[Diff since v1.0.1](v1.0.1...v1.1.0)

[Diff since v1.0.0](v1.0.0...v1.0.1)

**Merged pull requests:**
- Improve type stability in implicit (#21) (@gdalle)

[Diff since v0.3.1](v0.3.1...v1.0.0)

- no breaking changes. mainly just moving to 1.0.0 to make semver easier.
- add a pytorch example
- minor doc, CI, compat, updates

**Merged pull requests:**
- Remove the unused parametric types present after PR 13. (#14) (@juddmehr)
- Add user defined matrix multiplication function (#16) (@jmaack24)
- Update eigenvalues.jl (#17) (@BTV25)
- Allow ForwardDiff version 1 in [compat] (#19) (@fredrikekre)

**Closed issues:**
- Keep getting error with provide_rule (#18)

## ImplicitAD v0.3.1

[Diff since v0.3.0](v0.3.0...v0.3.1)



**Merged pull requests:**
- Qualify access to TrackedReal and TrackedArray in nonlinear.jl (#13) (@dingraha)

## ImplicitAD v0.3.0

[Diff since v0.2.2](v0.2.2...v0.3.0)


**Closed issues:**
- Differences with ImplicitDifferentiation.jl? (#4)
- Alternative method for providing partial derivatives? (#7)
- Partial derivative matrix cannot be safely re-used. (#9)

**Merged pull requests:**
- Avoid overwriting cached variables before the reverse pass (#10) (@taylormcd)
- add explicit_unsteady and implicit_unsteady (#11) (@taylormcd)
- Faster unsteady (#12) (@andrewning)

## ImplicitAD v0.2.2

[Diff since v0.2.1](v0.2.1...v0.2.2)



**Merged pull requests:**
- relax compatibility requirements (#6) (@taylormcd)

## ImplicitAD v0.2.1