Welcome to pyFANTOM!

FANTOM: Finite-element ANalysis and Topology Optimization Module

pyFANTOM is a fast, efficient, GPU and CPU-ready general topology optimization package. Built with an object-oriented design, pyFANTOM enables you to perform finite-element based topology optimization with ease and flexibility.

What is pyFANTOM?

pyFANTOM is a general-purpose topology optimization framework designed for flexibility and performance. The package provides:

  • Dual Backend Support: Run your optimization problems on both CPU and GPU (CUDA) with the same API

  • Multiple Mesh Types: Support for structured 2D/3D meshes and unstructured meshes

  • Efficient Solvers: Including CHOLMOD, CG, GMRES, and a custom MultiGrid solver optimized for topology optimization

  • Scalability: Solve mega-voxel problems (8M+ elements) on GPUs with 16GB VRAM thanks to matrix-free multi-grid implementation

  • Flexible Physics: Built-in linear elasticity support with extensible physics models

  • Interactive Visualization: 3D interactive visualizations using K3D for exploring your optimized topologies

Key Features

High Performance

  • GPU acceleration via CUDA backend

  • Matrix-free multi-grid solver for large-scale problems

  • Efficient sparse solvers (CHOLMOD, CG, GMRES, MultiGrid)

  • Solve mega-voxel problems (8M+ elements) on 16GB GPUs

Flexible & Extensible

  • Object-oriented design for easy customization

  • Support for custom physics models

  • Multiple optimization algorithms (PGD, MMA, OC)

  • Independent components (meshes, solvers, optimizers)

Multiple Problem Types

  • Structured 2D and 3D meshes

  • Unstructured meshes via pygmsh

  • CPU and GPU backends with the same API

  • Interactive 3D visualizations

Quick Start

Ready to get started? Check out our examples:

Or jump straight to the Installation guide and Introduction Tutorial for a complete walkthrough.

A Simple Example

Here’s a quick taste of what pyFANTOM can do:

from pyFANTOM.CPU import (
    StructuredMesh2D, StructuredStiffnessKernel, CHOLMOD,
    FiniteElement, StructuredFilter2D, MinimumCompliance, PGD
)
from pyFANTOM.Physics import LinearElasticity

# Setup physics and mesh
physics = LinearElasticity(E=1.0, nu=1/3, thickness=1.0, type='PlaneStress')
mesh = StructuredMesh2D(nx=256, ny=64, lx=4.0, ly=1.0, physics=physics)

# Setup solver and finite element analysis
kernel = StructuredStiffnessKernel(mesh=mesh)
solver = CHOLMOD(kernel=kernel)
FE = FiniteElement(mesh=mesh, kernel=kernel, solver=solver)

# Apply boundary conditions and setup optimization problem
filter = StructuredFilter2D(mesh=mesh, r_min=1.5)
problem = MinimumCompliance(FE=FE, filter=filter, volume_fraction=[0.4], penalty=3.0)
optimizer = PGD(problem=problem, change_tol=1e-4, fun_tol=1e-6)

# Run optimization
for i in range(200):
    optimizer.iter()
    if optimizer.converged():
        break

# Visualize result
problem.visualize_solution()

See the Getting Started guide for the complete example with boundary conditions and detailed explanations.

Documentation Table of Contents

Getting Started

Examples

API Reference

Other Sections