SETI@home banner


Candidate signals from a distant transmitter should get stronger and then weaker as the telescope moves over that point in the sky. Specifically, the power should increase and then decrease with a bell shaped curve (a gaussian curve). SETI@home clients search for this characteristic shape.

In these plots, the red line indicates the data gathered at the telescope. The yellow (smooth) line traces the "best fit gaussian" for the data set. The lower the Chi Sqr, the better the fit.

The first plot shows the SETI@home client program detecting a gaussian fit to a very strong signal; this test signal was artifically injected to insure the hardware and software were working correctly.

The second plot shows a gaussian fit found by a client in a typical work unit. The work unit contains only noise (no signal is present), but noise will occasionally look like a gaussian just by chance.

The distribution of many gaussian fits, plotted from a typical day's data: For each gaussian detected by a client, this graph plots that gaussian's peak power and chi square (chi square is a a measurement of how well the gaussian curve fits the data). The most interesting candidates are those in the lower right corner of the plot, with a low chi square (good fit) and a high power. In this case, there are no candidates with good scores, because this day's data is dominated by noise and contains very little radio frequency interference. The client programs do not report Gaussians with chi square fit greater than 10 (these are poor fits).

This graph plots the number of spike detections at each frequency. Almost all of these detections are from noise. The three peaks at 1419, 1420 and 1421 MHz are due to test signals which are constantly injected to make sure the system is working. The fourth peak is due to radio frequency interference.

SETI@home looks for signals at 15 different bandwidths, each twice as coarse as the previous one. This plot shows the number of signals detected as a function of bandwidth.