## Nebula results for SETI@home

This page shows results from Nebula, the SETI@home back end. For an explanation of the information here and how it's computed, see Nebula: Completing the SETI@home pipeline. For announcements and discussion, see the Nebula Blog.

This is a work in progress; we're still working out the scoring and RFI algorithms. These pages show the current state. Probably nothing here is an ET signal.

## Run details

Every so often, after changes to one or more of the algorithms, I run the Nebula pipeline, producing a new set of results.

Completion time of this pipeline run: 2019-06-03 20:43:03

Notes on this run:

20K birdies, 1.2M pixels: ~25M multiplets

The following generally shows the results of analyzing a subset of pixels (typically 100K out of 16M).

Multiplets   explain

Top scores

Top scores, normalized across signal types

Birdies   explain

Pixels   explain

Bookmarks   explain

You can 'bookmark' pixels, multiplets, and signals that are of interest (e.g. they look like an ET signal, or RFI removal didn't work properly).

## RFI removal

View graphs showing the distributions of signals removed by various RFI algorithms.

RFI zones (pre-Nebula)

RFI zones (Nebula)

## Snapshots

At the end of each run of the Nebula pipeline we create a 'snapshot' of the key output files. This lets us see the effects of changes in the pipeline.

• 2019-06-03 20:43:03

20K birdies, 1.2M pixels: ~25M multiplets

• 2019-05-06 13:19:34

Increase # of birdies to 20K (10K each bary, nonbary)

• 2019-04-25 23:49:21

Updated formula for time entropy

• 2019-04-24 01:18:19

Use appropriate signal duration in multiplet overlap pruning

• 2019-04-18 11:13:42

The power of a sinusoid in an FFT bin depends on where it is in the bin. Take this into account when computing birdie signals. Also, raise the power range of birdies.

• 2019-03-23 18:41:34

Limit birdie spikes to 8 per WU (this time for sure)

• 2019-03-10 13:17:59

Limit 8 spikes per WU for birdie signals, to mimic the client

• 2019-03-04 00:35:28

Generate birdie spikes in a way that reflects how the S@h client works.

• 2019-01-08 01:32:28

various bug fixes

• 2019-01-03 15:32:52

Include multiplet signals per observation; don't penalize frequency variation in non-bary multiplets.

• 2018-11-12 12:26:24

Model planetary occlusion for birdies (don't generate signals when imaginary transmitter is on far side of planet).

• 2018-11-02 22:43:38

Changed non-barycentric birdie generation. Now support planetary and nonplanetary (orbital) transmitters around a variety of planets, both HZ and non-Hz planets. Stellar sizes distributed as local mass function, HZ bounds set by luminosity.

• 2018-10-23 13:54:41

• RFI: add +/- 1220 Hz zone are 1420 MHz for all types except autocorr
• decrease nonbary birdie power range from 26-36 to 26-30
• add rotational component to drift of non-bary birdies
• 2018-10-05 14:44:37

When we find a multiplet, see if we can increase the score by sliding the window so that the multiplet is at the bottom

## Sky coverage

The part of the sky visible from the Arecibo telescope is divided into 16M 'pixels'. Here are some statistics about how many times, and for how long, SETI@home has observed each pixel.

This is based on our database of 'WU groups', each of which describes a 107-second period and contains a list of pointings (RA, dec, time) during that period. Early WU groups had one pointing record per 5 seconds. Later this was changed to 1 second, and later still to about 3 seconds.

The telescope sky direction changes at different rates:

• zero, when it's tracking
• .003 to .004 degrees/sec when it's drifting
• higher when it's slewing to a new location

This rate is called the 'angular velocity'. We're especially interested in the range from .0021 to .0105, because in that range we can look for Gaussian signals. We'll call that the 'Gaussian range'.

A 'pixel observation' is a string of consecutive pointings within one pixel. An observation is called 'Gaussian' if the angular velocity is within the Gaussian range at least once during the observation.

Here's some data about sky coverage: