Skip to content

OrdinaryDiffEqExtrapolation-v2.6.4

Choose a tag to compare

View all tags
@github-actions github-actions released this 03 Nov 02:50
OrdinaryDiffEqExtrapolation-v2.6.4
6efe4cd
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

OrdinaryDiffEqExtrapolation OrdinaryDiffEqExtrapolation-v2.6.4

Diff since OrdinaryDiffEqExtrapolation-v2.6.3

Merged pull requests:

Closed issues:

  • set_t! does not work for KenCarp4 and other extrapolating algorithms (#1353)
  • Second order ODE errors with Unitful units (#1397)
  • in place DAEProblem with jac_prototype broken. (#1966)
  • AutoFiniteDiff JVP recomputes f(x) on every Krylov matvec; nf accounting double-counts it (#4162)
  • SBDF solvers do not achieve expected convergence order (#4329)
  • StochasticDiffEq: JumpProblem callback stored in kwargs fires twice (#4439)
  • DelayDiffEq: callback save_positions decides the correctness of the history, not just the saved output (#4515)
  • OrdinaryDiffEqBDF QA (JET) fails on master: FBDF time_filter calls undefined functions in OOP path (#4518)
  • Pkg.test("OrdinaryDiffEqDifferentiation") fails because JET runs before the functional tests in the same process (#4542)
  • In-place interpolation uses the wrong sub-cache for alg_choice 2 and 4 with the default algorithm (#4555)
  • OrdinaryDiffEqBDF.jl v2.4.10 changed behavior significantly (#4573)
  • change_t_via_interpolation! behind a saved point leaves sol.t out of order (#4579)
  • AutoSpecialize callback wrapping shifts located continuous-event times (#4589)
  • DiffEqBase: Unitful ODE_DEFAULT_NORM computes sqrt(mean(|x|)), not RMS (#4596)
  • Rosenbrock DAE error estimate uses inv(abstol), which is wrong for vector abstol (#4598)
  • Float32: initdt falls back to dt=1e-6 via absolute eps guard, +27% steps on Lorenz (#4601)
  • DiffEqBase: FlexUnits, ReverseDiff, Tracker, Measurements, MonteCarloMeasurements ODE_DEFAULT_NORM compute sqrt(mean(|x|)), not RMS (#4610)
  • IMEXSSP3332 and IMEXSSP3433 are only first order accurate for a time-dependent explicit part (#4620)
  • ARS443: explicit weights be differ from Ascher–Ruuth–Spiteri (1997), §2.8 (#4621)
  • ReverseDiff.gradient through Rodas5(autodiff=AutoFiniteDiff()) throws a TrackedReal/TrackedArray LU TypeError (#4629)
  • ReverseDiff.gradient through Rodas5(autodiff=AutoFiniteDiff()) throws a TrackedReal/TrackedArray LU TypeError (#4630)
  • GPU group red on master since #4592: QNDF [jac=CSC, mass=diag_cu] gpu_mismatch in test/gpu/dae_tests.jl (#4638)
  • CompositeAlgorithm with 3+ algorithms never switches (choose_algorithm! no-op wins dispatch) (#4648)
  • Out-of-place Rosenbrock32 with a mass matrix goes Unstable where in-place converges (#4650)
  • Callback interpolation after a tstop-clamped step uses the restored proposal dt (#4655)
  • MethodOfSteps(FBDF()) corrupts history during adaptive overlapping DDE steps (#4659)
  • QNDF2 Unstable at t=0 on Robertson with user dt=1e-2 (#4675)
  • default_linear_interpolation is called with its arguments swapped, so its opt-outs never apply (#4684)
  • sol(t0; continuity = :left) returns the post-callback state without dense output (#4692)
  • Interpolating a Rosenbrock solution at exactly t0 returns NaN when a callback has an initialize function (#4693)
  • SRA: resize! leaves H0 stage buffers at the old length (DimensionMismatch on next step) (#4720)
  • sol(t0) returns NaN for Rosenbrock methods when a callback's initialize flags a derivative discontinuity (#4741)
  • save_discretes = false still saves the discretes at initialization (#4742)
  • reinit! on a DDE integrator leaves the history boundary at the old tspan[1], returning wrong history values (#4743)
  • Question: proof of contributor payments? (#4765)
  • ExplicitRK QA fails on master: JLArrays test extra loads LLVM 10 before AllocCheck's QA env (GPUCompiler IRBuilder UndefVarError) (#4773)
  • deleteat!/addat! on in-place integrators skip fsallast, nlsolver/J/W, and double-apply to aliased cache arrays (#4789)
  • DelayDiffEq: deleteat! throws UndefVarError, addat! misses solver buffers, resize! leaves grown history uninitialized (#4797)
  • DelayDiffEq: zero or negative constant_lags fail with DtLessThanMin / "unexpected time discrepancy" instead of a clear error (#4831)
  • Counted review request: PR #4714 at head 79d41dc (#4945)
  • [qa-hygiene] counted review round 2 of PR #4714 (head fa6d80f) — fresh-session reviewer task (#4946)

Contributors

ChrisRackauckas, inbounds, and 12 other contributors
Assets 2
Loading