v1.2.0

New mechanism for piecewise polynomials, implementing GEP 8. Lots of internal quality improvements. Require dags >= 0.5.1, which will emit a DeprecationWarning if used with older ttsim versions.

What's Changed

• Better selection mechanism for node.colormap in graphs. by @hmgaudecker in #67
• Get rid of UnitTest style tests. by @hmgaudecker in #68
• Run code simplification. by @hmgaudecker in #69
• Boilerplate update by @MImmesberger in #74
• Generalise type hints of time conversion functions by @hmgaudecker in #81
• Improve test coverage by @MImmesberger in #75
• Add ty static type checker to CI by @MImmesberger in #82
• Allow non-string annotations in vectorization. by @hmgaudecker in #83
• Prepare for dags 0.5.0 by @hmgaudecker in #86
• Refactor piecewise polynomial to use explicit interval notation (GEP-08) by @hmgaudecker in #76
• Migrate docstrings from NumPy to Google convention by @hmgaudecker in #89
• Trivial updates to code quality. by @hmgaudecker in #90

Full Changelog: v1.1.1...v1.2.0

v1.1.1

A number of bug fixes and infrastructure improvements. New feature: time converters targeting stocks. First release to support Python 3.14.

What's Changed

• Fix test failure on WSL2 due to different DAG execution order by @JuergenWiemers in #49
• Fix fail_if checks: input data tree was too greedy and environment had an ambiguous name. by @MImmesberger in #52
• Raise error if scalar is passed to function expecting an array by @MImmesberger in #53
• Fix _q to _m time conversion. by @MImmesberger in #55
• Add and expose time converters targeting stocks by @MImmesberger in #56
• Some chores by @hmgaudecker in #62
• Move from mypy to ty by @hmgaudecker in #63
• Add Python 3.14 support. by @hmgaudecker in #65

Full Changelog: v1.1.0...v1.1.1

v1.1.0

Updates to performance and plotting. Includes some changes in the behavior of input_data -> processed_data -> raw_results -> results, so we bump the minor version.

• #44 Add a default fixed colormap to plots and general improvements to plotting.
  (@hmgaudecker)

• #46 Optimization for fail_if.foreign_keys_are_invalid_in_data
  (@JuergenWiemers)

• #42 Pre-sort user data by p_id for performance (@JuergenWiemers,
  @hmgaudecker)

• #41 Improve performance of tt.shared.join and
  ttsim.interface_dag_elements.fail_if.foreign_keys_are_invalid_in_data
  (@JuergenWiemers)

• #40 Improve performance of aggregation_numpy and data_converters
  (@JuergenWiemers)

v1.0.0

Initial release of TTSIM.

• #38 Plotting: Replace nodes selection type with all_paths. (@MImmesberger)

• #37 Make it possible to pass all main args as class methods. (@MImmesberger)

• #32 Build inputs template and plotting DAG from specialized environment based on
  policy_inputs (@MImmesberger)

• #34 Optimize JAX performance in data preparation pipeline
  (@JuergenWiemers)

• #23 Remove orphaned policy inputs from the TT DAG. (@MImmesberger)

• #19 Clearer architecture (@hmgaudecker)

• #17 Add type for sparse dicts with int keys param. (@MImmesberger)

• #16 Add fail/warn mechanism to ColumnObjects and ParamFunctions.
  (@hmgaudecker)

• #15 Do not call len() on unsized arrays. (@hmgaudecker)

• #14 Do not loop over the attributes of Jax arrays in
  fail_if.backend_has_changed (@hmgaudecker)

• #13 Put plot_tt_dag and plot_interface_dag into ttsim.plot.dag namespace;
  rename to tt and interface. (@MImmesberger)

• #11 Fix several bugs in plot_tt_dag. (@MImmesberger)

• #9 Raise an error when passing data but no targets. (@hmgaudecker)

• #8 Allow for input data as targets. (@MImmesberger)

• #6 Fail if the leaf name of an object in the policy environment differs from the
  last element of the path (@MImmesberger, @hmgaudecker)

• Prior to this, all development happened in a single GETTSIM repository. See
  the GETTSIM changelog for the
  history.

v1.0a4

What's Changed

• Improve "packaging" of METTSIM by @hmgaudecker in #22
• Remove orphaned nodes from tt dag by @MImmesberger in #23
• Optimize JAX performance in data preparation pipeline by @JuergenWiemers in #34
• Build inputs template and plotting DAG from specialized environment based on policy_inputs by @MImmesberger in #32
• Remove token from publish-to-pypi (counterproductive with trusted publisher set up) by @hmgaudecker in #35
• Fix (?) security permissions for GHA workflow. by @hmgaudecker in #36

New Contributors

• @JuergenWiemers made their first contribution in #34

Full Changelog: v1.0a3...v1.0a4

v1.0a3

What's Changed

• Add type for sparse dict with integer keys param by @MImmesberger in #17
• Clearer architecture by @hmgaudecker in #19

Full Changelog: v1.0a2...v1.0a3

v1.0a2

What's Changed

• Fail if the leaf name of an object in the policy environment differs from the last element of the path by @MImmesberger in #6
• Allow for input data as targets. by @MImmesberger in #8
• Raise error when passing data but no targets by @hmgaudecker in #9
• Fix several bugs in plot_tt_dag by @MImmesberger in #11
• Turn on fail_if nodes when running policy tests by @MImmesberger in #12
• Put plotting functions in ttsim.plot.dag namespace by @MImmesberger in #13
• Do not loop over the attributes of Jax arrays in 'fail_if.backend_has_changed' by @hmgaudecker in #14
• Do not call len() on unsized arrays by @hmgaudecker in #15
• Add fail/warn mechanism to ColumnObjects and ParamFunctions by @hmgaudecker in #16

Full Changelog: v1.0a1...v1.0a2

v1.0a1

This is the initial release of TTSIM after extracting all core functions, which are unrelated to the German taxes and transfers system, from GETTSIM.

TTSIM is extensively tested and hence comes with its own little country model (for Middle Earth — METTSIM).

v1.0a

This is the initial release of TTSIM after extracting all core functions, which are unrelated to the German taxes and transfers system, from GETTSIM.

TTSIM is extensively tested and hence comes with its own little country model (for Middle Earth — METTSIM).