Skip to content

Chapter 18. Advanced Topics

Many programs do not compute their output by applying a fixed sequence of explicit operations. Instead, they define the output as the solution of another problem.

SectionTitle
1Chapter 18. Advanced Topics
2Differentiating Through Solvers
3Differentiable Optimization Layers
4Continuous-Time Adjoint Methods
5Neural ODEs
6Probabilistic Automatic Differentiation
7Quantum Differentiation
8Differentiable Programming Languages
9Verified Differentiation
10Unified Differentiable Infrastructure
Chapter 18. Advanced TopicsMany programs do not compute their output by applying a fixed sequence of explicit operations. Instead, they define the output as the solution of another problem.
6 min
Differentiating Through SolversA solver is a program that computes a value by search, iteration, or factorization. Instead of evaluating a closed-form expression, it finds a value that satisfies a condition.
8 min
Differentiable Optimization LayersAn optimization layer is a program component whose output is the solution of an optimization problem. Instead of computing
8 min
Continuous-Time Adjoint MethodsMany systems evolve continuously over time rather than through discrete layers. A state variable changes according to a differential equation:
8 min
Neural ODEsClassical neural networks apply a finite sequence of transformations:
7 min
Probabilistic Automatic DifferentiationClassical automatic differentiation computes derivatives of deterministic programs.
8 min
Quantum DifferentiationQuantum computation introduces a computational model fundamentally different from classical programs.
7 min
Differentiable Programming LanguagesAutomatic differentiation began as a transformation applied to numerical programs. A differentiable programming language instead treats differentiation as a native semantic...
8 min
Verified DifferentiationAutomatic differentiation systems are usually trusted because they implement mathematically established rules such as the chain rule, product rule, and linearization of...
7 min
Unified Differentiable InfrastructureAutomatic differentiation began as a numerical technique for computing gradients of scalar functions.
7 min