v0.8.3

ADNLPModels v0.8.3

Diff since v0.8.2

Merged pull requests:

• Add jprod benchmark (#256) (@tmigot)
• Add benchmark analyzer (#259) (@tmigot)
• Add gradient benchmark CI (#261) (@tmigot)
• Add more benchmark CI (#262) (@tmigot)
• Fix benchmarks (#263) (@tmigot)
• Add benchmark folder README (#265) (@tmigot)
• Faster Hessian package benchmark (#267) (@tmigot)
• Skip if too many constraints (for instance polygon) (#268) (@tmigot)
• Optimize sparse AD (#269) (@amontoison)
• Use a symmetric coloring for the computation of sparse Hessians (#271) (@amontoison)
• Update benchmark/Manifest.toml (#272) (@amontoison)
• Symmetric decompression (#273) (@gdalle)
• Don't test on nightly (#274) (@gdalle)
• [Benchmarks] Update Manifest.toml (#275) (@amontoison)
• [documentation] Use Documenter.jl v1.0 (#276) (@amontoison)

Closed issues:

• Symmetric coloring for efficient sparse hessian computation (#258)

v0.8.2

ADNLPModels v0.8.2

Diff since v0.8.1

Merged pull requests:

• Full support of sparse AD for NLS models (#239) (@amontoison)
• Add basic gradient benchmark (#248) (@tmigot)
• [documentation] Explain coloring and detector kwargs in sparse backends (#249) (@amontoison)
• Add more benchmark: jac and hess (#251) (@tmigot)
• Pass empty function instead of unused objective in NLS backend (#253) (@tmigot)

Closed issues:

• Bug sparse hessian with conditionals (#140)
• Add sparse backend for Hessian of residual for NLS models (#183)
• Test Graphs.jl coloring algorithm (#204)
• Let users configure sparsity detector and coloring algorithm (#242)

v0.8.1

ADNLPModels v0.8.1

Diff since v0.8.0

Merged pull requests:

• Replace ColPack with SparseMatrixColorings (#244) (@gdalle)
• [documentation] Fix the table of AD backends (#245) (@amontoison)

Closed issues:

• Replace ColPack with SparseMatrixColorings? (#237)
• Display issue in docs (#243)

v0.8.0

ADNLPModels v0.8.0

Diff since v0.7.2

Merged pull requests:

• Use SparseConnectivityTracer.jl (#230) (@amontoison)
• Add OptimalControl.jl in Breakage.yml (#233) (@amontoison)
• Upgrade to ColPack v0.4 (#235) (@gdalle)

Closed issues:

• Bug with upgrade to SymbolicUtils release 1.6 and 1.7 (#232)

v0.7.2

ADNLPModels v0.7.2

Diff since v0.7.1

Merged pull requests:

• Add GPU tests (#222) (@tmigot)
• [CI] Setup Buildkite (#223) (@amontoison)
• Version 0.7.2 (#227) (@tmigot)

v0.7.1

ADNLPModels v0.7.1

Diff since v0.7.0

Merged pull requests:

• Remove duplicate API for NLS (#200) (@tmigot)
• Remove Zygote in performance (#201) (@tmigot)
• Add new constructors for mixed models (#202) (@tmigot)
• Enable gpu for bound-constrained problems (#208) (@tmigot)
• [JSOTemplate.jl] Template compliance update (#211) (@abelsiqueira)
• Add CITATION.cff (#212) (@tmigot)
• Fix missing link in mixed.md (#213) (@tmigot)
• Bump to NLPModelsModifiers 0.7 (#214) (@tmigot)
• Extend show for NLS backends (#216) (@tmigot)
• Bump to NLPModels 0.21 (#219) (@tmigot)
• Version 0.7.1 (#220) (@tmigot)

Closed issues:

• Update CITATION (#207)
• Inaccurate display (#215)

v0.7.0

ADNLPModels v0.7.0

Diff since v0.6.2

Closed issues:

• Improve docs (#32)
• Tutorial: get started with ADNLPModel (#49)
• Add documentation on sparse jacobian (#106)
• Add documentation on available backends and new backend system (#135)
• Warnings about unused type parameters (#156)
• WARNING: method definition ... declares type variable AD but does not use it (#159)
• Test using NLPModel as backend using NLPModelsTest (#178)

Merged pull requests:

• Add optimized reverse jtprod (#121) (@tmigot)
• Add test/Project.toml (#146) (@tmigot)
• Add timer to backend constructors (#147) (@tmigot)
• Split files for AD backends definitions (#149) (@tmigot)
• Specialize unconstrained backend (#150) (@tmigot)
• Add ForwardDiff optimized jprod (#151) (@tmigot)
• Add ReverseDiff generic gradient (#152) (@tmigot)
• Add optimized ForwardDiff jtprod (#153) (@tmigot)
• Add Enzyme gradient (#155) (@tmigot)
• Remove unused types in ADNLSModels (#157) (@tmigot)
• Dispatch hprod for ci, lag and obj (#158) (@tmigot)
• Add hprod with ForwardDiff [WIP] (#160) (@tmigot)
• Bump NLPModelsTest to 0.9 (#162) (@tmigot)
• Update computation sparse jacobian (#163) (@tmigot)
• pre-allocate result of direction derivative in sparse matrix comp (#164) (@tmigot)
• Use ColPack instead of SparseDiffTools for coloration (#166) (@tmigot)
• Optimize hprod in sparse hessian (#167) (@tmigot)
• [CI] Remove the build with Linux ARMv8 (#170) (@amontoison)
• Add reverse-forward hprod (#171) (@tmigot)
• Add SparseDiffTools backends (#172) (@tmigot)
• Add predefined backends (#174) (@tmigot)
• Add NLPModel instead of an AD-backend (#176) (@tmigot)
• Add Symbolics as optional dep (#177) (@tmigot)
• Add test for mixed models (NLP) (#180) (@tmigot)
• Add NLS tests for mixed models (#182) (@tmigot)
• Update readme and docs (#187) (@tmigot)
• Version 0.7.0 (#188) (@tmigot)
• Add predefined backend tutorial (#190) (@tmigot)
• Add performance tips tutorial (#191) (@tmigot)
• Build a hybrid model tutorial (#192) (@tmigot)

v0.6.2

ADNLPModels v0.6.2

Diff since v0.6.1

Closed issues:

• Use SparseDiffTools with ADNLPModel? (#13)
• Support sparse Hessians and Jacobians (#44)
• "UndefVarError: ZygoteAD not defined" when trying to use Zygote as AD backend (#134)

Merged pull requests:

• Sparse hessian (#127) (@amontoison)
• Rename symbolic backends (#136) (@tmigot)
• Rename SparseForwardADJacobian backend (#139) (@amontoison)
• Fix bug for unconstrained ghjvprod (#142) (@tmigot)
• Compute coloration on the whole Hessian matrix instead of triangle (#143) (@tmigot)
• Version 0.6.2 (#145) (@tmigot)

v0.6.1

ADNLPModels v0.6.1

Diff since v0.6.0

Merged pull requests:

• Update files related to sparse Jacobian (#125) (@amontoison)
• Sparse hessian with Symbolics.jl (#126) (@amontoison)
• Add NLS in-place constructors with no nonlinear constraints (#128) (@tmigot)
• Update SparseADHessian structure for an optimized hess_coord! (#130) (@amontoison)
• Fix API for sparse hessian (#131) (@tmigot)
• Version 0.6.1 (#132) (@tmigot)

v0.6.0

ADNLPModels v0.6.0

Diff since v0.5.1

Closed issues:

• Use in place functions by default for constraints and residuals (#47)
• Unit tests segfault on M1 Mac (#57)

Merged pull requests:

• Setup in-place constructors (#101) (@tmigot)
• Add abstract AD types (#103) (@tmigot)
• Add in-place residual (#104) (@tmigot)
• Add SparseDiffTools jacobian (#105) (@tmigot)
• Update CommentPR.yml (#107) (@amontoison)
• Optimize src/ad.jl (#110) (@amontoison)
• [CI] Update .cirrus.yml (#112) (@amontoison)
• Refactor lagrangian function (#113) (@tmigot)
• Add sparse jacobian of residual (#114) (@tmigot)
• Prep jprod! (#116) (@tmigot)
• CompatHelper: bump compat for SparseDiffTools to 2, (keep existing compat) (#117) (@github-actions[bot])
• Increase coverage by removing out-of-place gradient function (#118) (@tmigot)
• Add optimized jprod with forward and reverse (#119) (@tmigot)
• Optimized hprod using SparseDiffTools (#122) (@tmigot)
• CompatHelper: bump compat for NLPModels to 0.20, (keep existing compat) (#123) (@github-actions[bot])
• Version 0.6.0 (#124) (@tmigot)