ILTpy v.1.1.0 released!

New releases, events, publications using ILTpy, conference presentations.
Post Reply
davistdaniel
Posts: 2
Joined: Sun Sep 14, 2025 4:38 pm

ILTpy v.1.1.0 released!

Post by davistdaniel »

ILTpy v.1.1.0 released!

See on PyPi or JuGit.

New Features
  • Static baseline feature support for multidimensional inversions; use parameters={sb:[True,False]} during init for pseudo 2D inversions.
  • Progress bar now shows convergence during inversion.
  • Added alt_g solver number in _solver_type attribute.
  • Read-only attributes for kernels; new iltkernel function for defining kernels.
  • UV workflow implementation for package management and testing.
Bug Fixes
  • Fixed static baseline feature in multidimensional inversions.
  • Fixed _solver_type not being defined for default solver.
  • Fixed _ndim attribute usage in plotting module.
  • Removed global state change for numpy/scipy errors; now using numpy's default RNG in iltstats.
  • Removed redundant consistency checks and improved error handling.
  • Refactored parameters.py: fixed bug with same Identity() kernel object assignment for unspecified dimensions.
  • Refactored init_g_red: removed g_red_temp computation, use np.any for reg_bc flag, fixed unnecessary s_gg computation in each iteration during zc_reg.
Performance & Refactoring
  • Refactored initialization in inversion and parameters modules.
  • Refactored reporting.py: removed str() on Path objects, multiple file opens, and code duplication for ndim>2 cases; switched to dicts instead of vars.
  • Refactored parameters.py: replaced np.ones()*scalar with np.full for alpha terms; refactored [_extend_dim] to remove np.append and duplicate code.
  • _generate_idx now uses lists for better performance.
  • Refactored iltstats in inversion.py: lightweight data_dict template, fresh IltData objects per parallel worker, removed duplicate g_norm computation.
  • Removed log formatting; switched to src folder layout per modern Python standards.
Testing & CI
  • Major test suite refactor: generates synthetic data dynamically, removing external file dependency; test coverage ~97%.
  • New benchmark tests and plotting module tests.
  • Comprehensive coverage for input, fit, utils, output modules.
  • CI: run benchmarks during merge requests.
  • CI: GitLab pipeline adapted for src layout.
Compatibility
  • Support for Python 3.14 and latest versions of SciPy and NumPy.
Documentation
  • Updated docs with info on UV tooling and contribution guide.
Post Reply