AutoDiffpyyy 1.0.0

Package for automatic differentiation by Harvard AC207 students

AC207 Final Project - AutoDiffpy

Team03 Members

Member	Email
Jinglun Gao	[e-mail]
Chuqin Zhao	[e-mail]
Chao Wang	[e-mail]
Jiaping Lin	[e-mail]

Introduction

The AutoDiffpy Python package is the project topic of the Harvard AC207 Final Project in Fall 2022, which we writed a python automatic differentiation library. AD is a very broad area spanning computer science and mathematics with applications in fields across science and engineering. Thus, we have build this package to support getting derivatives of large scalar functions and vector functions. Particularly, AutoDiffpy could perform forward automatic differentiation that takes dual numbers to compute derivatives sequentially, and could also perform reverse automatic differentiation that computes accurate derivatives to solve problems.

Simple Install

To install the AutoDiffpy, user could run the following command in the terminal.

!pip install AutoDiffpy

Documentation

The documentation for the AutoDiffpy package can be found here.