_v002_p01 import diffuse_8year
from flarestack.shared import plots_dir
from flarestack.icecube_utils.dataset_loader import data_loader, grl_loader
data_rate_dir = plots_dir + "data_rate/"
# for season in txs_sample_v2 + gfu_v002_p01:
for season in diffuse_8year:
data = data_loader(season["exp_path"])[season["MJD Time Key"]]
print(data_loader(season["exp_path"]).dtype.names)
sample_dir = data_rate_dir + season["Data Sample"] + "/"
if not os.path.isdir(sample_dir):
os.makedirs(sample_dir)
grl = grl_loader(season)
print(grl.dtype.names)
print(min(grl["start"]), max(grl["stop"]))
print(min(grl["run"]), max(grl["run"]))
print(grl[-10:])
plt.figure()
plt.hist(data, bins=50, histtype="stepfilled")
savepath = sample_dir + season["Name"] + ".pdf"
frac_s = np.sum(south < min_angular_err) / float(len(south))
fracs_south.append(frac_s)
print("North", np.sum(north < min_angular_err), float(len(north)), \
frac_n)
plt.figure()
plt.plot(means, fracs_north, label="Northern Sky")
plt.plot(means, fracs_south, label="Southern Sky")
plt.xlabel("log(E)")
plt.ylabel(r"Fraction of events sub-0.2$^\circ$ error")
plt.legend()
plt.savefig(sample_dir + season["Name"] + "_energy.pdf")
plt.close()
exp = data_loader(season["exp_path"])
mask = exp["sigma"] < min_angular_err
plt.figure()
plt.hist(np.degrees(exp["sigma"]), bins=100)
plt.yscale("log")
plt.savefig(sample_dir + season["Name"] + "_corrected.pdf")
plt.close()
print()
print(season["Name"], "(read in with data_loader)")
print(np.sum(mask), "events have an angular error less than", end=' ')
print(np.degrees(min_angular_err), "degrees")
