vizopt¶

vizopt is a mathematical optimization library for data visualization. It provides a general framework for defining and solving layout optimization problems — such as bubble layouts and label placement — using JAX for automatic differentiation and JIT compilation.

Features¶

• General optimization framework — define multi-objective loss functions from composable terms
• Gradient descent via Adam — efficient JAX-based optimization with JIT compilation
• Bubble layout — circular node layouts with hierarchical inclusion constraints
• NetworkX integration — works directly with NetworkX graphs
• Pydantic validation — optional input validation for problem templates

Examples¶

See the examples gallery for worked examples.

Quick Example¶

Installation¶

pip install vizopt

Requires Python 3.13+.