- Work on the presentation of your paper.
Agenda
- Share the work you have done on your presentation (10 mins ea.)
- Focus on Conclusion
- Show us the structure of your paper
- Brief example of how to set up workflows in PaperPile
- Discuss reproducibility.
- Results need to be generated within your notebook
- (or as a consequence of executing your notebook).
- Single script should reproduce all outputs:
reproduce.sh
- "It works on my computer!" → reproducibility with binder.
- There different version of python
- different versions of packages we use (pandas, etc.)
- mybinder is a tool that we can specify the specific versions of these things and generate a link that some one can click on and open a notebook with these things installed.
- Create a file of YOUR environment:
- In your paper directory
- Create a folder: binder
mkdir binder cd binder
- create the environment file: This will create a file with all installed packages including versions
conda env export -f environment.yml
- Create a reproduce.sh file:
- copy and paste the file from AMonninger
- Replace jupyter notebook names with the file names of your paper
- give the script permission to be executable using your terminal
ls -las
(shows if it is executable)chmod u+x reproduce.sh
ls -las
(reproduce.sh file should be green now)
- run the file using
/bin/bash reproduce.sh
- NOTE: I build the paper using myst and save it as a pdf. To specify options here, adjust your myst.yml file
- Sync your GitHub fork to assure that binder can access all information
- Results need to be generated within your notebook
- In class work time.
- Bring your questions.
- Receive help with Python, Jupyter,
build-jb
,myst-start
LitMaps, Paperpile, … anything!
- Q&A About the final draft.
Homework
-
Work on your Final Draft
-
Revisit your PaperPile
- denote with a tag the citations that you actually reference in your draft
- Set up a workflow that will automatically update your bibfile when you change the papers you are referencing
-
Please create a final Litmap with all the papers you used(should be > 10 and < 100). Make sure to have tagging done right(a solid number of tags + show which papers address multiple tags in a single paper)