notebook.community

Edit and run



In [1]:

    
import numpy as np
import matplotlib.pyplot as plt



In [2]:

    
import radprofile as RP



In [3]:

    
RPFILE = 'test_profile'



In [4]:

    
profile = RP.get_profile( RPFILE )



In [5]:

    
plt.errorbar(profile.rmid, profile.surbri, xerr=0.5*(profile.rright-profile.rleft), \
             yerr=profile.surbri_err, \
             marker='', capsize=0, ls='', color='k', lw=1, alpha=0.3)
plt.loglog()
plt.xlim(0.3,300.0)

plt.title('Test that profile was read properly', color='k')









    Out[5]:





<matplotlib.text.Text at 0x10c2b9e50>



In [6]:

    
RP.plot_profile(profile, color='b', ls='', capsize=0)
plt.loglog()
plt.xlim(0.3,300.0)

plt.title('Test radprofile.plot_profile function', color='b')









    Out[6]:





<matplotlib.text.Text at 0x10c2f6390>

Test error propagation -- addition



In [7]:

    
ADD_ERR = 0.2

prop_add_result = np.sqrt(profile.surbri_err**2 + ADD_ERR**2)



In [8]:

    
profile.minus(0.5, 0.2) # subtract 0.4 +/- 0.2



In [9]:

    
# The printed result should be zero
print np.sum(profile.surbri_err - prop_add_result)

0.0

Try out log-min error bars



In [10]:

    
RP.plot_profile(profile, logmin=0.1, color='r', ls='', capsize=0)
plt.loglog()
plt.xlim(0.3,300.0)
plt.ylim(0.1,1.e4)

plt.title('Test log-min error bars', color='r')









    



Utilizing logmin error bars






    Out[10]:





<matplotlib.text.Text at 0x10c79a7d0>

Test error propagation -- multiply and divide



In [11]:

    
## Multiplication
MULT_FAC = 1.0
MULT_ERR = 0.5

prop_mult_result = np.sqrt( MULT_FAC**2 * profile.surbri_err**2 + profile.surbri**2 * MULT_ERR**2 )
#prop_mult_result = profile.surbri*MULT_FAC * \
#    np.sqrt( (profile.surbri_err/profile.surbri)**2 + (MULT_ERR/MULT_FAC)**2 )



In [12]:

    
profile.multiply(MULT_FAC, MULT_ERR)



In [13]:

    
# The printed result should read zero
print np.sum(profile.surbri_err - prop_mult_result)









    



-1.40443212615e-14

Close enough??



In [14]:

    
print prop_mult_result









    



[  8.82956797e+02   1.69714034e+03   1.15694420e+03   6.45567335e+02
   3.13833167e+02   1.58432100e+02   9.93252224e+01   5.86590579e+01
   3.20940876e+01   2.04300922e+01   1.35852844e+01   7.45040012e+00
   4.37159453e+00   3.43615240e+00   2.25467930e+00   1.33152423e+00
   8.43097181e-01   5.10306774e-01   2.99473383e-01   9.33753105e-02]



In [15]:

    
print profile.surbri_err









    



[  8.82956797e+02   1.69714034e+03   1.15694420e+03   6.45567335e+02
   3.13833167e+02   1.58432100e+02   9.93252224e+01   5.86590579e+01
   3.20940876e+01   2.04300922e+01   1.35852844e+01   7.45040012e+00
   4.37159453e+00   3.43615240e+00   2.25467930e+00   1.33152423e+00
   8.43097181e-01   5.10306774e-01   2.99473383e-01   9.33753105e-02]

Yes, looks like a numerical thing



In [16]:

    
## Division
DIV_FAC = 1.0
DIV_ERR = 0.1

prop_div_result = np.sqrt(profile.surbri_err**2 + (profile.surbri/DIV_FAC)**2*DIV_ERR**2) / DIV_FAC



In [17]:

    
profile.divide(DIV_FAC, DIV_ERR)



In [18]:

    
# The printed result should be zero
print np.sum(profile.surbri_err - prop_div_result)

0.0