v5.9.0

NeuralPDE v5.9.0

Diff since v5.8.0

Merged pull requests:

• [WIP] Bayesian PINN solver for ODE (#692) (@AstitvaAggarwal)
• Fix ODEs system and add test (#714) (@KirillZubov)
• CompatHelper: add new compat entry for Functors at version 0.4, (keep existing compat) (#715) (@github-actions[bot])
• CompatHelper: add new compat entry for MCMCChains at version 6, (keep existing compat) (#716) (@github-actions[bot])
• CompatHelper: add new compat entry for LogDensityProblems at version 2, (keep existing compat) (#717) (@github-actions[bot])
• CompatHelper: add new compat entry for AdvancedHMC at version 0.5, (keep existing compat) (#718) (@github-actions[bot])
• Incorporating given points into training (#723) (@sdesai1287)
• [WIP] API for BNNODE (#724) (@AstitvaAggarwal)
• CompatHelper: bump compat for Optimisers to 0.3, (keep existing compat) (#726) (@github-actions[bot])
• Rename samples field of WeightedIntervalTraining to points and clean tstops_loss logic (#727) (@sathvikbhagavan)
• Bump actions/checkout from 3 to 4 (#730) (@dependabot[bot])
• CompatHelper: bump compat for ComponentArrays to 0.15, (keep existing compat) (#732) (@github-actions[bot])
• CompatHelper: add new compat entry for MonteCarloMeasurements at version 1, (keep existing compat) (#736) (@github-actions[bot])
• CompatHelper: bump compat for SciMLBase to 2, (keep existing compat) (#738) (@github-actions[bot])

Closed issues:

• Have a training strategy for PreDefined Points (#644)
• Bayesian Inference for NNODE (#682)
• Type issue in Julia v1.9 (#696)
• Adaptive reweighting or modify neural network to solve complex system. (#708)
• Issued with sin and cos function after update version of NeuralPDE (#710)
• API for BNNODE. (#719)
• one-dimensional heat equation (#720)

v5.8.0

NeuralPDE v5.8.0

Diff since v5.7.0

Closed issues:

• Parameters estimation using NeuralPDE.jl is not working for PDEs (#681)
• Demo not running (#690)
• Parameters estimation using NeuralPDE.jl is not working for PDEs (#694)
• Failed to precompile NeuralPDE (#698)
• Precompilation warning (#709)

Merged pull requests:

• CompatHelper: bump compat for ComponentArrays to 0.14, (keep existing compat) (#699) (@github-actions[bot])
• CompatHelper: bump compat for Flux to 0.14, (keep existing compat) (#700) (@github-actions[bot])
• CompatHelper: bump compat for Flux to 0.14 for package docs, (keep existing compat) (#701) (@github-actions[bot])
• CompatHelper: bump compat for Lux to 0.5, (keep existing compat) (#704) (@github-actions[bot])
• CompatHelper: bump compat for Lux to 0.5 for package docs, (keep existing compat) (#705) (@github-actions[bot])

v5.7.0

NeuralPDE v5.7.0

Diff since v5.6.1

Closed issues:

• Push nnode further (#46)
• Treating states properly, second encounter (#605)

Merged pull requests:

• Explored some new training strategies for large neural networks. (#635) (@sdesai1287)
• added additional loss against data for NNODE (#666) (@AstitvaAggarwal)
• Try domainsets v0.6 (#675) (@ChrisRackauckas)
• Test new DiffEqBase (#676) (@ChrisRackauckas)
• fix formatting (#677) (@ChrisRackauckas)

v5.6.1

NeuralPDE v5.6.1

Diff since v5.6.0

Merged pull requests:

• update Symbolics and SymbolicUtils (#663) (@xtalax)
• Relax the IDE test for v1.6 (#671) (@ChrisRackauckas)
• test against PDESysLibrary (#674) (@xtalax)

v5.6.0

NeuralPDE v5.6.0

Diff since v5.5.0

Closed issues:

• @Nograd not defined (#661)
• Typo on the page example (#664)

Merged pull requests:

• Update README.md (#667) (@naceurCRAAG)
• Test master (#668) (@ChrisRackauckas)
• Update ks.md (#669) (@JunaidAkhter)

v5.5.0

NeuralPDE v5.5.0

Diff since v5.4.0

Closed issues:

• PDESystemLibrary.jl a comprehensive resource for discretizer packages. (#638)
• Additional loss is not logged (#649)
• NeuralPDE fail to compile (#660)

Merged pull requests:

• ComponentArray ERROR in Lux (#646) (@Xinyu-Li-1997)
• Log additional loss (#650) (@De-Souza)
• ComponentArray ERROR in Lux (2) (#651) (@De-Souza)
• Update pdesystem.md (#652) (@AstitvaAggarwal)
• Describe Flux vs Lux (#653) (@ChrisRackauckas)
• enable dependabot for GitHub actions (#655) (@ranocha)
• Bump actions/cache from 1 to 3 (#656) (@dependabot[bot])
• Bump actions/checkout from 1 to 3 (#657) (@dependabot[bot])
• Depwarn (#658) (@ChrisRackauckas)
• rebase (#662) (@xtalax)

v5.4.0

NeuralPDE v5.4.0

Diff since v5.3.0

Closed issues:

• CUDA: NAN (#466)
• QuasiRandomTraining(sampling_alg = SobolSample()) is not working correctly (#609)
• Using NeuralPDE with Boundary Conditions on the GPU (#612)
• ERROR: Optimization algorithm not found. (#618)
• Lots of warnings on example (#619)
• ERROR: LoadError: UndefVarError: SciMLProblem not defined (#620)
• Higher approximation order numeric_derivative (#641)

Merged pull requests:

• Correct sampling (#610) (@YichengDWu)
• canonize docs (#613) (@ArnoStrouwen)
• [skip ci] badges (#615) (@ArnoStrouwen)
• reproducible docs (#616) (@ArnoStrouwen)
• [skip ci] doc compat (#617) (@ArnoStrouwen)
• lib cov (#621) (@ArnoStrouwen)
• Use the parameterless type for doing adapting (#632) (@ChrisRackauckas)
• [skip ci] LanguageTool (#637) (@ArnoStrouwen)
• format markdown (#639) (@ArnoStrouwen)
• update to ArrayInterface v7 (#645) (@ChrisRackauckas)

v5.3.0

NeuralPDE v5.3.0

Diff since v5.2.0

Closed issues:

• How to using additional_loss input 3D data. (#589)
• How to use GPU to [chain1,chian2]? (#594)
• ERROR: MethodError: no method matching NeuralPDE.Phi (#595)
• Poisson Example not working on Julia 1.8 (#602)
• The implementation of finite differences throws away its advantage over AD (#607)

Merged pull requests:

• unify docs (#590) (@ArnoStrouwen)
• [skip ci] badges (#596) (@ArnoStrouwen)
• doc cov (#597) (@ArnoStrouwen)
• MassInstallAction: Install the Invalidations workflow on this repository (#601) (@devmotion)
• Adapt states to GPU correctly (#604) (@YichengDWu)
• Use different step sizes for different orders of derivatives (#608) (@YichengDWu)

v5.2.0

NeuralPDE v5.2.0

Diff since v5.1.1

Closed issues:

• Upstreaming ComponentArrays overloads for Adapt required for GPU (#584)
• How to use real data？ (#588)

Merged pull requests:

• Remove ComponentArrays overloads (#587) (@MilkshakeForReal)

v5.1.1

NeuralPDE v5.1.1

Diff since v5.1.0

Closed issues:

• Training on GPU with Lux (#579)

Merged pull requests:

• Don't initialize parameters twice (#586) (@MilkshakeForReal)