diffstep 1.3.0

A symbolic differentiation library.

diffstep

diffstep is a lightweight Python library for algorithmic differentiation.

It lets you:

• Differentiate expressions like x^2 + 3x + 2
• Evaluate gradients at specific points

Installation

Install from PyPI:

pip install diffstep

Quick Start

Basic Access (from diffstep import diffstep)

from diffstep import diffstep

diffstep("x^2 + 3x + 2")
# 2*x + 3

Advanced Access (import diffstep)

import diffstep

diffstep.differentiate("x^2 + 3x + 2")
# 2*x + 3

diffstep.gradient("x^2 + 3x + 2", 1)
# 5

diffstep.parse_expression("sin(x) + cos(x)")
# Prints a readable AST tree

Public API

• diffstep(expression: str) Main entry point. Returns the simplified symbolic derivative.

• differentiate(expression: str) Alias of diffstep().

• gradient(expression: str, x_value: float) Evaluates the derivative at a specific x value.

• normalise_expression(expression: str) Returns the normalized expression string used internally by the parser.

• parse_expression(expression: str, pretty: bool | None = None) Parses an expression to AST form. By default, it prints a pretty ASCII tree. Pass pretty=False to return the raw AST node instead.

• DiffStep(expression: str) Deprecated alias for backward compatibility.

Supported Expression Features

• Variables: x
• Constants and arithmetic: +, -, *, /, ^
• Common functions such as sin, cosec, arctan, ln, and sqrt

Project Structure

• diffstep/core/: validation, tokenization, parsing, AST nodes
• diffstep/differentiation/: differentiation rules
• diffstep/simplify/: simplification logic
• diffstep/printer/: AST-to-expression string conversion
• diffstep/pipeline.py: user-facing workflow functions

License

MIT License. See LICENSE.txt for details.