mdecourse / constraint-solver

Pure python geometric constraint solver

constraint-solver

Pure python geometric constraint solver

import gcs

See gcs/solve_elements.py for a description of the mechanics behind this package. Essentially, gcs implements a constraint solver that breaks a problem into fully constrained sets of equations/variables and solves each set in order of its graph dependencies.

This algorithm was loosely inspired by some light research I did on CAD constraint solvers, but the idea to split equations/constraints into fully constrained sets was original.

Installing

This repo contains conda environment definitions that work with this package (environment.yml and environment-windows.yml).

To install on windows (remove sh if not using bash and use create-windows-env.bat):

sh create_windows-env.sh
pip install -e .

Running

A couple sample scripts exist in test/.

• test/make_constraint_graph.py: shows how to set up a solver, solve it, and view its geometry and constraint graph
• test/problem2_cstr_del_add: tests adding/deleting constraints on a sample problem. Makes an mpl animation showing the solver being updated as the constraint controlling the circle's radius changes

Sample Results

Geometry for problem2:

Constraint graph before running gcs algorithm:

Constraint graph split into fully constrained sets:

Solved problem2:

Re-solved after point-on-circle constraint removed and circle radius changed: