Usage¶
Qanat is meant to be used as a command-line tool. The main entry point is the qanat command which takes an argument specifying the action to be performed. The following actions are supported:
qanat init: - creates a new qanat project in the specified directory
qanat dataset: - dataset level operations
qanat experiment: - experiment level operations
qanat status: - prints the status of the current qanat project
qanat config: - operations config of qanat project
qanat cache: - operations on cache of qanat project
Assumptions¶
Qanat assumes a few thing in the way you organise your workflow and your experiments:
You want to work from the terminal mostly.
You are comfortable with the command-line and shell scripting.
You are able to split your experiment workflow into 3 steps:
executable preparation: A script responsible for the execution of the experiment (must handle pre-processing, execution and post-processing tasks)
runner choice: A way to run the executable on a machine (local, htcondor, slurm:todo)
analysis: One or several scripts that analyse the results of the experiment and produce a summary of the results.
You give Qanat the reponsibility for launching the experiments and analysing the results according to your runner choice. The results are storted in a directory created by Qanat.
You have access to your datasets via a shared filesystem that is accessible from all the machines you want to run your experiments on. (Later will be added option to mount on the fly).
The experiments are specified by executables which accept command-line arguments and produce output files that will be later analysed thanks to other scripts.
You are able to specify the parameters of your experiments in a configuration file or through command-line arguments.
Concepts and Vocabulary¶
Qanat is based on a few concepts that are important to understand in order to use it efficiently. See the glossar for a list of terms and their definitions: