go-autodiff documentation
go-autodiff is the planned home of automatic differentiation for go-scikit: forward and reverse modes, with Jacobian and Hessian helpers that feed go-numerics/optimize gradients and the JacFunc of go-numerics/integrate, both of which take user-supplied derivatives or finite differences today. It is in wave 2 (PLAN.md item 8).
Status: planned (wave 2). The organisation has no library and no release tag. These pages state the scope, the plan and the references each library will be validated against. The plan itself is PLAN.md of go-scikit/roadmap, and the state of each item is in STATUS.md.
Plan
In PLAN.md, wave 2:
- item 8 (wave 2): forward (dual) and reverse (tape) modes, Jacobians and Hessians for optimize and integrate; size M.
It builds on go-ndarray; it feeds go-numerics/optimize and go-numerics/integrate, and later the neural networks of go-mlkit.
The libraries
| Library | Status | What it will provide |
|---|---|---|
| Automatic differentiation | planned (wave 2) | Planned for wave 2: forward and reverse automatic differentiation with Jacobian and Hessian helpers. |
Every library of the go-scikit family is pure Go (CGO_ENABLED=0), validated
against the tool that is the reference of its domain, run as an oracle from Go,
and measured against it on one core. None of that is claimed here yet: there is
nothing to measure.
What each library’s pages hold
| Library | Pages |
|---|---|
| Automatic differentiation | overview |
Created on by tannevaled
Modified on by tannevaled