Auto-eD (https://autoed.herokuapp.com) is a web application to help students learn the basics behind automatic differentiation by helping them to visualize the computational graph structure which underlies these computations in forward and reverse mode.
An automatic differentiation unit that teaches the basics of automatic differentiation with this tool to help with visualization can be found on Read the Docs.
Thus, this repository offers three learning levels for students seeking to learn more about automatic differentiation and its computational structure:
- Students who are completely new to automatic differentiation should work through the unit on Read the Docs, which is complemented by the web app.
- Students who have been introduced to the basic concepts behind automatic differentiation and who are seeking a tool to support their understanding should use the web app directly.
- Students who are advanced programmers and want to advance their understanding of automatic differentiation through code may download the full package. These students should refer to DeveloperDocumentation.ipynb for more details.
This work is an extension of a final project by Lindsey Brown, Xinyue Wang, and Kevin Yoon for Harvard IACS CS207: Systems Development for Computational Science.