Description
Implicit forward and adjoint differentiation over lowered relations.
Structs§
- Accepted
Linearization - A linearization whose primal residual has been independently accepted.
- Accepted
Output Linearization - One accepted relation and output projection bound for a derivative action.
- Adjoint
Gradient - Accepted adjoint and total objective gradient.
- Discrete
Adjoint Checkpoint - One explicit accepted-state boundary inside a discrete adjoint trajectory.
- Discrete
Trajectory Adjoint - Reverse accumulation result over one validated discrete trajectory.
- Forward
Output Sensitivity - Accepted implicit state sensitivity and its selected output projection.
- Parameter
Jacobian - Matrix-free parameter Jacobian action
R_pfrom the same relation. - State
Jacobian - Matrix-free state Jacobian action
R_wderived from one relation JVP/VJP.
Functions§
- adjoint_
gradient - Solve the transposed state system and form the total parameter gradient.
- adjoint_
objective_ gradient - Compose one accepted relation and one objective intended for its accepted point.
- adjoint_
output_ gradient - Pull one complete output cotangent back through an accepted implicit relation.
- discrete_
trajectory_ adjoint - Compose reverse-mode cotangents over accepted discrete step relations.
- forward_
output_ sensitivity - Solve the implicit forward system and project it into complete outputs.
- forward_
sensitivity - Solve
R_w dw = -R_p dpat one accepted linearization.