finite_element_method

A small Rust crate that collects finite‑element method (FEM) building blocks used as the foundation for my browser‑based FEA project fea_app.

This crate is intentionally educational and practical. It focuses on clarity and explicit data flow (elements → assembly → global system), rather than on being a general‑purpose or feature‑complete FEM framework.

---

Overview

The goal of this crate is to make the core ideas of the finite‑element method tangible in code:

• how elements are defined and parameterized,
• how local stiffness contributions are assembled,
• how the global system is prepared for downstream solvers.

It is designed to be read, experimented with, and extended.

---

What’s inside

• Mesh and data model
  • nodes, element connectivity, and basic material / section properties
• Element implementations
  • a small set of classic structural elements (e.g. truss / beam / plate variants)
• Assembly helpers
  • routines to compute and assemble global stiffness contributions
• Geometry utilities
  • small helpers (e.g. planar convex hull) used by element logic

Solvers live in separate repositories:

• Iterative methods (CG / PCG): iterative_solvers
• Direct solver experiments: colsol
• GPU compute & visualization pipeline: implemented in fea_app

---

Design notes

• Readability first
  This code is written to understand FEM internals end‑to‑end, not to hide them behind abstractions.

• Explicit data flow
  Types and functions aim to make mathematical structure and data movement visible.

• Scoped ambition
  This is not a production FEM package. It intentionally omits features such as: nonlinear analysis, contact, advanced integration schemes, or a large element catalog.

---

Documentation

Some background notes and design explanations live in the docs/ directory:

• Elements axes direction
• Element types

---

Relationship to fea_app

fea_app integrates these building blocks into a complete pipeline:

mesh input → FEM assembly → solver → visualization (WebGL / WebGPU)

This crate provides the FEM side (elements and assembly).
Companion crates provide solvers and GPU compute experiments.

---

Status

Active learning and exploration.
The API may evolve as the broader project grows.

---

License

MIT OR Apache‑2.0 (see Cargo.toml).