search

COSMIC-SETI / seticore

A high-performance implementation of some core SETI algorithms that can be included in other programs.
MIT License
0 stars 0 forks source link

Investigation of SNR #2

Open Chenoachem opened 2 weeks ago

Chenoachem commented 2 weeks ago

@NMStiegler has been investigating issue related to the SNR calculations for seticore. We are finding inconsistencies in the values and a high number of false positives based on the threshold of 8 sigma. Here we will detail the investigation on how to fix this.

Some suggestions on a general level are:

  1. Replace the calulcation for noise in seticore and seticore viewer. Low Risk - modifies current calculation but doesn’t impact other functions of seticore. 2.Replace entire SNR calculation in seticore with the logic from BLISS Medium Risk - modifies current calculation but doesn’t impact other functions of seticore. 2.5.Remove the PFB response from the spectra within seticore Medium Risk - modifies current calculation but doesn’t impact other functions of seticore.
  2. Replace seticore with BLISS High Risk - seticore runs the stamp process and beamformer. So this would all require integration.
david-macmahon commented 2 weeks ago

And there's also:

  1. Replace the SNR threshold with one more suited to the low-N statistics of short integrations

These options are not all mutually exclusive. IMHO, removing the PFB passband should definitely be implemented in seticore (in general, not just for COSMIC).