Update DiffEqBase requirement from 6 to 6, 7.21

[dependabot]

Updates the requirements on DiffEqBase to permit the latest version.

Release notes

Sourced from DiffEqBase's releases.

DiffEqBase-v7.21.1

DiffEqBase DiffEqBase-v7.21.1

Diff since DiffEqBase-v7.21.0

• Release OrdinaryDiffEqBDF 2.4.7 (#4490)
• Unify companion global error estimators into GlobalErrorEstimation (#4492)
• Add adjoint-based global error control as a SciMLSensitivity extension (#3953)
• ExpRK: cache ishermitian (rebase of #4305, + review fixes) (#4497)
• Mark discrete time-grid search helpers as Enzyme inactive (#4496)
• GlobalAdjoint: every-step trajectory scope and step-grid control (#4495)
• GlobalAdjoint: defer to adjoint_sensitivities default instead of pinning an adjoint (#4498)
• Release GlobalDiffEq 1.7.1 (#4501)
• Fix GlobalDiffEq precompilation on Julia 1.13 and companion accuracy (#4500)
• Make mass-matrix identity/zero/diagonal checks GPU-safe (#4502)

---

🤖 Posted by an AI agent — harness: Claude Code, model: claude-opus-5[1m]. Chris asked for this release.

https://claude.ai/code/session_014FEzNTLFutCmTEAZ3zBg5R

Merged pull requests:

• GlobalDiffEq: add adjoint-based global error control as a SciMLSensitivity extension (#3953) (@​ChrisRackauckas-Claude)
• fix: save all simultaneous events in a VCC (#4129) (@​AayushSabharwal)
• Add Friedli Exponential Runge-Kutta solver (#4481) (@​dhairya-motta)
• Release OrdinaryDiffEqBDF 2.4.7 (#4490) (@​ChrisRackauckas)
• Unify companion global error estimators into GlobalErrorEstimation (#4492) (@​ChrisRackauckas-Claude)
• GlobalAdjoint: every-step trajectory scope and step-grid control (closes #4493, #4494) (#4495) (@​ChrisRackauckas-Claude)
• Mark discrete time-grid search helpers as Enzyme inactive (#4496) (@​ChrisRackauckas-Claude)
• ExpRK: cache ishermitian (rebase of #4305, + review fixes) (#4497) (@​ChrisRackauckas-Claude)
• GlobalAdjoint: defer to adjoint_sensitivities default instead of pinning an adjoint (#4498) (@​ChrisRackauckas-Claude)
• GlobalDiffEq: fix Julia 1.13 precompilation and InterpolatingMode companion accuracy (#4500) (@​ChrisRackauckas-Claude)
• Release GlobalDiffEq 1.7.1 (#4501) (@​ChrisRackauckas)
• Make mass-matrix identity/zero/diagonal checks GPU-safe (fix Julia 1.13 GPU DAE scalar-indexing regression) (#4502) (@​ChrisRackauckas-Claude)
• Fix BDF step rejection for backward-in-time integration (#4503) (@​ChrisRackauckas-Claude)
• Release GlobalDiffEq 1.7.2 (#4506) (@​ChrisRackauckas)
• Release OrdinaryDiffEqBDF 2.4.8 (#4507) (@​ChrisRackauckas)
• Release OrdinaryDiffEqCore 4.17.1 (#4508) (@​ChrisRackauckas)
• Release OrdinaryDiffEqDefault 2.6.1 (#4509) (@​ChrisRackauckas)
• Release OrdinaryDiffEqFIRK 2.8.5 (#4510) (@​ChrisRackauckas)
• Release OrdinaryDiffEqRosenbrock 2.7.2 (#4511) (@​ChrisRackauckas)
• Release StochasticDiffEqCore 2.2.2 (#4512) (@​ChrisRackauckas)
• qa: allow JET 0.12 in OrdinaryDiffEqCore/BDF/NonlinearSolve QA envs (#4516) (@​ChrisRackauckas-Claude)
• Add regression tests for issues verified fixed in recent releases (#4517) (@​ChrisRackauckas-Claude)
• Fix BDF rejected-step controller for backward (tdir < 0) solves (#4519) (@​ChrisRackauckas-Claude)
• Release DiffEqBase 7.21.1 (#4521) (@​ChrisRackauckas)

Closed issues:

• Unrecognized keyword arguments: [:du0] after remake (#2138)
• Incorrect solution.stats.nf in implicit methods on SecondOrderODEProblems (#2537)

... (truncated)

Changelog

Sourced from DiffEqBase's changelog.

OrdinaryDiffEq.jl v7 / DifferentialEquations.jl v8 Breaking Changes

This release bumps to SciMLBase v3, RecursiveArrayTools v4, and includes breaking changes across DiffEqBase, OrdinaryDiffEqCore, and all solver sublibraries. It also coincides with the DifferentialEquations.jl v8 umbrella release, which is itself a breaking change to the user-facing meta-package.

DifferentialEquations.jl v8: scope reduction

DifferentialEquations.jl v8 no longer re-exports the full SciML solver suite. Previously, using DifferentialEquations pulled in OrdinaryDiffEq, StochasticDiffEq, DelayDiffEq, BoundaryValueDiffEq, Sundials, JumpProcesses, SteadyStateDiffEq, LinearSolve, NonlinearSolve, Optimization, etc. — a large default surface that drove up using time and made it unclear which package any given solver actually came from.

In v8, using DifferentialEquations only loads OrdinaryDiffEq. All other solver families have been removed from the umbrella. If your code relied on DifferentialEquations for SDEs, DDEs, BVPs, jumps, steady states, or any non-ODE solver, you will need to add the topic-specific package to your project explicitly.

Migration

Find the topic you need a solver for and add the corresponding sublib(s) directly. The DiffEqDocs tutorials and solver pages now specify, per algorithm, which package it ships from. Common cases:

Topic	Old (DiffEq v7 umbrella)	New (DiffEq v8)
ODEs	using DifferentialEquations	using OrdinaryDiffEq (or using OrdinaryDiffEqTsit5, OrdinaryDiffEqRosenbrock, … for individual solver families)
Stochastic ODEs	using DifferentialEquations	using StochasticDiffEq
Delay ODEs	using DifferentialEquations	using DelayDiffEq
Boundary value problems	using DifferentialEquations	using BoundaryValueDiffEq (or one of BoundaryValueDiffEqMIRK, BoundaryValueDiffEqFIRK, BoundaryValueDiffEqShooting, …)
Jump processes	using DifferentialEquations	using JumpProcesses
Steady state	using DifferentialEquations	using SteadyStateDiffEq
DAEs (mass matrix or implicit)	using DifferentialEquations	using OrdinaryDiffEq (mass matrix), using Sundials (IDA), or topic sublib
Sundials wrappers (CVODE, IDA, ARKODE)	using DifferentialEquations	using Sundials
Linear / nonlinear / optimization	using DifferentialEquations	using LinearSolve / using NonlinearSolve / using Optimization

For ODE work specifically, prefer importing only the sublib you need (e.g. using OrdinaryDiffEqTsit5: Tsit5) rather than the umbrella using OrdinaryDiffEq — the v7 ecosystem split lets you trim using time substantially. The DiffEqDocs tutorials and solver index annotate every algorithm with its host sublib.

Why

Removing the meta-package's broad re-exports lets each topic's package version cycle independently, eliminates the long using DifferentialEquations precompile chain for users who only need ODEs, and makes the dependency graph for any given script honest about what's actually being loaded.

This change is independent of the OrdinaryDiffEq v7 changes below — OrdinaryDiffEq v7 ships with DifferentialEquations v8, but you can also use OrdinaryDiffEq v7 directly without the umbrella package at all.

OrdinaryDiffEq.jl v7 Breaking Changes

This release bumps to SciMLBase v3, RecursiveArrayTools v4, and includes breaking changes across DiffEqBase, OrdinaryDiffEqCore, and all solver sublibraries.

Themes of the v7 release

Most of the breaking changes fall into a small set of recurring themes. Keep these in mind while reading the migration table — they explain why an individual change exists and often suggest the right migration direction:

• Time to first solve (TTFS) reduction. Direct deps on Static.jl, StaticArrayInterface.jl, Polyester.jl, and StaticArrays.jl were dropped; using OrdinaryDiffEq now loads only the default solver set; ODEFunction switched to AutoSpecialize. All of this means less code loaded and more precompilation caching on first solve.

• Type stability everywhere. All Bool solver/solve keyword arguments (autodiff, verbose, alias, lazy, …) were replaced by typed objects. Passing a Bool no longer changes dispatch in ways the compiler cannot specialize on, and the reverse is no longer allowed to silently fall back through slow generic paths. For example:

  # v6 — a Bool
  Rosenbrock23(autodiff = true)
  solve(prob, alg; verbose = false, alias = true)

... (truncated)

Commits

• 5cc92dd Fix BDF rejected-step controller for backward (tdir < 0) solves (#4519)
• 438c16c Release DiffEqBase 7.21.1 (#4521)
• a44c3a7 Add 4th-order Friedli Exponential Runge-Kutta solver (#4481)
• 7b0f4ae fix: save all simultaneous events in a VCC (#4129)
• 64d932c Add regression tests for issues verified fixed in recent releases (#4517)
• 310f147 qa: allow JET 0.12 in OrdinaryDiffEqCore/BDF/NonlinearSolve QA envs (#4516)
• 9adcec0 Release StochasticDiffEqCore 2.2.2 (#4512)
• 063a3ce Release OrdinaryDiffEqRosenbrock 2.7.2 (#4511)
• 910a1b7 Release OrdinaryDiffEqFIRK 2.8.5 (#4510)
• 8f4dac8 Release OrdinaryDiffEqDefault 2.6.1 (#4509)
• Additional commits viewable in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

---

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)