Changes in version 0.2.2 (2026-09-14)                  

  - References in the Description field now use the CRAN-required
    authors (year) <doi:...> form, as requested in manual review. No
    code changes.

                        Changes in version 0.2.1                        

  - The Rust static archive is now removed from the build tree once the
    shared library is linked. R (>= 4.6.0) scans the symbol tables of
    linked static libraries, and the Rust standard library's bundled
    runtime objects reference exit/abort entry points that are never
    linked into the final shared library; leaving the archive behind
    produced a spurious checking compiled code warning on CRAN's Linux
    checks.

                        Changes in version 0.2.0                        

  - New dmdc(): Dynamic Mode Decomposition with control (Proctor,
    Brunton and Kutz, 2016), identifying the forced linear system
    x_{t+1} = A x_t + B u_t from snapshot pairs and control inputs.
    Supports joint estimation of A and B, a known-B mode that estimates
    only A, an optional reduced-order output projection, and autonomous
    multi-trajectory fits from explicit pairs (U = NULL).

  - New predict() method for dmdc objects simulates the identified
    system under a given control input sequence, or its zero-input free
    response.

  - The bundled Rust backend is updated to 'koopman-dmd' 0.2.0, which
    also fixes the full-A fallback and centered amplitudes in core DMD.

                        Changes in version 0.1.0                        

  - Initial release.

  - Core Dynamic Mode Decomposition (dmd()) with truncated SVD, optional
    mean centering, and automatic rank selection.

  - Extended DMD via lifting functions (polynomial, polynomial_cross,
    trigonometric, delay), for systems whose dynamics are nonlinear in
    the original coordinates.

  - Hankel-DMD (hankel_dmd()) using time-delay embedding, for scalar or
    low-dimensional signals.

  - Generalized Laplace Analysis (gla()) for computing Koopman
    eigenfunctions directly from weighted time averages.

  - Harmonic time averages (harmonic_time_average(), hta_convergence())
    and mesochronic harmonic plots (mesochronic_compute()) for phase
    space analysis, with orbit classification via
    classify_phase_space().

  - Built-in maps for experimentation: Chirikov standard, Froeschlé,
    extended standard, Hénon, and logistic.

  - Analysis helpers: dmd_spectrum(), dmd_stability(), dmd_error(),
    dmd_residual(), dmd_dominant_modes(), and dmd_reconstruct().

  - Numerics are implemented in Rust and reached through 'extendr'. The
    complete Rust dependency tree is bundled in src/rust/vendor.tar.xz,
    so the package builds without network access.