API Reference

CVXPYlayers provides the CvxpyLayer class for each supported framework. The API is consistent across frameworks, with minor differences for framework integration.

• PyTorch API
  • CvxpyLayer
  • Usage Example
  • GPU Usage
  • Integration with nn.Module
• JAX API
  • CvxpyLayer
  • Usage Example
  • Computing Gradients
  • JIT Compilation
  • Vectorization with vmap
• MLX API
  • CvxpyLayer
  • Usage Example
  • Computing Gradients
  • Value and Gradient
  • Apple Silicon Optimization
  • Notes

Quick Reference

All frameworks share the same constructor signature:

CvxpyLayer(
    problem,           # cvxpy.Problem
    parameters,        # list of cvxpy.Parameter
    variables,         # list of cvxpy.Variable
    solver=None,       # optional solver
    gp=False,          # geometric program mode
    verbose=False,     # solver verbosity
    solver_args=None,  # default solver arguments
)

Framework Differences

Framework	Base Class	Call Method
PyTorch	torch.nn.Module	forward(*params)
JAX	Callable class	__call__(*params)
MLX	Callable class	__call__(*params)