Looking more into deep learning and missing power on large scales

Our Dark Sector group at JPL has recently started an initiative to expand our skillsets towards deep learning, eventually coming up with new ideas to tackle the challenges that we’ll face in the current and upcoming generation of cosmological surveys.

I’ve thus started to dig into the deep learning book by Francois Chollet, which I can highly recommend. It also comes with a set of companion Python notebooks, which are a great addition to the material.

On the front of component separation techniques for cosmology surveys such as Planck, we’ve encountered a problem while only doing the component separation on smaller patches of then sky and then ’tiling’ the full sky. In the power spectrum level, this procedure removes power from the faint cosmology background field at scales larger than our patch size. We’re currently attempting to verify whether we can measure the effective transfer function of this effect and then simply correct for this.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s