waxMorph

Contents

waxMorph#

A differentiable cell-based framework for three-dimensional morphogenesis.

Cells build tissues through local exchanges of force and information. waxMorph represents tissue as interacting spheroidal cells and couples a prescribed forward simulator with a differentiable learned inverse emulator. The two share a state representation, and each has its own dynamics.

waxMorph pipeline: forward simulation and inverse learning of shape assembly
waxMorph represents tissue as interacting polarized spheroids, runs forward and inverse models of morphogenesis on that representation, and recovers spatially organized latent signals.

Explore#

Installation

Install the package and pick the simulation, learning, or jax extras for your workflow.

Installation
Tutorials

End-to-end notebooks: run a forward simulation and train a learned emulator from a mesh pair.

Tutorials
Gallery

Play reconstructed morphogenesis trajectories in an interactive 3D viewer, with a timeline and a free-flying camera.

Gallery
How-to guides

Task-focused recipes for people who already know waxMorph and want to get something specific done.

How-to guides
Explanation

The modelling choices: cell state, locality, differentiable physics, forward and inverse modes.

Explanation
API reference

Public modules with PyTorch defaults and explicit JAX/Equinox counterparts.

Reference
Contributing

Development setup, testing across both backends, and the conventions your changes should follow.

Contributing