Asimov: a toolkit for managing and automating parameter estimation

Lightweight, but build for scale

Asimov began life as a project for managing very large analyses from gravitational wave searches, but it's just as happy working on small projects.

Get started by following the installation documentation.

Using pip

Asimov is included in the python package index, making it easy to install with a single command on most systems

$ pip install asimov

We recommend always installing inside a virtual environment.

Using conda

Asimov is also available from conda-forge, and can be installed in conda environments.

$ conda install -c conda-forge asimov

Analyse a gravitational wave

Get started on your own gravitational wave analysis by following the getting started guide.

Create new analyses with just a few lines of configuration

We often want to run multiple analyses on the same data with just a few settings changed.

Asimov makes this easy, and makes sure that the analyses are internally consistent, thanks to its analysis blueprints feature.

Analysis sets can be saved and applied to multiple subjects, for example, the entire set of analyses required to reproduce a GWTC catalogue analysis can be applied to an event with a single command.

kind: analysis
name: get-data
pipeline: gwdata
download:
  - frames

---
kind: analysis
name: noise-analysis
pipeline: bayeswave
needs: get-data

---
kind: analysis
name: main-analysis
pipeline: bilby
needs: noise-analysis

Asimov takes care of setting up each pipeline consistently, so that you don't need to learn how the configurations for each analysis work to set them up correctly.

Keeping things running smoothly

Asimov works with the htcondor scheduling system to keep track of your analyses.

It can produce web-pages with details of all your analyses so that they can easily be checked on the go.

Asimov will attempt to fix and restart pipelines which have stopped running, or alert you if it's not able to fix the problem.

$ asimov monitor

Easy to extend

Asimov allows you to hook-up your own analysis pipeline with just a small amount of code, allowing you to take advantage of its powerful automation features.

Visit the pipeline development guide to learn how to make your analysis work with asimov.

Rigorous and reproducable

Asimov makes it easy for you to distribute exact details of your analyses, making rerunning and reproducing your analyses straightforward. Simply include a copy of your asimov blueprints in your data release to allow for analyses to be repeated and remixed.

A full recipe book

Explore a set of curated analysis blueprints for supported analysis pipelines in our recipe book. You no longer need to spend all day reading the documentation for a new analysis method, just use one of our recipes and start analysing!

Manage and automate complex scientific workflows with Asimov