User guide

graphlow is a differentiable mesh-graph library for mesh processing. It combines geometry metrics, mesh topology, and sparse graph operators behind one Python API so that the same workflow can be used with torch and phlower_tensor backends.

This guide is organized by the questions new users usually have first:

• Installation
• Quickstart
• Logging (optional)
• Core concepts
• Migrating to 0.1.x mesh API
• Examples and API reference

What you can do with graphlow

Typical workflows include:

• load a PyVista or VTK mesh into a differentiable TensorMesh

• compute geometry metrics such as volumes, normals, centroids, and quality

• build topology-derived sparse operators such as incidences, adjacency, and Laplacians

• combine geometry and graph operators in autograd-based optimization loops

If you want the shortest path to a working example, start with Quickstart. If you need package installation details first, go to Installation.