Page MenuHomeHEPForge

TesterFunctions.py
No OneTemporary

TesterFunctions.py

#!usr/bin/env python
import os, sys
import numpy as np
import scipy.stats as sp
import math
global mu_test
mu_test=1
def Var_mu_comb(b_count,s_count,db_count, ds_count):
#Construct the inverse variance matrix from expected vals of the 2nd derivatives of the Log Likelihood
Var_matrix_inv=np.zeros([(len(b_count)+len(s_count)+1),(len(b_count)+len(s_count)+1)])
#add exception handling if s=/=b
#loop over all counts
for i in range(0, len(b_count)+len(s_count)):
# if mu_test*s_count[i]+b_count[i] == 0:
# print 'g'
#mu mu
#Var_matrix_inv[0,0] += s_count[i]**2/(mu_test*s_count[i]+b_count[i])
#mu b_i/b_i mu
if i < len(b_count):
Var_matrix_inv[0,0] += s_count[i]**2/(mu_test*s_count[i]+b_count[i])
Var_matrix_inv[i+1,0]=Var_matrix_inv[0,i+1] = s_count[i]/(mu_test*s_count[i]+b_count[i])
if db_count[i]**2 > 0.0:
Var_matrix_inv[i+1,i+1]=1/(mu_test*s_count[i]+b_count[i]) + 1/db_count[i]**2
else:
Var_matrix_inv[i+1,i+1]=1/(mu_test*s_count[i]+b_count[i])
if i>=(len(b_count)):
Var_matrix_inv[i+1,0]=Var_matrix_inv[0,i+1] = (mu_test*s_count[i-len(b_count)])/(mu_test*s_count[i-len(b_count)]+b_count[i-len(b_count)])
if ds_count[i-len(b_count)]**2 >0.0:
Var_matrix_inv[i+1,i+1]=(mu_test**2)/(mu_test*s_count[i-len(b_count)]+b_count[i-len(b_count)]) + 1/ds_count[i-len(b_count)]**2
else:
Var_matrix_inv[i+1,i+1]=(mu_test**2)/(mu_test*s_count[i-len(b_count)]+b_count[i-len(b_count)])
if i < len(s_count):
Var_matrix_inv[len(b_count)+1+i,i+1] = Var_matrix_inv[i+1,len(b_count)+1+i] = mu_test/(mu_test*s_count[i]+b_count[i])
if np.linalg.det(Var_matrix_inv) == 0:
Var_matrix = np.zeros([(len(b_count)+1),(len(b_count)+1)])
#Invert and return it
else:
Var_matrix = np.linalg.inv(Var_matrix_inv)
return Var_matrix
def confLevel(sigCount, bgCount, bgErr, sgErr):
varMat= Var_mu_comb(bgCount,sigCount,bgErr, sgErr)[0,0]
# q_mu_a = qMu_Asimov(mu_test,bgCount,sigCount,bgErr)
mu_hat = 0
if varMat ==0:
return 0
else:
q_mu=0
p_val=0
q_mu = (mu_test-mu_hat)**2/(varMat)
if 0 < q_mu <= (mu_test**2)/(varMat):
p_val=sp.halfnorm.sf(np.sqrt(q_mu))
elif q_mu > (mu_test**2)/(varMat):
p_val=sp.halfnorm.sf( (q_mu + (mu_test**2/varMat))/(2*mu_test/(np.sqrt(varMat))) )
return float('%10.6f' % float(1-p_val))

File Metadata

Mime Type
text/x-python
Expires
Sat, Dec 21, 6:43 PM (9 m, 46 s)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
4023854
Default Alt Text
TesterFunctions.py (2 KB)

Event Timeline