API Reference

Moreau provides a unified API across CPU and GPU backends, with integrations for PyTorch and JAX.

• Core API
  • Solver
  • CompiledSolver
  • Cones
  • SolverType
  • Settings
  • IPMSettings
  • Data Types
  • Device Functions
• PyTorch API
  • Solver
  • Functional API
  • Gradient Computation
  • GPU Usage
  • Data Types
• JAX API
  • Solver
  • Functional API
  • Gradient Computation
  • Batching with vmap
  • JIT Compilation
  • Data Types

Quick Reference

Core Classes

Class	Description
moreau.Solver	Single problem solver
moreau.CompiledSolver	Batched solver with shared structure
moreau.Cones	Cone specification
moreau.Settings	Solver configuration
moreau.IPMSettings	IPM algorithm settings
moreau.SolverType	Solver algorithm type enum

Framework Integrations

Framework	Main Class	Features
PyTorch	moreau.torch.Solver	autograd, functional API
JAX	moreau.jax.Solver	jit, vmap, grad

Data Types

Type	Description
Solution	Primal/dual solution (x, z, s)
BatchedSolution	Batched solution arrays
WarmStart	Warm start point for single problem
BatchedWarmStart	Warm start point for batched problems
SolveInfo	Solver metadata (status, timing)
BatchedSolveInfo	Batched solver metadata
TuneResult	Auto-tune benchmark results
SolverStatus	Solve outcome enum
TorchSolution / TorchBatchedSolution	PyTorch solution types
TorchSolveInfo / TorchBatchedSolveInfo	PyTorch metadata types
JaxSolution	JAX solution NamedTuple
JaxSolveInfo	JAX metadata NamedTuple