Module types 

Types for differentiable optimization (OptNet-style LP/QP layers).

Provides configuration, result, and gradient structures for differentiable quadratic and linear programming.

Structs

DiffLPConfig

Configuration for differentiable LP solving.

DiffLPResult

Result of a differentiable LP forward solve.

DiffOptGrad

Gradient of a loss w.r.t. all optimization layer parameters.

DiffOptParams

Parameters for a generic QP/LP optimization layer.

DiffOptResult

Result of a generic optimization layer forward pass.

DiffQPConfig

Configuration for differentiable QP solving.

DiffQPResult

Result of a differentiable QP forward solve.

ImplicitGradient

Implicit gradients of the loss w.r.t. problem parameters.

KKTSystem

KKT system residuals for monitoring convergence.

Enums

BackwardMode

Mode of backward differentiation through the optimization layer.

DiffOptStatus

Solver status for an optimization layer.