Changelog¶

For the full changelog, see:

Recent Changes¶

[0.10.0] - 2026-08-10¶

Fixed: - compute_sidelobe_level reported a sample on the main-lobe skirt instead of a sidelobe. It excluded the main beam out to one half-power beamwidth each side of the peak, but the first null of a tapered aperture sits at roughly 1.3 to 1.8 times the HPBW. The main lobe is now excluded out to its first null. - Reported sll_db values change, in some cases by more than 20 dB. A 32x32 Taylor -35 dB design at broadside went from -14.0 dB to -35.2 dB. Re-run any recorded trade study or requirement verification that used sll_db. - Same cause, also fixed: sll_db was non-monotonic in taper depth, and nearly insensitive to phase_bits off broadside.

[0.9.0] - 2026-07-22¶

Added: - NSGA-II multi-objective optimization (optimize_pareto, optional [mdao] extra); requirements as constraints - Constraint-aware DOE (validate="architecture") — no more per-case construction errors - Sobol global sensitivity indices (--sens-method sobol) - Search timeline metrics (PRF + search volume -> revisit rate, occupancy); see MDAO Tools - Thermal-reliability coupling: dissipated power -> junction temperature -> Arrhenius MTBF - Interactive plotly trade plots and self-contained report embeds

Changed: - Weighted-sum penalty normalized by requirement scale; round-robin interleaved timeline

[0.8.0] - 2026-07-22¶

Added: - Real ITU-R P.676-13 (line-by-line) and P.838-3 propagation with vendored coefficient tables - NRL sea clutter and Barton constant-gamma ground clutter (published models) - Swerling 0-4 threading through required SNR, integration gain, and Pd - Analytic error budget (quantization, RMS errors) wired into both antenna paths - Validation suite vs published references; validation table - Seed threading, provenance stamps, sorted batch results, pasys doe --resume - mypy at zero errors, enforced in CI

Changed (numbers shift): - Atmospheric/rain losses, clutter sigma0, CFAR loss vs Pfa, exact required SNR (golden margin -0.24 dB), radar honors cascaded NF (golden +1.16 dB SNR)

Removed: - Hand-fit placeholder coefficients, compute_cost_per_db, empirical Swerling/CFAR step tables

[0.7.0] - 2026-07-22¶

Added: - Digitization-level trades (element / subarray / analog) driving ADC count, data rate, compute, and power - ADC aperture-jitter SNR model and system dynamic range with array processing gain - Digital-section and receive-chain power in the DC budget; radar duty cycle - Exact Swerling 0-4 detection statistics (noncentral chi-square / gamma mixtures) - DBF architecture trade example and golden-case regression test

Fixed: - DOE runner dropped digital.*/reliability.* variables - augment_doe repeated original samples when reusing a seed - Radar equation double-counted pulse integration gain

[0.4.0] - 2026-02-01¶

Added: - Digital array model for digital beamforming calculations - RF cascade model for noise figure and gain cascade analysis - MkDocs documentation site with API reference, user guides, and tutorials

Fixed: - Ruff linting errors in models module - MathJax rendering in documentation

[0.3.0] - 2024-01-15¶

Added: - Radar detection model with pulse integration - CLI commands: pasys run, pasys doe, pasys report, pasys pareto - HTML and Markdown report generation

[0.2.0] - 2024-01-01¶

Added: - Design of Experiments (DOE) with LHS, random, and grid - Pareto frontier extraction and TOPSIS ranking - Requirements verification system

[0.1.0] - 2023-12-15¶

Added: - Initial release - Architecture configuration - Communications link budget model - Basic visualization