This image performs the model-independent multiscale source search.
It analyzes a set of directions on a sphere (originally developed for the directions of neutrinos in equatorial coordinates) to detect the most pronounced overdense region.
The input can be supplied as a text file, with one line encoding one position. Each line contains right ascension, declination and, for historical reasons, four further numbers that are not evaluated. All values within a line are separated by whitespaces.
If you want to use custom input, it must be named events.txt and has to be located at /input/ within the container. A minimum of 3000 data points is highly recommended.
The output is a skymap in Hammer-Aitov-projection showing the evaluated densities with and without events. To obtain the output, a folder has to be mounted to /output within the container.
The recommended calls would therefore be something like:
docker run -v 'pwd'/outputFromMultiscale:/output km3net/multiscale:20170307
for default input or
docker run -v /absolute/path/that/contains/your/inputfile:/input -v 'pwd'/outputFromMultiscale:/output km3net/multiscale:20170307
to use your own input.
This demonstration does not compute p-values or significances, as the required amount of pseudo experiments takes a long time. The full source-code can be found at https://github.com/sgeisselsoeder/multiscale.