Skip to contents

Changelog

spmR (development version)

  • Multi-stock native status-determination alternatives retain each stock’s own catch instead of assigning the last stock’s catch to every stock.

  • Multi-stock Ricker projections keep each stock’s productivity coefficient separate, so changing stock order preserves the projections.

  • Multi-stock zero-catch initialization indexes the catch year and stock separately, preventing an out-of-bounds matrix write.

  • The GT2026 adapter rejects spawning months other than January before staging because its initial spawning-biomass verification assumes beginning-of-year spawning.

  • tier3_scenario_table() summarizes each stock separately for multi-stock inputs and retains the existing column layout for single-stock results.

  • The GT2026 report defaults to an audit-only render, so the plain Quarto command works in a fresh source checkout without ignored local run directories. Completed results are selected explicitly with saved_run.

  • The GT2026 example preserves the author’s confirmed 47-value SSB history and provides a portable fully worked example with the actual inputs, editable R commands, verified saved results, three Alternative 2 figures and CSV tables. Spawning biomass and catch use logarithmic y-axes; fishing mortality shows dashed F40 and F35 reference lines from the saved run. Its conditional results are published as a reference with the remaining scientific assumptions stated explicitly.

  • The standalone GT2026 report reads saved controls through the validated positional parser, so supported spm.dat comment labels no longer cause completed runs to be rejected.

spmR 0.4.3

  • write_spm_metadata() embeds validated version-2 declarations as comments in each species file by default. validate_spm_inputs() and runSPM() read those declarations automatically; an explicit metadata = "name.json" retains the separate JSON workflow.
  • The self-contained native engine reads the embedded declarations directly and advertises SPMR_INLINE_INPUT=1. A standalone run needs spm.dat and its species files. Legacy native sidecars remain readable, and the R runner generates a sidecar for older format-2 executables.
  • The GT2026 adapter accepts wrapped recruitment histories and an explicit per sex declaration. At the 0.4.3 release, the supplied example had 47 recruitment values and 45 SSB values; the adapter rejected that mismatch without padding or discarding values. Later source revisions are evaluated by the live input audit.

spmR 0.4.2

  • The ADMB projection engine now compiles from inst/admb/spm.tpl alone; the format-2 input checks are included in the template, and a separate C++ header is no longer required.
  • Package vignettes and website articles use Quarto, with a canonical GT2026 audit/staging example and a separate guide to its input definitions and conditional diagnostic runs.
  • run_gt2026_projection() in examples/gt_2026/run.R stages the supplied Greenland turbot inputs, records input repairs and metadata, and separates explicitly requested diagnostic weight proxies from supplied population weights. check_gt2026_inputs() lists the remaining source-data issues.

spmR 0.4.1

  • Split-sex species files now contain a male spawning weight-at-age vector, wt_M, immediately after wt_F. The R validator and ADMB engine both read and validate the new field. Pooled-sex files continue to use wt_F for both internal sex groups.
  • Version-2 projections no longer read or require tacpar.dat. They continue to require TAC_ABC = 1; legacy fitted and external TAC modes remain rejected.

spmR 0.4.0

  • New projections require input metadata format 2 and a compatible executable. Legacy output remains readable with run = FALSE.
  • Recruitment basis is explicit. The complete history is converted to total recruitment before calculating arithmetic and harmonic means, variability, simulations, and stock–recruitment inputs. Each projected total is allocated equally to the two sexes once.
  • Female and male population weights are separate from spawning and fishery weights. Missing male population weights require an explicit, recorded substitute for split-sex inputs. Metadata validation checks age order, dimensions, finite values, and compatible units.
  • Total biomass uses population weights throughout projections and biomass reference calculations. Male reference calculations use male selectivity.
  • Stock–recruitment fits use bounds tied to the normalized recruitment history and mean-corrected lognormal draws. Unit-scaled fixtures check the fitted relationships. New fitted runs require a maximum gradient below 1e-4 and a positive-definite Hessian.
  • New runs validate inputs before execution, check the executable’s format support, use a clean execution directory, and write SHA-256 provenance. Existing results survive a failed run.
  • Format 2 supports direct ABC projections (TAC_ABC = 1), recruitment modes 1 and 2, abundance scaling N_scalar = 1, and weight in kg. Auxiliary recruitment modes 3 and 4 and the experimental RTMB projection runner require further implementation before use with this format.

spmR 0.3.0

  • as_spm_result() provides a validated common result format for projection model backends while preserving model-specific columns.
  • runSPM() now dispatches through internal model adapters and returns an spm_result while remaining compatible with legacy ADMB and experimental RTMB calls.
  • tier3_scenario_table() summarizes simulation output into assessment-ready rows for the seven Tier 3 projection alternatives.

spmR 0.2.1

  • The experimental RTMB path can reuse existing output when rendering its comparison vignette. This path remains a simplified prototype and is not yet a full RTMB translation or numerically equivalent replacement for the ADMB model.