Seasons Of Code

Photometric Redshift - Kumar Ayush

WnCC - Seasons of Code

Seasons of Code is a programme launched by WnCC along the lines of the Google Summer of Code. It provides one with an opprtunity to learn and participate in a variety of interesting projects under the mentorship of the very best in our institute.

List of Projects

Photometric Redshift

Photometric Redshift

A scientific computing project aimed at calculating redshifts of extra-galactic objects from photometric data.

Redshifts caluclated from spectroscopic data is accurate but it is time expensive to measure the spectrum. There is a correlation between photometric data which is easier and faster to measure and the redshift which can be used to calculate photometric redshifts. We are currently trying to write a neural network in Keras to compute photometric z from photometric magnitudes.

Week 1

  • Make SkyServer accounts
  • Get familiar with SQL
  • Look into how CasJobs work
  • Look into DR12 data schema

Week 2

  • Get data from DR12.SpecPhotoAll
  • Filtering bad data out
  • Trying K-nearest neighbour training from scikit learn
  • Try simple Neural Network Models on Keras
  • Single hidden layer with mse loss and RMSprop optimizer works average, with sigmoid activation on last layer
  • Our inputs are psf_Mag_u/g/i/r/z and/or their linear combinations but all of them give the same output
  • We suspect we still have a lot of bad data in various form and heavier pre-processing is needed