SciMLBase

SciMLBase.jl is the core interface definition of the SciML ecosystem. It is a low dependency library made to be depended on by the downstream libraries to supply the common interface and allow for interexchange of mathematical problems.

v3.0 Breaking Changes

RecursiveArrayTools v4: Solution types are now AbstractArrays (#1297)

Most impactful change. AbstractVectorOfArray (and thus ODESolution, DDESolution, RODESolution, DAESolution) now subtypes AbstractArray:

• sol[i] returns the ith scalar element (column-major), not the ith timestep. Use sol.u[i] or sol[:, i] for timesteps.
• length(sol) returns total elements (prod(size(sol))). Use length(sol.u) for number of timesteps.
• iterate(sol) iterates scalar elements. Use sol.u for timestep iteration.
• map(f, sol) maps over elements. Use map(f, sol.u) for timesteps.

Ensemble RNG redesign (#1252)

• prob_func(prob, i, repeat) → prob_func(prob, ctx) where ctx::EnsembleContext
• output_func(sol, i) → output_func(sol, ctx)
• EnsembleContext includes sim_id, repeat, rng, sim_seed, worker_id, master_rng
• New seed/rng/rng_func kwargs on solve() for deterministic, thread-count-independent ensemble solves

Removed deprecated APIs

• u_modified! renamed to derivative_discontinuity! (#1289)
• Removed deprecated.jl: old type aliases (DEAlgorithm, DEProblem, DESolution, etc.), constructors, deprecated accessors (#1291)
• Removed backward compat shims in remake.jl and MLStyle extension (#1292)
• Removed old iterators: tuples, intervals, TimeChoiceIterator (#1290)
• Removed Symbol→ReturnCode.T conversion (Base.convert(::Type{ReturnCode.T}, ::Symbol) and symbol_to_ReturnCode) — deprecated for years across v1/v2 with a warning printed on every use. Comparing sol.retcode == :Success no longer works; use SciMLBase.successful_retcode(sol), which correctly accepts every success-ish ReturnCode value (Success, StalledSuccess, ExactSolutionLeft, ExactSolutionRight, FloatingPointLimit, …), not just Success.

Simplified getproperty

• Removed redundant getproperty overloads on solution abstract types (#1293)
• Removed deprecated getproperty aliases (.destats, .x, .lb/.ub, .minimizer, .minimum) (#1294)

Other breaking changes

• Replaced Moshi with plain Julia structs for Clocks — 23% precompilation improvement (#1295)
• ODEFunction uses DEFAULT_SPECIALIZATION (AutoSpecialize) for convenience constructors (#1300)
• Propagate interp/dense to DiffEqArrays from solution callables (#1297)
• is_discrete_time_domain(nothing) now returns false (#1306)

Migration Guide

Old (v2)	New (v3)
sol[i] (timestep)	sol.u[i] or sol[:, i]
length(sol) (timesteps)	length(sol.u)
for u in sol	for u in sol.u
u_modified!(integrator, true)	derivative_discontinuity!(integrator, true)
prob_func(prob, i, repeat)	prob_func(prob, ctx) — use ctx.sim_id, ctx.repeat
output_func(sol, i)	output_func(sol, ctx)
sol.destats	sol.stats
sol.retcode == :Success	SciMLBase.successful_retcode(sol)
ODEFunction{true}(f) (FullSpecialize)	Now uses AutoSpecialize by default

v2.0 Breaking Changes

The breaking changes in v2.0 are:

• IntegralProblem has moved to an interface with IntegralFunction and BatchedIntegralFunction which requires specifying prototypes for the values to be modified instead of nout and batch. https://github.com/SciML/SciMLBase.jl/pull/497
• ODEProblem was made temporarily into a mutable struct to allow for EnzymeRules support. Using the mutation throws a warning that this is only experimental and should not be relied on. https://github.com/SciML/SciMLBase.jl/pull/501
• BVProblem now has a new interface for TwoPointBVProblem which splits the bc terms for the two sides, forcing a true two-point BVProblem to allow for further specializations and to allow for wrapping Fortran solvers in the interface. https://github.com/SciML/SciMLBase.jl/pull/477
• SDEProblem constructor was changed to remove an anti-pattern which required passing the diffusion function g twice, i.e. SDEProblem(SDEFunction(f,g),g, ...). Now this is simply SDEProblem(SDEFunction(f,g),...). https://github.com/SciML/SciMLBase.jl/pull/489