Changelog

0.5.0 (2021-12-01)

✨ Features

  • ✨ Feature: Megacomplex Models (#736)

  • ✨ Feature: Full Models (#747)

  • ✨ Damped Oscillation Megacomplex (a.k.a. DOAS) (#764)

  • ✨ Add Dataset Groups (#851)

  • ✨ Performance improvements (in some cases up to 5x) (#740)

πŸ‘Œ Minor Improvements:

  • πŸ‘Œ Add dimensions to megacomplex and dataset_descriptor (#702)

  • πŸ‘Œ Improve ordering in k_matrix involved_compartments function (#788)

  • πŸ‘Œ Improvements to application of clp_penalties (equal area) (#801)

  • ♻️ Refactor model.from_dict to parse megacomplex_type from dict and add simple_generator for testing (#807)

  • ♻️ Refactor model spec (#836)

  • ♻️ Refactor Result Saving (#841)

  • ✨ Use ruaml.yaml parser for roundtrip support (#893)

  • ♻️ Refactor Result and Scheme loading/initializing from files (#903)

  • ♻️ Several refactoring in glotaran.Parameter (#910)

  • πŸ‘Œ Improved Reporting of Parameters (#910, #914, #918)

  • πŸ‘Œ Scheme now excepts paths to model, parameter and data file without initializing them first (#912)

🩹 Bug fixes

  • 🩹 Fix/cli0.5 (#765)

  • 🩹 Fix compartment ordering randomization due to use of set (#799)

  • 🩹 Fix check_deprecations not showing deprecation warnings (#775)

  • 🩹 Fix and re-enable IRF Dispersion Test (#786)

  • 🩹 Fix coherent artifact crash for index dependent models #808

  • 🩹 False positive model validation fail when combining multiple default megacomplexes (#797)

  • 🩹 Fix ParameterGroup repr when created with β€˜from_list’ (#827)

  • 🩹 Fix for DOAS with reversed oscillations (negative rates) (#839)

  • 🩹 Fix parameter expression parsing (#843)

  • 🩹 Use a context manager when opening a nc dataset (#848)

  • 🚧 Disallow xarray versions breaking plotting in integration tests (#900)

  • 🩹 Fix β€˜dataset_groups’ not shown in model markdown (#906)

πŸ“š Documentation

  • πŸ“š Moved API documentation from User to Developer Docs (#776)

  • πŸ“š Add docs for the CLI (#784)

  • πŸ“š Fix deprecation in model used in quickstart notebook (#834)

πŸ—‘οΈ Deprecations (due in 0.7.0)

  • glotaran.model.Model.model_dimension -> glotaran.project.Scheme.model_dimension

  • glotaran.model.Model.global_dimension -> glotaran.project.Scheme.global_dimension

  • <model_file>.type.kinetic-spectrum -> <model_file>.default_megacomplex.decay

  • <model_file>.type.spectral-model -> <model_file>.default_megacomplex.spectral

  • <model_file>.spectral_relations -> <model_file>.clp_relations

  • <model_file>.spectral_relations.compartment -> <model_file>.clp_relations.source

  • <model_file>.spectral_constraints -> <model_file>.clp_constraints

  • <model_file>.spectral_constraints.compartment -> <model_file>.clp_constraints.target

  • <model_file>.equal_area_penalties -> <model_file>.clp_area_penalties

  • <model_file>.irf.center_dispersion -> <model_file>.irf.center_dispersion_coefficients

  • <model_file>.irf.width_dispersion -> <model_file>.irf.width_dispersion_coefficients

  • glotaran.project.Scheme(..., non_negative_least_squares=...) -> <model_file>dataset_groups.default.residual_function

  • glotaran.project.Scheme(..., group=...) -> <model_file>dataset_groups.default.link_clp

  • glotaran.project.Scheme(..., group_tolerance=...) -> glotaran.project.Scheme(..., clp_link_tolerance=...)

  • <scheme_file>.maximum-number-function-evaluations -> <scheme_file>.maximum_number_function_evaluations

  • <model_file>.non-negative-least-squares: true -> <model_file>dataset_groups.default.residual_function: non_negative_least_squares

  • <model_file>.non-negative-least-squares: false -> <model_file>dataset_groups.default.residual_function: variable_projection

  • glotaran.parameter.ParameterGroup.to_csv(file_name=parameters.csv) -> glotaran.io.save_parameters(parameters, 'file_name=parameters.csv)

🚧 Maintenance

  • 🩹 Fix Performance Regressions (between version) (#740)

  • πŸ§ͺπŸš‡ Add integration test result validation (#754)

  • πŸ”§ Add more QA tools for parts of glotaran (#739)

  • πŸ”§ Fix interrogate usage (#781)

  • πŸš‡ Speedup PR benchmark (#785)

  • πŸš‡πŸ©Ή Use pinned versions of dependencies to run integration CI tests (#892)

  • 🧹 Move megacomplex integration tests from root level to megacomplexes (#894)

  • 🩹 Fix artifact download in pr_benchmark_reaction workflow (#907)

0.4.0 (2021-06-25)

✨ Features

  • Add basic spectral model (#672)

  • Add Channel/Wavelength dependent shift parameter to irf. (#673)

  • Refactored Problem class into GroupedProblem and UngroupedProblem (#681)

  • Plugin system was rewritten (#600, #665)

  • Deprecation framework (#631)

  • Better notebook integration (#689)

🩹 Bug fixes

  • Fix excessive memory usage in _create_svd (#576)

  • Fix several issues with KineticImage model (#612)

  • Fix exception in sdt reader index calculation (#647)

  • Avoid crash in result markdown printing when optimization fails (#630)

  • ParameterNotFoundException doesn’t prepend β€˜.’ if path is empty (#688)

  • Ensure Parameter.label is str or None (#678)

  • Properly scale StdError of estimated parameters with RMSE (#704)

  • More robust covariance_matrix calculation (#706)

  • ParameterGroup.markdown() independent parametergroups of order (#592)

πŸ”Œ Plugins

  • ProjectIo β€˜folder’/’legacy’ plugin to save results (#620)

  • Model β€˜spectral-model’ (#672)

πŸ“š Documentation

  • User documentation is written in notebooks (#568)

  • Documentation on how to write a DataIo plugin (#600)

πŸ—‘οΈ Deprecations (due in 0.6.0)

  • glotaran.ParameterGroup -> glotaran.parameterParameterGroup

  • glotaran.read_model_from_yaml -> glotaran.io.load_model(..., format_name="yaml_str")

  • glotaran.read_model_from_yaml_file -> glotaran.io.load_model(..., format_name="yaml")

  • glotaran.read_parameters_from_csv_file -> glotaran.io.load_parameters(..., format_name="csv")

  • glotaran.read_parameters_from_yaml -> glotaran.io.load_parameters(..., format_name="yaml_str")

  • glotaran.read_parameters_from_yaml_file -> glotaran.io.load_parameters(..., format_name="yaml")

  • glotaran.io.read_data_file -> glotaran.io.load_dataset

  • result.save -> glotaran.io.save_result(result, ..., format_name="legacy")

  • result.get_dataset("<dataset_name>") -> result.data["<dataset_name>"]

  • glotaran.analysis.result -> glotaran.project.result

  • glotaran.analysis.scheme -> glotaran.project.scheme

  • model.simulate -> glotaran.analysis.simulation.simulate(model, ...)

0.3.3 (2021-03-18)

  • Force recalculation of SVD attributes in scheme._prepare_data (#597)

  • Remove unneeded check in spectral_penalties._get_area Fixes (#598)

  • Added python 3.9 support (#450)

0.3.2 (2021-02-28)

  • Re-release of version 0.3.1 due to packaging issue

0.3.1 (2021-02-28)

  • Added compatibility for numpy 1.20 and raised minimum required numpy version to 1.20 (#555)

  • Fixed excessive memory consumption in result creation due to full SVD computation (#574)

  • Added feature parameter history (#557)

  • Moved setup logic to setup.cfg (#560)

0.3.0 (2021-02-11)

  • Significant code refactor with small API changes to parameter relation specification (see docs)

  • Replaced lmfit with scipy.optimize

0.2.0 (2020-12-02)

  • Large refactor with significant improvements but also small API changes (see docs)

  • Removed doas plugin

0.1.0 (2020-07-14)

  • Package was renamed to pyglotaran on PyPi

0.0.8 (2018-08-07)

  • Changed nan_policiy to omit

0.0.7 (2018-08-07)

  • Added support for multiple shapes per compartment.

0.0.6 (2018-08-07)

  • First release on PyPI, support for Windows installs added.

  • Pre-Alpha Development