Jupyter Binder Use Case Reproducibility Open Science Best Practice

Scenario

Jane has written a (math) paper based on experimentations. She would like anyone to be able to reproduce her calculations.

a binder logbook screenshot

Suggestion of solution

  1. Describe the experimentation as Jupyter notebooks, mixing prose, code, and outputs (think of them as logbooks);

  2. Publish them on a public repository (e.g. on GitHub);

  3. Make that repository Binder-ready by describing the software stack required; for details, see the Binder documentation;

  4. Bonus: make the paper itself active

    To do: explore using e.g. latexml+thebe?.

Some instances

To do

Time and expertise required

Assuming Jane is familiar with version control and Jupyter (basic lab skills taught at Software Carpentry, that the experiments were prepared as notebooks, and the software required is packaged (conda, debian, docker container, …), the publishing part could take two hours the first time, and half an hour later on.

What’s new since OpenDreamKit started

OpenDreamKit contribution