v7.91.0

SciMLSensitivity v7.91.0

Diff since v7.90.0

Merged pull requests:

• CompatHelper: bump compat for Optimization to 5 for package docs, (keep existing compat) (#1276) (@github-actions[bot])
• removed unecessary repack call (#1285) (@simenhu)
• Bump actions/checkout from 4 to 6 (#1290) (@dependabot[bot])
• Add Julia ecosystem to Dependabot configuration (#1291) (@ChrisRackauckas-Claude)
• Update OrdinaryDiffEqCore requirement from 1 to 1, 2.0 in the all-julia-packages group across 1 directory (#1293) (@dependabot[bot])

Closed issues:

• retcode: Failure in first example from tutorials (#1249)
• Compute the gradient of an ODE directly with Enzyme.jl (#1274)
• Using Stochastic Differential Equations broken in current julia version, but works in v1.9 (#1277)
• ComponentArray based problem causes lots of issues with Zygote.gradient() (#1283)

v7.90.0

SciMLSensitivity v7.90.0

Diff since v7.89.0

Merged pull requests:

• Add GK Adjoint to concrete_solve.jl, and the respective tests (#1273) (@maxesit)

Closed issues:

• (potential regression?) Case of parameter fitting suddenly throw error (MethodError: no method matching +(::@NamedTuple{…}, ::Base.RefValue{…})) (#1212)

v7.89.0

SciMLSensitivity v7.89.0

Diff since v7.88.0

Merged pull requests:

• Re-enable EnzymeAdjoint testing (#1232) (@ChrisRackauckas)
• feat: add Gauss Kronrod Adjoint (#1235) (@maxesit)
• Use explicit imports (#1240) (@ChrisRackauckas)
• Setup explicit imports in the code (#1241) (@ChrisRackauckas)
• Remove printing while running tests (#1245) (@DhairyaLGandhi)
• Add spell checking configuration (#1252) (@ChrisRackauckas)
• Add typos configuration and fix spelling error (#1253) (@ChrisRackauckas)
• Add downgrade CI workflow with Core1 test group (#1254) (@ChrisRackauckas)
• Update CI to test on Julia 1, lts, and pre versions (#1255) (@ChrisRackauckas)
• Apply JuliaFormatter to fix code formatting (#1256) (@ChrisRackauckas)
• Initialize line split fixes following Catalyst.jl PR #1306 (#1258) (@ChrisRackauckas)
• Update forwarddiffsensitivity_sparsity_components.jl to not have jac_prototype (#1263) (@ChrisRackauckas)

v7.88.0

SciMLSensitivity v7.88.0

Diff since v7.87.1

Merged pull requests:

• test: fix parameter initialization test (#1237) (@AayushSabharwal)

v7.87.1

SciMLSensitivity v7.87.1

Diff since v7.87.0

Merged pull requests:

• Update chaotic_ode.md (#1231) (@ChrisRackauckas)
• Update Mooncake error throw derivatives (#1234) (@ChrisRackauckas)
• Improve SciMLStructures support (#1236) (@ChrisRackauckas)

Closed issues:

• Forward differentiation with Enzyme doesn't work (#1226)
• Mooncake gives the wrong aggregator and thus does not give contextualized error messages (#1230)

v7.87.0

SciMLSensitivity v7.87.0

Diff since v7.86.1

Merged pull requests:

• Add EnzymeAdjoint (#1209) (@ChrisRackauckas)
• Add MooncakeAdjoint (#1227) (@ChrisRackauckas)
• Fix tests on master and separate MTK (#1229) (@ChrisRackauckas)

Closed issues:

• Grandient calculation fails when using a parameter-dependent SciMLOperator (#1139)
• EnzymeAdjoint (#1148)

v7.86.1

SciMLSensitivity v7.86.1

Diff since v7.86.0

Merged pull requests:

• Use SciMLBase.Void on EnzymeVJP functions (#1222) (@ChrisRackauckas)

v7.86.0

SciMLSensitivity v7.86.0

Diff since v7.85.0

Merged pull requests:

• actually add remake_zero (#1221) (@ChrisRackauckas)

Closed issues:

• Inconsistent Gradients between ForwardDiff and Zygote, MTK and Manual (#1219)

v7.85.0

SciMLSensitivity v7.85.0

Diff since v7.84.0

Merged pull requests:

• Documentation to use UDEs on the Brusselator equation. (#1215) (@sharvmurgai)
• SteadyStateAdjoint fix for applying dp twice (#1217) (@jClugstor)
• Use remake_zero! (#1220) (@ChrisRackauckas)

Closed issues:

• Wrong gradient when using AutoSpecialize and embedding the ODEProblem in a NamedTuple (#1181)
• EnzymeVJP slow down due to allocations from repeated make_zero (#1202)
• Zygote and ForwardDiff give different results for gradient of NonlinearSolve (#1210)

v7.84.0

SciMLSensitivity v7.84.0

Diff since v7.83.0

Merged pull requests:

• Add Mooncake to Alternative AD Frontends (#1151) (@willtebbutt)
• Fix GPU tests (#1213) (@ChrisRackauckas)
• Try to make doc build more robust (#1214) (@ChrisRackauckas)