parameters_impact.py
5.46 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
#!/bin/python
from numpy import zeros, size, nditer, average, savetxt, set_printoptions
from src.read import select_events
from src.analytic import degre, yr
set_printoptions(precision=2)
Redshifts=["0.04","0.14","0.4","1","2"]
powerlaw_index=1.2
###############################################################################
print "Compute mean obs vs EGMF strength ..."
EGMFs=["12","13","14","15","16","17","18","19","20","21","22"]
theta_mean = zeros((size(Redshifts)+1,size(EGMFs)+1))
theta_mean1 = theta_mean.copy()
theta_mean2 = theta_mean.copy()
dt_mean = theta_mean.copy()
dt_mean1 = theta_mean.copy()
dt_mean2 = theta_mean.copy()
ratio_gen = theta_mean.copy()
def compute_mean_delay_and_theta(i,j):
if i==0:
theta_mean[i,0] = 0
theta_mean1[i,0]= 0
theta_mean2[i,0]= 0
theta_mean[i,j] = 10**(-float(EGMFs[j-1]))
theta_mean1[i,j]= 10**(-float(EGMFs[j-1]))
theta_mean2[i,j]= 10**(-float(EGMFs[j-1]))
dt_mean[i,0] = 0
dt_mean1[i,0]= 0
dt_mean2[i,0]= 0
dt_mean[i,j] = 10**(-float(EGMFs[j-1]))
dt_mean1[i,j]= 10**(-float(EGMFs[j-1]))
dt_mean2[i,j]= 10**(-float(EGMFs[j-1]))
ratio_gen[i,0]= 0
ratio_gen[i,j]= 10**(-float(EGMFs[j-1]))
else:
theta_mean[i,0] = Redshifts[i-1]
theta_mean1[i,0] = Redshifts[i-1]
theta_mean2[i,0] = Redshifts[i-1]
dt_mean[i,0] = Redshifts[i-1]
dt_mean1[i,0] = Redshifts[i-1]
dt_mean2[i,0] = Redshifts[i-1]
ratio_gen[i,0] = Redshifts[i-1]
fileId="Simulations/z="+Redshifts[i-1]+"/EGMF=1e-"+EGMFs[j-1]+"G"
weight, energy, time, theta, phi, Esource, gen = select_events(fileId,Erange=[1,1e3],powerlaw_index=1.2)
dt_mean[i,j]=average(time,weights=weight)/yr
theta_mean[i,j]=average(theta,weights=weight)/degre
cond = (gen==2)
nb_gen1 = sum(weight[cond])
dt_mean1[i,j]=average(time[cond],weights=weight[cond])/yr
theta_mean1[i,j]=average(theta[cond],weights=weight[cond])/degre
cond = (gen==4)
nb_gen2 = sum(weight[cond])
dt_mean2[i,j]=average(time[cond],weights=weight[cond])/yr
theta_mean2[i,j]=average(theta[cond],weights=weight[cond])/degre
ratio_gen[i,j-1] = nb_gen2/nb_gen1
it=nditer(theta_mean1, flags=['multi_index'], op_flags=['readwrite'])
while not it.finished:
compute_mean_delay_and_theta(it.multi_index[0],it.multi_index[1])
it.iternext()
savetxt("Results/theta_mean_vs_EGMF.dat" ,theta_mean ,fmt='%1.4e')
savetxt("Results/theta_mean_vs_EGMF-gen=1.dat",theta_mean1,fmt='%1.4e')
savetxt("Results/theta_mean_vs_EGMF-gen=2.dat",theta_mean2,fmt='%1.4e')
savetxt("Results/delay_mean_vs_EGMF.dat" ,dt_mean ,fmt='%1.4e')
savetxt("Results/delay_mean_vs_EGMF-gen=1.dat",dt_mean1 ,fmt='%1.4e')
savetxt("Results/delay_mean_vs_EGMF-gen=2.dat",dt_mean2 ,fmt='%1.4e')
savetxt("Results/ratio_gen_EGMF.dat" ,ratio_gen ,fmt='%1.4e')
print(ratio_gen)
###############################################################################
print "Compute mean obs vs EGMF coherence length ..."
L_Bs=["0.001","0.010","0.100","1","10","100","1000"]
theta_mean=zeros((size(Redshifts)+1,size(L_Bs)+1))
dt_mean=theta_mean.copy()
def compute_mean_delay_and_theta(i,j):
if i==0:
theta_mean[i,0]=0
dt_mean[i,0]=0
theta_mean[i,j]=L_Bs[j-1]
dt_mean[i,j]=L_Bs[j-1]
else:
theta_mean[i,0] = Redshifts[i-1]
dt_mean[i,0] = Redshifts[i-1]
if (L_Bs[j-1]=="1"):
fileId="Simulations/z="+Redshifts[i-1]+"/EGMF=1e-15G"
else:
fileId="Simulations/z="+Redshifts[i-1]+"/lambda_B="+L_Bs[j-1]+"Mpc"
weight, energy, time, theta, phi, Esource, generation = select_events(fileId,Erange=[1,1e3],powerlaw_index=1.2)
dt_mean[i,j]=average(time,weights=weight)/yr
theta_mean[i,j]=average(theta,weights=weight)/degre
it=nditer(theta_mean, flags=['multi_index'], op_flags=['readwrite'])
while not it.finished:
compute_mean_delay_and_theta(it.multi_index[0],it.multi_index[1])
it.iternext()
set_printoptions(precision=2)
print(theta_mean)
print "========================================"
print(dt_mean)
savetxt("Results/theta_mean_vs_LB.dat",theta_mean,fmt='%1.4e')
savetxt("Results/delay_mean_vs_LB.dat",dt_mean,fmt='%1.4e')
###############################################################################
print "Compute mean obs vs jet opening angle ..."
opening_angle = [90,75,60,45,30,15,10,5,2,1,0.7,0.5,0.1,1e-2] # degres
def compute_theta_mean(i,j):
if i==0:
if j==0:
theta_mean[i,j]=0
dt_mean[i,0]=0
else:
theta_mean[i,j]=opening_angle[j-1]
dt_mean[i,j]=opening_angle[j-1]
else:
if j==0:
theta_mean[i,j]=float(Redshifts[i-1])
else:
fileId="Simulations/z="+Redshifts[i-1]+"/EGMF=1e-15G"
weight, energy, time, theta, phi, Esource, gen = select_events(fileId,Erange=[1,1e3],powerlaw_index=1.2,tjet=opening_angle[j-1])
theta_mean[i,j]=average(theta,weights=weight)/degre
dt_mean[i,j]=average(time,weights=weight)/yr
theta_mean=zeros((size(Redshifts)+1,size(opening_angle)+1))
dt_mean=theta_mean.copy()
it=nditer(theta_mean, flags=['multi_index'], op_flags=['readwrite'])
while not it.finished:
compute_theta_mean(it.multi_index[0],it.multi_index[1])
it.iternext()
set_printoptions(precision=3)
print(theta_mean)
savetxt("Results/theta_mean_vs_jet_opening.dat",theta_mean,fmt='%1.4e')
savetxt("Results/delay_mean_vs_jet_opening.dat",dt_mean,fmt='%1.4e')