MLy_Workbench - VasSkliris/mly GitHub Wiki

This package depends a lot on the right files being at the right position and sometimes with the right name. To make that easier for the user I introduced MLy_Workbench filesystem which is just a preset of empty files in which I ask the user to work in. If you try to work outside of it, there might be path errors. To start working with mly you should go to a desired directory open a python shell and type:

from mly import CreateMLyWorkbench
CreateMLyWorkbench()

This will create a file called MLy_Workbench. The structure of that file will be as shown bellow:

MLy_Workbench

• datasets
  • burst
  • cbc
  • noise
    • optimal
    • real
    • sudo_real
• injections
  • bursts
  • cbcs
• ligo_data
  • 2048 (or any other sample frequency you want)
• trainings

datasets have all the datasets that you will use to train networks. To create the datasets you will use functions provided with mly, for more information about that visit Generators page.

injections has all the injection files ready to used for generation of datasets. Injections are not provided with this package. For now it is up to the user to create them. For more information on how to do that go to Injections page.

ligo_data includes real LIGO and Virgo data that you will use for generation of datasets or false alarm testing of the networks.

trainings is a directory that will include the subdirectories of all the networks you create and train. It is suggested to the user to create different directory for each network.

After you create MLy_Workbench you are ready to create your datasets, but first let's make sure you have injections.