optymus is a Python library for solving optimization problems in engineering and scientific computing. Built on JAX for automatic differentiation, it provides efficient gradient computation and GPU acceleration. The library supports continuous optimization, topology optimization, and combinatorial optimization (vehicle routing).

Getting Started

1. Install optymus:

   uv add optymus

2. Basic optimization example:

   from optymus import Optimizer
   from optymus.benchmark import Mccormick
   
   import jax.numpy as jnp
   
   f = Mccormick()
   initial_point = jnp.array([2.0, 2.0])
   
   opt = Optimizer(f_obj=f,
                   x0=initial_point,
                   method='bfgs')
   
   opt.report()

3. Vehicle Routing Problem:

   import numpy as np
   from optymus.routing import solve_vrp
   
   coordinates = np.array([[0,0], [1,2], [3,1], [4,3], [2,4]], dtype=float)
   demands = np.array([0, 3, 5, 4, 6], dtype=float)
   
   result = solve_vrp(
       coordinates=coordinates,
       demands=demands,
       vehicle_capacity=10,
   )
   print(result.routes)           # e.g. [[1, 4], [2, 3]]
   print(result.total_distance)

4. Topology optimization with engineering domains:

   from optymus.benchmark import MbbDomain
   from optymus.methods import polymesher
   
   # MBB beam domain with boundary conditions
   domain = MbbDomain
   
   # Generate polygonal mesh
   result = polymesher(domain=domain, num_elements=100)

Citation

If you use optymus in your research, please cite:

@misc{optymus2024,
  author = {da Costa, Kleyton and Menezes, Ivan and Lopes, Helio},
  title = {Optymus: Optimization Methods in Python},
  year = {2024},
  note = {GitHub Repository},
  url = {https://github.com/quant-sci/optymus}
}