Skip to content

asympTotic lIkelihood Tests for daRk mAtTer sEarch

Notifications You must be signed in to change notification settings

StFroese/TITRATE

Repository files navigation

asympTotic lIkelihood-based Tests for daRk mAtTer sEarch

This package is based on the paper Asymptotic formulae for likelihood-based tests of new physics.

(your daily dose of meme)

Why does this package exist?

Well, I'm currently doing my PhD on dark matter search with Imaging Air Cherenkov Telescopes (IACTs) and during my research, I looked into a good share of DM papers... Turns out none of the ones I've read really explained how they are calculating their upper limits and most of them lack a good explanation for upper limit vs dark matter mass plots.

So to understand what's going on I went back in time...on arxiv. I thought, for sure the CERN people will know this stuff since calculating upper limits is daily business for them. I quickly found the paper mentioned above and it only took me three months to understand it.

(╯°□°)╯︵ ┻━┻

What have I learned?

  1. Statistical tests are difficult to understand
  2. The decentralized $\chi^2$-distribution is the final boss
  3. A lot of researchers use a test statistic that is physically not meaningfull (signal strength can be smaller than zero)
  4. A lot of researchers calculate the median upper limits and bands for the expected signal by using a bunch of toy MCs, which is not necessarily needed if asymptotics are valid

What does this package offer?

  1. Adding Asimov datasets to gammapy ✔️
  2. Adding test statistics to gammapy ✔️
  3. Validation of test statistics ✔️
  4. Calculation of ULs ✔️
  5. Calculation of median ULs and bands with asymptotic formulae and asimov datasets (faster than toy MCs) ✔️

Disclaimer

This is not a finished version yet and can contain drastic changes

About

asympTotic lIkelihood Tests for daRk mAtTer sEarch

Resources

Stars

Watchers

Forks

Packages

No packages published

Languages