## DiffEqFlux v1.15.0

[Diff since v1.14.0](https://github.com/SciML/DiffEqFlux.jl/compare/v1.14.0...v1.15.0)


**Closed issues:**
- Support for complex matrix ODEs (#74)

**Merged pull requests:**
- Tensor prod layer [WIP] (#313) (@d-netto)
- update requires for CUDA.jl (#323) (@ChrisRackauckas)

## DiffEqFlux v1.14.0

[Diff since v1.13.0](https://github.com/SciML/DiffEqFlux.jl/compare/v1.13.0...v1.14.0)


**Closed issues:**
- Supervised NNODE Example Broken? (#318)

**Merged pull requests:**
- Use Float32s in ComponentArray example (#315) (@jonniedie)
- rename docs/src/examples markdown files (#317) (@JeremyFongSP)
- CompatHelper: bump compat for "Zygote" to "0.5" (#319) (@github-actions[bot])
- [WIP] Add pde constrained example (#320) (@JeremyFongSP)

## DiffEqFlux v1.13.0

[Diff since v1.12.1](https://github.com/SciML/DiffEqFlux.jl/compare/v1.12.1...v1.13.0)


**Closed issues:**
- Continuous normalizing flows (#46)
- FFJORD (#47)
- Correct way to index into parameter vector? (#281)
- Library-tized FFJORD (#282)
- Use sciml_train for multiple initial conditions (#306)

**Merged pull requests:**
- Use JuliaGPU's GitLab CI setup (#284) (@christopher-dG)
- Correct toml/FFJORD (#291) (@d-netto)
- Some cleanup after FFJORD PR (#295) (@ChrisRackauckas)
- CompatHelper: add new compat entry for "Distributions" at version "0.23" (#296) (@github-actions[bot])
- CompatHelper: add new compat entry for "Zygote" at version "0.4" (#297) (@github-actions[bot])
- CompatHelper: bump compat for "Adapt" to "2.0" (#299) (@github-actions[bot])
- FFJORD for multivariate distributions (#302) (@d-netto)
- Update Neural ODE MNIST Example (#303) (@avik-pal)
- Added example with ComponentArrays (#304) (@jonniedie)
- FFJORD: docstrings and small change in loss function for multivariate gaussian test (#305) (@d-netto)
- Documentation (#307) (@d-netto)
- FFJORD: tutorial (#308) (@d-netto)
- Remove using OrdinaryDiffEq (#309) (@ChrisRackauckas)
- more FFJORD cleanup (#310) (@ChrisRackauckas)
- Fix in documentation/tutorial (#311) (@d-netto)
- Update ffjord.jl docstring typo (#314) (@Vaibhavdixit02)

## DiffEqFlux v1.12.1

[Diff since v1.12.0](https://github.com/SciML/DiffEqFlux.jl/compare/v1.12.0...v1.12.1)


**Closed issues:**
- Implicit differential equation solver with internal autodiff of the Jacobian (#31)
- BSplines to approximate gradient of data  (#75)
- Neural optimal control documentation (#174)
- README Stable docs links broken (#251)
- Indexing solution in loss function gives zero gradient with default sensealg (#259)
- RadauAII5 and typeassert issues in autodiff (#271)
- Error in LV-Univ example (#275)

**Merged pull requests:**
- add analytical tgrad for mass matrix methods (#224) (@ChrisRackauckas)
- Add mini batch docs (#267) (@collinwarner)
- add nested AD stiff tests and update docs (#274) (@ChrisRackauckas)
- Update docs (#277) (@mkg33)
- Update LICENSE (#280) (@ViralBShah)
- Add Documentation for LV-NN-Stiff.md (#285) (@JeremyFongSP)
- Fix for RecursiveArrayTools broadcast fix (#286) (@ChrisRackauckas)

## DiffEqFlux v1.12.0

[Diff since v1.11.0](https://github.com/SciML/DiffEqFlux.jl/compare/v1.11.0...v1.12.0)



**Merged pull requests:**
- Remove concrete_solve (#273) (@ChrisRackauckas)

## DiffEqFlux v1.11.0

[Diff since v1.10.3](https://github.com/SciML/DiffEqFlux.jl/compare/v1.10.3...v1.11.0)


**Closed issues:**
- Using reverse-mode autodiff with a second order ODE problem (#48)
- [Bug Report] Rosenbrock23 gives no method matching finite_difference_jacobian (#257)
- NewtonSecondOrderAdjoints code in document not working (#258)
- LV-Univ example code doesn't match intro text (#262)
- reconstruct_Foo instead of reconstruct NeuralODE (#264)

**Merged pull requests:**
- Second Order Neural ODEs (#254) (@ChrisRackauckas)
- [WIP] Add Documentation to MNIST example (#256) (@JeremyFongSP)
- Docs fix: remove parameter `p` from call of `predict_rd` (#260) (@metanoid)
- Update LV-delay.md to include training step (#261) (@metanoid)
- Delete nfe variable in code (#263) (@JeremyFongSP)
- Remove unused functor definitions (#265) (@ChrisRackauckas)
- Update layers.jl (#269) (@ChrisRackauckas)

## DiffEqFlux v1.10.3

[Diff since v1.10.2](https://github.com/SciML/DiffEqFlux.jl/compare/v1.10.2...v1.10.3)



**Merged pull requests:**
- Doc typi (#247) (@jlperla)
- missing global in docs example (#248) (@dehann)
- fuse 2: refused (#252) (@ChrisRackauckas)

## DiffEqFlux v1.10.2

[Diff since v1.10.1](https://github.com/SciML/DiffEqFlux.jl/compare/v1.10.1...v1.10.2)


**Closed issues:**
- FastChain adjoint definition is not compatible with acting on a matrix (#192)
- Outer vs Inner constructors style (#230)
- Google season of Docs (#237)
- FYI only - Interesting paper on "Differentiation of Blackbox Combinatorial Solvers" (#239)
- Neural ODE example with GPU does not work (#243)

**Merged pull requests:**
- make all inner constructors for consistency (#245) (@ChrisRackauckas)
- Fix adjoints with matrices and default neural ODEs to ZygoteVJP (#246) (@ChrisRackauckas)

## DiffEqFlux v1.10.1

[Diff since v1.10.0](https://github.com/SciML/DiffEqFlux.jl/compare/v1.10.0...v1.10.1)



**Merged pull requests:**
- fix default adjoints for GPU (#240) (@ChrisRackauckas)
- Add GPU-based MNIST example to the docs (#242) (@ChrisRackauckas)

## DiffEqFlux v1.10.0

[Diff since v1.9.0](https://github.com/SciML/DiffEqFlux.jl/compare/v1.9.0...v1.10.0)


**Closed issues:**
- Gradients w.r.t. scalar indexing of a DiffEqArray (#160)
- Store and return the current best loss from sciml_train on Flux.jl optimizers (#170)
- Broadcasting problem in Zygote.gradient with FastChain() (#171)
- Issue with batch training of NeuralODE with signals as inputs (#225)
- Docstrings and documentation for fast layers (#231)

**Merged pull requests:**
- Utilize view strided fixes (#152) (@ChrisRackauckas)
- CompatHelper: bump compat for "Optim" to "0.21" (#222) (@github-actions[bot])
- Documenter (#228) (@Emmanuel-R8)
- gigantic doc update (#232) (@ChrisRackauckas)
- Deploy Docs (#233) (@kanav99)
- Add MLSL global optimization (#234) (@AStupidBear)
- Gigantic documentation update 2 (#235) (@ChrisRackauckas)

## DiffEqFlux v1.9.0

[Diff since v1.8.1](https://github.com/SciML/DiffEqFlux.jl/compare/v1.8.1...v1.9.0)


**Closed issues:**
- BBO breaks (#199)
- Force stopping BBO from callback fails (#205)

**Merged pull requests:**
- Add callback kwarg to BlackBoxOptim dispatch (#201) (@abhigupta768)
- See what the tests say (#213) (@ChrisRackauckas)
- Add TikTak global optimization to sciml_train (#217) (@ranjanan)
- Move TikTak to requires.jl (#219) (@ranjanan)
- Add QuadDIRECT global optimizer  (#220) (@ranjanan)
- analytical tgrad for neural ODEs (#223) (@ChrisRackauckas)

## DiffEqFlux v1.8.1

[Diff since v1.8.0](https://github.com/SciML/DiffEqFlux.jl/compare/v1.8.0...v1.8.1)



**Merged pull requests:**
- Failed when loss function returned more than one value to pass to cb (#203) (@ChrisDeGrendele)
- Adding `kwarg` for Optim options (#208) (@YingboMa)

## DiffEqFlux v1.8.0

[Diff since v1.7.1](https://github.com/SciML/DiffEqFlux.jl/compare/v1.7.1...v1.8.0)


**Closed issues:**
- Theoretical underpinning (#193)
- Solving Forward Backward SDE  (#197)
- FastChain in loss (#198)

**Merged pull requests:**
- improve logging of sciml_train (#194) (@AStupidBear)
- Save best solution rather than last one in sciml_train (#195) (@ChrisDeGrendele)
- Add BlackBoxOptim dispatch to sciml_train (#196) (@abhigupta768)

## DiffEqFlux v1.7.1

[Diff since v1.7.0](https://github.com/JuliaDiffEq/DiffEqFlux.jl/compare/v1.7.0...v1.7.1)


**Closed issues:**
- Setup progress bar of sciml_train to be platform agnostic (#158)
- sciml_train fails on GPU  (#182)
- `sciml_train` error when using `NewtonTrustRegion` optimizer (#185)
- Neural SDE does not train diffusion parameters (#189)

**Merged pull requests:**
- Provide default progress loggers and use new ProgressLogging API (#184) (@devmotion)
- Test DiffEqBase promotion (#188) (@ChrisRackauckas)
- fix and better test neural SDE on FastChain (#191) (@ChrisRackauckas)

## DiffEqFlux v1.7.0

[Diff since v1.6.0](https://github.com/JuliaDiffEq/DiffEqFlux.jl/compare/v1.6.0...v1.7.0)


**Closed issues:**
- Singular mass matrices not compatible with hardcoded adjoints (#34)
- StackOverflowError on a simple example (#149)
- Setup progress bar of sciml_train to be platform agnostic (#158)

**Merged pull requests:**
- Add progress bar to sciml_train, fixes #158 (#180) (@abhigupta768)
- support second order (#181) (@ChrisRackauckas)

See github.com/JuliaDiffEq/DiffEqFlux.jl/releases/tag/v1.3.1 for release notes