Unit Testing

In case you have not prepared your code to implement unit test, please use the following example code for shear estimation.

Download and Install

git clone https://github.com/mr-superonion/meaShear.git
cd meaShear
pip install .

About the code

This code implements a simple estimator of the weak gravitational lensing shear -- i.e., the distortion of light from distant source galaxies as they pass through the large-scale structure of the Universe on their way to us. The deflection of light rays by the matter distribution causes changes in galaxy brightness, size, and shape. Coherent changes in shape can be used to measure those deflections and learn about the distribution of dark matter in the Universe.

The code in this repository implements one step in that process: measurement of galaxy shapes of images, after accounting for blurring of the images due to the light passing through the atmosphere and telescope optics.

shear estimation

ring test

• Simulate a group galaxies with intrinsic ellipticity average to zero and distort them by known shear to test the shear estimation.

import meaShear
import numpy as np
import astropy.io.fits as pyfits
import numpy.lib.recfunctions as rfn
# Read PSF image
psfData=pyfits.getdata('data/psf_test.fits')
# Read GAL image
print('Reading the simulated image distorted by g_1= 0.02, g2=0.00')
galDatAll=pyfits.getdata('data/gal_test.fits')
imgList=[galDatAll[i*64:(i+1)*64,0:64] for i in range(4)]

ellRes=[]
for i in range(4):
    # input image and PSF to estimate e and R
    eR=meaShear.measureEllip(imgList[i],psfData,scale_par=0.85,weight_par=100.)
    ellRes.append()

ellRes =   rfn.stack_arrays(ellRes,usemask=False)

# average over response to estimate shear
g1_est=np.average(ellRes['fpfs_e1'])/np.average(ellRes['fpfs_RE'])
g2_est=np.average(ellRes['fpfs_e2'])/np.average(ellRes['fpfs_RE'])
print('estimated shear is: g1= %.5f, g2= %.5f' %(g1_est,g2_est))

Test Development List

• Code performance (accuracy)
  • Use ring test (45 degree rotated galaxies) to test the accuracy of the estimator on noiseless galaxies
• Inputs Control (interaction with users)
  • type control: Inputs galaxies and PSF images are ndarrays
  • type control: The tuning parameters are real
  • value control: If the input image has values that are problematic, like NaN or Inf, the code does something reasonable
  • value control: the tuning parameters in reasonable ranges (0<scale_par<1), (weight_par>0)

Useful functions

np.testing.assert_almost_equal(out_come, expectation, 5) # 1e-5 accuracy
assert_raises(TypeError,your_function,input_1,input_2,input_3,input_4)
assert_raises(ValueError,your_function,input_1,input_2,input_3,input_4)

Example

You can find a testing example here, and you can run the test by

pip install nose

cd tests
nosetests -w tests -v