Version History

• v0.1.0:

  • Implemented necessary features to run together with AixCaliBuHa and EnSTATS

• v0.1.1:

  • Fixed bugs necessary to work with AixCaliBuHa and EnSTATS and refactor functions based on feedback

• v0.1.2:

  • Move CalibrationClass to AixCaliBuHa

  • Add scipy.optimize.differential_evolution as an optimizer

• v0.1.3:

  • Move conversion.py to utils and make preprocessing a direct module

  • Introduce current_best_iterate as a parameter for optimization to ensure that the best solution to a problem is still saved even if an iteration step causes an error

  • Make interrupt of optimization through Keyboard-Interrupt possible

  • Adjust Goals functions to make slicing of multiple time-intervals possible.

• v0.1.4:

  • Create a tutorial with juypter-notebook

  • Introduce MultiIndex (from pandas) and make TimeSeriesData extend of the standard DataFrame

  • Adjust all classes, most notably Goals. This class will go into AixCaliBuHa, as it is only relevant for Calibrations.

• v0.1.5:

  • Remove dlib and PyQt5 from setup.py and delete TunerParas.show()

  • Refactor Optimizer so the framework parameter is only necessary to call optimize()

• v0.1.6:

  • Issue23: Change conversion functions to correctly handle multiheaders

  • Issue24: Add converters and correctly inherit from pd.DataFame in TimeSeriesData

  • Add functions for configuration using yaml

  • Add regex functions for extraction of modelica variables

  • Add option to directly return results as a dataframe using simulateMultiResultsModel.

• v0.1.7:

  • Issue13: Add an FMU API as simulation api

  • Issue36: Use Python Logging as default logger in all modules

  • Issue38: Fix future warning in pandas loffset

  • Issue39: Add simres from modelicares to avoid version errors with matplotlib.

  • Issue41: Fix inheritance of pd.DataFrame in TimeSeriesData.

    This also enables converting any single-index df to the required multi-index format (See #37).
    

  • Issue42: Move TunerParas to AixCaliBuHA

  • Add unit-tests for the dymola-api.

  • Use f-strings only

  • Update setup.py and requirements

  • Use TimeSeriesData in classes as output in the simulation api

• v0.1.8:

  • Drop support for Python 3.6 (See #44)

    • Breaking changes in pandas forces us to increase the version required

    • Pandas dropped the support for Python 3.6

• v0.2.0:

  • Prepare JOSS publication. Fixed #48

  • Prepare workshop using new examples. Fixes #49

  • Add multiprocessing to Simulation APIs. Fixed #5

  • Genereal restructuring of examples, tutorials and tests

  • Several new functions for easier access / utility

  • Add inputs option for DymolaAPI

• v0.2.1:

  • Fix installation error by using scikit-learn instead of sklearn

• v0.2.2:

  • Allow custom methods in StatisticsAnalyzer

  • Allow structural parameter changes in DymolaAPI

• v0.2.3:

  • Fix bad import in setup.py

  • Remove requirement for PyTables to enable support on python >3.9 #5

  • Add further save- and load functions for .csv

• v0.3.0:

  • Add parallelization option for optimization backends

• v0.3.1:

  • Avoid DType Warning #53

• v0.3.2:

  • Correct bounds for simulation Variables #56

  • Add script to reproduce simulation studies and store info in an archive #27

• v0.3.3:

  • Fix memory error for large parameter variations durring multiproccesing #76

  • Add option to set only one savepath for saving files of parameter varaiations #74

• v0.3.4:

  • Fix reproduction for new pip versions and old Dymola versions #80

  • Add support for the data format parquet with compression options #81

• v0.3.5:

  • Add simulation time estimation and checking of free hard drive space for saving #86

  • Add saving of directories for reproduction and tests for reproduction functions #87

• v0.3.6

  • Fix mp issue #95

  • fix tests for newer dependency versions

  • Remove warnings and update frequency if TimeSeriesData

  • Add CI based pypi release

• v0.3.7

  • Fix setup.py pydantic version

• v0.3.8

  • Fix sampling error for non-numeric values #69 #102

• v0.3.9

  • Fix error in loading newer ds-data from Dymola #70

• v0.3.10

  • Fix fmpy version #109

  • Fix data format for index frequency of tsd #107

  • Fix bug in function clean_and_space_equally_time_series #111

• v0.3.11

  • Add dymola_exe_path to set of kwargs in DymolaAPI #115

• v0.3.12

  • Port handling Add dymola_exe_path to set of kwargs in DymolaAPI #118

• v0.3.13

  • Fixes in simulate options and add Dymola CI tests #28, #113, #117

• v0.3.14

  • Fix retranslation #121

  • refactor cd to working_directory #61

  • Propagate names of simres to TimeSeriesData to speed up mat-loading

• v0.4.0

  • Preprocessing functions now copy the input DataFrame by default with the new inplace argument (default False). #125

  • Add inplace option to TimeSeriesData function (default True)

• v0.4.1

  • fix reproduction with modifiers #82

  • enable model_name=None upon startup of DymolaAPI #130

  • Only create logger handler if not already done by root-logger

• v0.4.2

  • Do not use chars_to_strings from scipy #132

• v0.4.3

  • Fix missing .egg for new Dymola release #138

• v0.5.0

  • Added bayesian-optimization to the possible optimizers #129

• v0.5.1

  • Add save-logs option for SimulatioAPI #141

  • Minor bug fixes #143