CHANGELOG.md

[3.26.11] - 2026-05-21

Fixed

• Preserve computed structural dataset variables when subsampling simulations.

[3.26.10] - 2026-05-17

Changed

• Derive microsimulation weights from household weights instead of entity-specific weight variables.

[3.26.9] - 2026-05-17

Changed

• Exclude pseudo-inputs and calculated values from simulation input exports by default.

[3.26.8] - 2026-05-17

Changed

• Disallow variable definitions that combine formula, adds/subtracts, and uprating computation modes.

[3.26.7] - 2026-05-17

Changed

• Disallowed formula-time randomness through random(), directing model authors to use input seed or draw variables instead.

[3.26.6] - 2026-05-17

Fixed

• Raised a clear error when numeric simulation inputs contain NaN values.

[3.26.5] - 2026-05-14

Fixed

• Fixed uprating dependency sorting for cloned parameter subtrees with duplicate parameter names.

[3.26.4] - 2026-05-14

Fixed

• Use safe YAML loaders and remove dynamic eval from parameter breakdown handling.

[3.26.3] - 2026-05-14

Fixed

• Preserve values passed through dispatched set_input handlers across cache invalidation.
• Apply chained parameter uprating dependencies in deterministic dependency order.

[3.26.2] - 2026-05-14

Fixed

• Avoid constructing macro-cache metadata during calculations when macro-cache reads are disabled.

[3.26.1] - 2026-05-07

Changed

• Added provider-neutral AI engineering guidance for tests, documentation review, pull requests, and changelog fragments.

Fixed

• Fixed enum encoding for HDF5-style UTF-8 byte-string arrays containing non-ASCII enum member names.

[3.26.0] - 2026-05-04

Added

• Added dependency-free runtime metadata for bundle validation.

[3.25.4] - 2026-04-30

Fixed

• Preserve user-provided ETERNITY inputs across cache invalidation when they were set for an ordinary period.

[3.25.3] - 2026-04-28

Changed

• Updated GitHub Actions workflows for Node 24-compatible action runtimes.

[3.25.2] - 2026-04-21

Fixed

• Simulation.subsample now invalidates _fast_cache. Previously, entries populated by to_input_dataframe held pre-subsample arrays that survived build_from_dataset and were returned by the fast-path short-circuit in Simulation.calculate whenever a caller passed decode_enums=False, triggering "size X != Y = count" projection errors in downstream code (e.g. person.household("household_weight", period) inside person_weight).

[3.25.1] - 2026-04-18

Fixed

• Iterate only existing holders in _invalidate_all_caches. The 3.24.4-era implementation walked every variable in the tax-benefit system and lazy-created a Holder for each one, inflating apply_reform from milliseconds to seconds in downstream packages with thousands of variables (policyengine-us YAML full-suite went from ~17 min to ~51 min per job and started timing out at the 1-hour GitHub Actions limit). Untouched variables have no holder and therefore nothing to wipe, so iterating population._holders.values() on each population recovers the original performance while keeping the set_input preservation behaviour introduced in #475.

[3.25.0] - 2026-04-18

Added

• Add a downstream preflight smoke test (tests/smoke/test_country_init.py) that instantiates CountryTaxBenefitSystem() from the currently-published policyengine-us and policyengine-uk against the core under test. Catches the class of cross-repo regressions that the existing RUN_SMOKE_TESTS-gated microsimulation smoke test misses — breakdown-validator errors at parameter load, _fast_cache-init assumptions, and any other init-time changes that happen to break downstream country models. Also extends the smoke job in .github/workflows/pr.yaml to install policyengine-uk alongside policyengine-us.

Fixed

• Guard invalidated_caches mutation sites in Simulation against the attribute being missing, matching the pattern used for _fast_cache in #474. Country-package subclasses that override __init__ without calling super().__init__ can skip initialising this private attribute, so purge_cache_of_invalid_values and invalidate_cache_entry now fall back gracefully instead of raising AttributeError.

[3.24.4] - 2026-04-18

Fixed

• Preserve set_input values across apply_reform. The H3 cache invalidation wiped every variable's _memory_storage._arrays, which also wiped user-provided dataset inputs loaded via set_input. Country-package subclasses calling set_input during construction and then applying a structural reform (the policyengine-uk pattern) silently lost their datasets. Now tracks set_input provenance and replays those values after the invalidation wipe; formula-output caches are still invalidated as before.
• Restore nested-branch input inheritance and cover situation-dict set_input. Three follow-ups on top of the Simulation.set_input preservation: (1) Holder.set_input also records _user_input_keys so situation-dict inputs routed through SimulationBuilder.finalize_variables_init survive apply_reform, not only inputs set via Simulation.set_input; (2) Holder.get_array walks up simulation.parent_branch before falling back to default, so a sub-branch (e.g. no_salt cloned from itemizing) still sees inputs set on its ancestor — the C1 fallback-to-default-only broke the country-package nested-branch pattern; (3) GroupPopulation.clone now passes the cloned population (not the source) to holder.clone, so group-entity holders on a get_branch clone point at the cloned simulation and branch-aware lookups resolve correctly. Unblocks PolicyEngine/policyengine-us#8066 (the tax_unit_itemizes integration test crashing with TypeError: int() argument ... not 'NoneType' under core 3.24.x because state_fips got wiped, plus a follow-up infinite recursion in tax_liability_if_itemizing once the state_fips wipe was resolved).

[3.24.3] - 2026-04-17

Changed

• Update CONTRIBUTING.md and README.md to document the towncrier changelog.d/ workflow. The old changelog_entry.yaml + make changelog flow was deprecated some time ago; the contributor docs still instructed new contributors to use it, so PRs kept adding changelog_entry.yaml that the PR CI's "Check changelog fragment" step rejects.

[3.24.2] - 2026-04-17

Fixed

• Guard all _fast_cache mutation sites in Simulation against the attribute being missing. Simulation.__init__ initialises self._fast_cache = {}, but country-package subclasses (e.g. policyengine_uk.Simulation) legitimately override __init__ without calling super().__init__, so set_input, delete_arrays, and purge_cache_of_invalid_values were raising AttributeError on those subclasses.
• Compare assert_near operands at float32 when one operand is already float32, otherwise at float64. Keeps the H6 fix that catches dollar-level differences on multi-million-dollar float64 values, without surfacing float32 storage rounding (e.g. 8.91 stored as float32 vs the Python literal 8.91) as spurious test failures in YAML tests against float-typed Variables.

[3.24.1] - 2026-04-17

Fixed

• Add clear_parent_cache and mark_as_modified to ParameterScale so Parameter.update() no longer raises AttributeError when backdating parameters that live inside a scale bracket. Regression surfaced by policyengine_uk.backdate_parameters(... "2015-01-01") after the _fast_cache work in #448/#436.

[3.24.0] - 2026-04-17

Added

• Added ruff check linting configuration with E and F rules to catch common Python errors.
• Support Python 3.9 and 3.10 (in addition to 3.11–3.14). On Python 3.9, NumPy is pinned to <2.1.
• Support for Python 3.14.

Changed

• Upgrade breakdown enum mismatch from WARNING to ValueError for early detection of parameter key errors.
• Migrated from changelog_entry.yaml to towncrier fragments to eliminate merge conflicts.
• Migrate from pip to uv for dependency management in Makefile and CI workflows.
• Optimize simulation calculate with fast cache and vectorial parameter lookups.
• Switched code formatter from black to ruff format.

Fixed

• Simulation.apply_reform now invalidates every cached value (_fast_cache, in-memory holder storage, on-disk holder storage, invalidated_caches) and cascades the invalidation into every branch. Previously those caches still contained pre-reform values, so subsequent calculate calls returned stale values even though the tax-benefit system had been mutated.
• Compare values in float64 instead of float32 inside assert_near and tools.assert_near. Previously both operands were downcast to float32 before the diff, so values above ~16M silently lost precision and tests expecting, e.g., 25_000_001 happily passed on 25_000_000. Note: one existing tax-scale test relied on the 0.3-is-representable-in-float32 accident and has been updated to request a tolerance of 1e-10.
• Pass minlength=self.count to numpy.bincount in GroupPopulation.sum (no-role branch) and GroupPopulation.nb_persons (no-role branch) so the result always has one cell per entity. Previously, when the highest-indexed entity had zero members, bincount returned a shorter array that silently misaligned downstream broadcasting.
• Fix clone() to update parent references and clear_parent_cache to invalidate root node cache.
• Refuse to interpret introspection / dunder attribute lookups as dataset-key loads in Dataset.__getattr__. Previously any missing attribute (hasattr, copy.deepcopy, pickle.dumps, debugger repr) triggered a real H5 / CSV load — leaking file handles, triggering network downloads for FLAT_FILE + url pipelines, and crashing during serialization with cryptic errors. __getattr__ now raises AttributeError for names starting with _.
• Raise a descriptive ValueError when calling Period.get_subperiods on an ETERNITY period. Previously the call crashed with TypeError: 'float' object cannot be interpreted as an integer because size is float("inf") for ETERNITY periods, and the YEAR subperiod branch passed it straight to range(...).
• Fix _fast_cache invalidation bug in set_input and add cache tests.
• Use a stable hash (SHA-256) when seeding numpy.random from situation inputs and variable names, so random() produces reproducible results across Python processes regardless of PYTHONHASHSEED. Also sort keys when hashing situation inputs so equivalent situations built in different dict orders produce the same seed.
• Stop Holder.get_array from silently returning values stored under an arbitrary branch when the requested branch has no value for the period. The fallback now only returns the default branch's value (or None if that is also empty), preventing reform vs. baseline cross-contamination. Also pass branch_name through the auto-carry-over path and sort known periods temporally (by period.start) instead of lexicographically.
• Drop the Singleton metaclass from SimulationMacroCache. Each simulation now constructs a fresh cache instance instead of mutating a process-wide one, so cache_folder_path / cache_file_path from one simulation can no longer overwrite another's, and country_version is read from the actual TaxBenefitSystem currently being cached (instead of whichever one constructed the singleton first).
• Fix invalid YAML in push workflow (with: and token: on the same line) that was causing all post-merge runs to fail with zero jobs, blocking version bumps and PyPI publishes.
• Isolate reform parameters from the baseline by cloning the parameter tree and resetting _parameters_at_instant_cache when a Reform is constructed. Previously both were shared by reference, so any direct self.parameters.path.update(...) in Reform.apply() mutated the baseline in place and stale cached ParameterNodeAtInstant objects kept returning the baseline's pre-reform view.
• Fix several surgical Medium-severity bugs: @uprated now uses numpy.all(old == 0) instead of Python all(old) when deciding whether to fall back to the previous formula (bug M1); single-point axes in SimulationBuilder no longer divide by zero (bug M8); Dataset.download(release://org/repo/tag/file-with/slashes) parses file paths containing slashes correctly (bug M10); every requests.get/post in the core now has an explicit timeout (bug M11); and the default datetime.date variable value is a fixed datetime.date(1970, 1, 1) instead of the timezone-dependent datetime.date.fromtimestamp(0) (bug M13).
• Respect branch_name in InMemoryStorage.delete and OnDiskStorage.delete. Previously these wiped every branch's data regardless of the requested branch_name (either entirely when period was None, or by period containment only when period was set), so reform cache invalidation also cleared the baseline's cache.
• Correctly subtract parameter values when a subtracts list element refers to a parameter (previously the parameter value was added instead of subtracted, making the formula off by 2x the parameter value).
• Key _tax_benefit_system_cache in policyengine_core.tools.test_runner on the baseline object itself (via weakref.WeakKeyDictionary) instead of id(baseline). CPython reuses object ids after GC, so a collected baseline could hit a stale cache entry belonging to a completely different TBS, silently sharing test setup across unrelated baselines.
• Migrate push workflow to use GitHub App token (APP_ID / APP_PRIVATE_KEY) instead of the expired POLICYENGINE_GITHUB PAT, so the versioning job can push the "Update PolicyEngine Core" commit that triggers Test + Publish. Matches the pattern already used by policyengine-us, policyengine-api, and several other PolicyEngine repos.

Removed

• Support for Python 3.10 (following SPEC 0 policy).

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

3.23.6 - 2026-01-25 14:03:25

Fixed

• Fixed pandas 3.0 compatibility in ParameterNodeAtInstant.getitem() by converting pandas StringArray to numpy array before using for fancy indexing (fixes

3.23.5 - 2026-01-25 04:19:17

Fixed

• Fixed pandas 3.0 compatibility in Enum.encode() by using positional access (.iloc[0]) for pandas Series instead of label-based access (array[0]), which fails with KeyError when Series has a non-integer index (fixes

3.23.4 - 2026-01-24 15:41:39

Fixed

• Fixed pandas 3.0 compatibility issues with StringDtype and StringArray

3.23.3 - 2026-01-17 18:14:48

Fixed

• Handle empty HUGGING_FACE_TOKEN gracefully in CI environments. Empty strings are now treated the same as None, and non-interactive environments (like CI) return None instead of hanging on getpass prompt.

3.23.2 - 2026-01-15 22:17:08

Added

• Documentation for Reform.from_dict() method with examples covering basic reforms, multiple parameters, bracket syntax, period formats, and Microsimulation integration.

3.23.1 - 2025-12-14 23:36:30

Fixed

• Optimisation improvements for loading tax-benefit systems (caching).

3.23.0 - 2025-12-03 22:11:27

Changed

• Invalid enum values now raise ValueError instead of logging a warning and returning index 0. This prevents silent data corruption when incorrect enum strings are passed to simulations.

3.22.2 - 2025-12-03 01:03:13

Fixed

• Allow NumPy 2.3+ for Python 3.14 compatibility (fixes temporary elision bug causing incorrect microsimulation results)

3.22.1 - 2025-12-02 15:43:09

Fixed

• Issue of running on windows because of not having UTF-8 encoding.

3.22.0 - 2025-12-01 15:19:17

Changed

• Optimised enum encoding using searchsorted instead of np.select.
• Optimised empty_clone using object.new() instead of dynamic type creation.
• Added lru_cache to period and instant string parsing.
• Vectorised random() function using PCG hash instead of per-entity RNG instantiation. Note that this changes random value sequences - simulations using random() will produce different (but still deterministic) values.
• Added warning logging for invalid enum string values during encoding.

3.21.0 - 2025-11-27 16:18:36

Added

• Subdirectory support for hf:// URLs (e.g., hf://owner/repo/path/to/file.h5).
• Google Cloud Storage support with gs:// URL scheme (e.g., gs://bucket/path/to/file.h5).

3.20.1 - 2025-10-01 17:01:19

Fixed

• NumPy 2.x structured array dtype compatibility when using axes with parameters that have different field subsets (e.g., ACA rating areas)

3.20.0 - 2025-08-12 15:53:31

Added

• Easy way to update parameters with uprating behaviour.

3.19.4 - 2025-07-28 09:17:29

Fixed

• UK microsimulation class handling.

3.19.3 - 2025-07-24 18:10:45

Added

• regression test ensuring base Projector.transform raises

Fixed

• Projector.transform now raises NotImplementedError

3.19.2 - 2025-07-24 17:03:30

Fixed

• Fix NumPy 2.1.0 random seed overflow issue by ensuring seeds are always non-negative

3.19.1 - 2025-07-24 14:00:15

Fixed

• Add test retry mechanism to handle intermittent CI failures

3.19.0 - 2025-07-23 00:48:51

Changed

• added support for python 3.13.0
• upgraded dependency to numpy 2.1.0

3.18.0 - 2025-07-22 20:16:08

Changed

• Update microdf_python dependency to >=1.0.0.

3.17.1 - 2025-07-12 14:22:53

Fixed

• Bug causing to_input_dataframe to include bad time periods.

3.17.0 - 2025-06-05 20:14:47

Added

• Default input and output periods to Simulation constructor.

3.16.6 - 2025-04-09 09:40:30

Added

• Error handling for HF downloads.

3.16.5 - 2025-04-02 09:54:58

Fixed

• Subsampling logic.

3.16.4 - 2025-03-25 15:06:42

Fixed

• Bug causing some reforms to fail.

3.16.3 - 2025-02-06 15:38:10

Fixed

• Fix file writing on windows.

3.16.2 - 2025-01-28 07:08:07

Fixed

• Bug in parametric reforms.

3.16.1 - 2024-12-20 12:56:31

Changed

• Explicitly set local directory when downloading datasets from Hugging Face

3.16.0 - 2024-12-18 15:12:25

Changed

• Refactor all Hugging Face downloads to use the same download function
• Attempt to determine whether a token is necessary before downloading from Hugging Face
• Add tests for this new functionality

3.15.0 - 2024-12-05 14:47:01

Added

• Chained reform handling in TaxBenefitSystems.

3.14.1 - 2024-12-02 16:35:14

Changed

• Replaced coexistent standard/HuggingFace URL with standalone URL parameter
• Fixed bugs in download_from_huggingface() method
• Create utility function for pulling HuggingFace env var

3.14.0 - 2024-11-29 16:08:25

Added

• Support for HuggingFace-hosted H5 file inputs.

3.13.0 - 2024-11-27 13:12:44

Added

• HuggingFace upload/download functionality.

3.12.5 - 2024-11-20 13:13:13

Changed

• update the psutil requirement to version 6
• update the furo requirment to <2025
• update the markupsafe requirement to <3

3.12.4 - 2024-11-11 14:20:12

Fixed

• Datasets writing downloaded data now use an atomic_write to write it to disk. This prevents other processes attempting to read a partial file or clobbering each other.

3.12.3 - 2024-11-04 16:29:34

Fixed

• Bug in labour supply responses.

3.12.2 - 2024-11-01 21:39:35

Added

• Compatibility settings for editable installs

3.12.1 - 2024-11-01 11:36:53

Fixed

• Bug causing Enums to fail in some simulations.

3.12.0 - 2024-10-30 18:46:15

Changed

• update the ipython requirement to version 8

3.11.1 - 2024-10-29 20:04:02

Changed

• Replace custom implementation of microdf with deployed version

3.11.0 - 2024-10-24 11:04:12

Fixed

• Bug causing subsample to not work in some situations.

3.10.0 - 2024-10-17 15:59:10

Added

• Two tests related to extensions that were previously removed

Changed

• Shallow copy GroupEntities and PopulationEntity when cloning TaxBenefitSystem object

3.9.0 - 2024-10-09 20:29:35

Changed

• Shallow copy entities between TaxBenefitSystem objects when cloning

3.8.2 - 2024-10-01 19:33:10

Changed

• Updated README.md

3.8.1 - 2024-09-27 10:43:12

Changed

• Set test runner's default period as underlying simulation's default period
• Prevented crashing when absolutely no date is provided anywhere for tests

3.8.0 - 2024-09-26 12:00:25

Added

• Randomness based on entity IDs as seeds.
• OpenFisca-Core imports.

3.7.1 - 2024-09-26 00:13:13

Changed

• Renamed sim_macro_cache.py to simulation_macro_cache.py

3.7.0 - 2024-09-24 18:10:25

Added

• Simulation subsampling.

3.6.6 - 2024-09-09 13:52:47

Changed

• Disable macro cache by default.

3.6.5 - 2024-09-04 21:19:20

Fixed

• Corrected logic with Simulation.check_macro_cache to prevent undesirable caching

3.6.4 - 2024-09-03 18:42:55

Changed

• Ensure that every instance of 'Simulation' class has 'dataset' attribute, even if equal to 'None'
• Add better safeguards to 'check_macro_cache' method of 'Simulation' class

3.6.3 - 2024-09-02 22:01:19

Fixed

• Syntax error in Simulation class

3.6.2 - 2024-09-02 13:04:21

Added

• Added class SimulationMacroCache for macro simulation caching purposes.

3.6.1 - 2024-08-31 16:13:07

Added

• Added class SimulationMacroCache for macro simulation caching purposes.

3.6.0 - 2024-08-28 16:41:26

Fixed

• Bugs in typing and Dataset saving.

3.5.3 - 2024-08-21 14:34:52

Changed

• Fixed script to initialize a reform within the simulation

3.5.2 - 2024-08-17 19:28:38

Changed

• change mypy version from 1.11.1 to <2

3.5.1 - 2024-08-08 15:17:05

Fixed

• fixed bug where uprating was causing nan.

3.5.0 - 2024-08-08 15:16:12

3.4.0 - 2024-08-07 12:42:36

Added

• Flat file export option to simulations.

3.3.0 - 2024-08-02 09:09:51

Added

• Simulation loading from dataframes.
• Simulation start_instant attribute.

3.2.0 - 2024-07-28 19:28:05

Changed

• Reform syntax to increase flexibility.

3.1.0 - 2024-07-26 00:28:31

Added

• Flat file dataset option.
• Microdata weighting classes from microdf.
• Eternity dictionary reform options.

3.0.0 - 2024-06-26 17:08:24

Changed

• Added support for Python 3.12

2.23.1 - 2024-06-06 09:09:41

Added

• V
• i
• s
• u
• a
• l
• i
• z
• a
• t
• i
• o
• n
• o
• p
• t
• i
• o
• n
• w
• h
• e
• n
• r
• u
• n
• n
• i
• n
• g
• t
• e
• s
• t
• s

2.23.0 - 2024-06-04 19:43:13

Added

• max_value and min_value in Variable class.

2.22.0 - 2024-05-31 11:12:13

Fixed

• Bug causing reforms affecting data to not affect caches.

2.21.8 - 2024-05-14 15:33:07

Fixed

• Uncap coverage.

2.21.7 - 2024-05-14 15:27:27

Fixed

• Remove name metadata tag in docs.

2.21.6 - 2024-05-13 16:26:53

Changed

• Replaced unsafe numpy-Python comparison with use of numpy dtype to convert byte-string arrays to Unicode ones within enums

2.21.5 - 2024-05-08 15:07:51

Fixed

• Bug in macro cache logic.

2.21.4 - 2024-05-08 13:39:53

Added

• Added .vscode and venv to .gitignore file.

2.21.3 - 2024-05-08 13:36:00

Added

• Dependency upgrade for pytest, from 7 to 8.

2.21.2 - 2024-05-08 08:44:48

Fixed

• Generation of enum-typed variables

2.21.1 - 2024-05-07 15:44:58

Fixed

• Bug in macro caching logic.

2.21.0 - 2024-05-07 12:16:05

Added

• Ability to turn off read and write features in the macro cache.

2.20.0 - 2024-05-06 14:24:08

Added

• Macro impact caching.
• Dictionary-input start-stop date reform handling.

Fixed

• Uprating bugs.

2.19.2 - 2024-05-06 12:47:34

Added

• IPython to dependencies

2.19.1 - 2024-04-29 15:06:58

Fixed

• FutureWarning issue with bools and enums.

2.19.0 - 2024-04-16 12:51:42

Added

• Baseline parameter node.

2.18.0 - 2024-03-22 10:42:19

Added

• Parameters can now be uprated based upon fixed cadences

2.17.1 - 2024-03-05 13:01:42

Fixed

• Breakdown-added parameters no longer precede uprating indices.

2.17.0 - 2024-03-04 11:27:40

Changed

• Add method for propagating units to descendants for ParameterScale objects

2.16.1 - 2024-02-29 16:03:09

Fixed

• Fixed a bug in Core monthlyisation.

2.16.0 - 2024-02-14 14:49:15

2.15.0 - 2024-02-08 14:32:38

Added

• Variable inheritance in declarations.

2.14.0 - 2024-02-06 17:28:06

Changed

• A variable must have a label.
• If a variable has formulas then it must not have adds/subtracts, and vice versa.

2.13.1 - 2024-02-06 11:28:22

Added

• Improvements to error handling in bad variable declarations.

2.13.0 - 2024-01-02 15:26:52

Added

• Simulation helper to extract individual households from a microsimulation.

2.12.1 - 2023-12-27 13:28:08

Fixed

• API to reform functionality made more editable.

2.12.0 - 2023-12-26 18:31:31

Added

• API ID to reform object functionality.

2.11.5 - 2023-12-21 10:55:47

Fixed

• Randomness issue.

2.11.4 - 2023-12-19 11:12:24

Added

• Tools for branching

2.11.3 - 2023-12-16 15:34:37

Fixed

• Random seed fixed for each simulation.

2.11.2 - 2023-12-14 11:38:52

Fixed

• Bug causing random behaviour differences between situations with and without axes.

2.11.1 - 2023-11-30 18:45:01

Fixed

• Bug causing subtracts to not accept strings.

2.11.0 - 2023-11-22 15:55:52

Fixed

• Reforms always apply before parameter utilities.

2.10.0 - 2023-11-22 15:05:43

Fixed

• Reforms always apply before parameter utilities.

2.9.0 - 2023-11-12 13:23:23

Added

• PolicyEngine API integration for reforms.

2.8.2 - 2023-10-09 17:52:02

Changed

• Unpinned lower NumPy pin.

2.8.1 - 2023-10-05 13:24:20

Fixed

• Bug causing YAML tests to not populate Simulation.input_variables.

2.8.0 - 2023-10-05 11:41:53

Added

• Support for adds/subtracts targeting parameters.
• Support for adds/subtracts targeting year ranges where no parameters exist.

Changed

• Deprecated sum_of_variables.

2.7.1 - 2023-10-05 11:00:40

Fixed

• Fix bug affecting microsimulation runs in countries which use automatic period adjustments.

2.7.0 - 2023-10-04 19:05:58

Added

• Automatic period adjustment helper functionality.

Changed

• Default error threshold for tests widened to 1e-3.

2.6.0 - 2023-10-04 14:51:40

Fixed

• Bump numpy version
• Fix numpy record compatibility issue

2.5.0 - 2023-08-02 12:54:44

Added

• Chart formatting utilities.

2.4.0 - 2023-07-21 14:55:15

Added

• Chart formatting utilities.

2.3.0 - 2023-06-30 08:22:27

Added

• Optional argument for saving datasets to a specific file path.
• Safety check for Path objects not being strings.

2.2.1 - 2023-06-02 15:22:29

Fixed

• A bug causing variable not found errors to sometimes fail to throw.

2.2.0 - 2023-05-31 11:11:09

Fixed

• Bug causing unhelpful errors when variables are not found.

2.1.1 - 2023-05-24 15:11:39

Fixed

• Fixed pytest deprecated usages since version 7, and fixed a bug that prevented new package from publishing.

2.1.0 - 2023-05-23 13:22:04

Added

• Python 3.10 support.

Fixed

• PyTest syntax updated.

2.0.4 - 2023-05-23 07:35:11

Fixed

• Bug causing simulations containing multiple group entities and axes to fail to simulate.

2.0.3 - 2023-04-14 10:06:10

Changed

• Removed node homogenuity check for speed.

2.0.2 - 2023-03-31 18:27:52

Changed

• Added ability to pull datasets from GitHub releases.

2.0.1 - 2023-03-23 08:55:46

Fixed

• A bug where public datasets would fail to download.

2.0.0 - 2023-03-23 02:37:29

Changed

• Datasets API simplified and made more consistent.

1.12.4 - 2023-03-08 21:35:17

Fixed

• Removed bad dependency fixes.

1.12.3 - 2023-03-04 22:58:51

Fixed

• Branches stay persistent between variables.

1.12.2 - 2023-02-09 21:59:16

Added

• Option for loading non-year-specific datasets.

1.12.1 - 2023-01-28 15:58:38

Added

• Abolition parameters for each variable under gov.abolitions.

1.12.0 - 2023-01-27 22:28:10

Added

• Likely high time savings for simulations making heavy usage of reforms and branches.

1.11.4 - 2023-01-11 23:44:12

Fixed

• Labels and names aren't propagated in parameters.

1.11.3 - 2023-01-10 17:32:23

1.11.2 - 2023-01-04 13:05:49

Added

• Error message for an unknown variable.

1.11.1 - 2023-01-04 12:51:00

Added

• A metadata option to specify which policies are modelled and not modelled.

1.11.0 - 2023-01-03 20:50:02

Added

• Adds/subtracts option for parameter names.

1.10.24 - 2023-01-02 20:40:49

Changed

• Pin all package versions.

1.10.23 - 2022-12-18 06:12:52

Fixed

• Fix incorrect type annotation in policyengine_core/variables/variable.py

1.10.22 - 2022-12-18 06:07:32

Added

• Fixed mypy errors in several files.

1.10.21 - 2022-12-17 21:51:59

Changed

• Removed seemingly unused "N_" function

1.10.20 - 2022-12-17 16:36:53

Fixed

• Fixed mypy errors in policyengine_core/tracers/tracing_parameter_node_at_instant.py

1.10.19 - 2022-12-17 09:23:10

Fixed

• Fixed mypy errors in tests/core/test_entities.py

1.10.18 - 2022-12-15 15:27:35

Fixed

• Bugs with Enum inputs.

1.10.17 - 2022-12-14 15:33:46

Added

• Added basic_inputs, a metadata field for the basic inputs needed to get a country model running.

1.10.16 - 2022-12-13 23:40:47

Added

• Added mypy to our test suite to prevent new type regressions

1.10.15 - 2022-12-12 15:28:47

Added

• Hidden input field for variables.

1.10.14 - 2022-12-11 23:52:13

Fixed

• Added documentation on populating changelog entries. Also created a CONTRIBUTING.md symlink to make contributing guidelines easier to find.

1.10.13 - 2022-12-11 23:46:33

Fixed

• Git config now ignores JetBrains IDE config files

1.10.12 - 2022-12-07 13:41:29

Fixed

• Uprating bug affecting the UK.

1.10.11 - 2022-12-07 12:51:49

Fixed

• Include scale brackets in descendants.

1.10.10 - 2022-12-05 16:21:04

Fixed

• Removed warning on Enum decoding.

1.10.9 - 2022-12-03 15:21:15

Fixed

• Bug with Enum decoding.

1.10.8 - 2022-12-02 18:11:10

Fixed

• Singleton files have a shorter moduleName.

1.10.7 - 2022-11-30 20:33:21

Fixed

• Various safety checks.

1.10.6 - 2022-11-30 15:19:56

Fixed

• README.md files in parameter trees now inform metadata.

1.10.5 - 2022-11-28 10:37:23

Added

• Functionality for getting all known branch-period caches.

1.10.4 - 2022-11-27 16:16:29

Fixed

• Removed debug lines.

1.10.3 - 2022-11-27 16:08:33

Fixed

• Fixed a bug causing Windows tests to fail.

1.10.2 - 2022-11-27 14:33:53

Fixed

• Bug in defined-for causing it to be ignored.
• Branched simulations now cache variables specifically for the branch and time period, not sharing with the parent simulation.

1.10.1 - 2022-11-24 00:33:05

Added

• Documentation page on reforms.

Fixed

• Moved the uprating logic to after formulas are checked.

1.10.0 - 2022-11-24 00:15:28

Added

• Variables can now specify an 'uprating' parameter.
• Variables can now specify 'adds' and 'subtracts' variable dependencies.
• Variable folders are now automatically stored in a tree.

Fixed

• Bugs in cloned simulations.

1.9.0 - 2022-11-08 15:00:26

Fixed

• Various fixes for PolicyEngine's server app.

1.8.1 - 2022-11-02 01:31:48

1.8.0 - 2022-10-23 13:41:39

Changed

• Parameter scale components now included in get_descendants().

Fixed

• Folder paths in Datasets are parsed as Paths.
• Empty input: fields are parsed as {}.

1.7.0 - 2022-10-21 12:39:50

Added

• Metadata dictionary input to Variables.

1.6.0 - 2022-10-19 22:44:19

Added

• Derivative calculation in simulations.

1.5.0 - 2022-10-19 16:01:51

Added

• Branched simulations.

1.4.1 - 2022-10-18 18:13:09

Added

• Plotly to dependencies

1.4.0 - 2022-10-18 15:28:22

Added

• Performance improvements
• Import sorting and code health improvements

Fixed

• Axes now work in simulations

1.3.0 - 2022-10-14 17:18:21

Added

• Tests and documentation.

1.2.0 - 2022-10-14 10:14:40

Added

• Parameter operations from openfisca-tools
• Dataset classes from openfisca-tools

1.1.3 - 2022-10-12 18:08:43

Fixed

• Git blame reassigns OpenFisca-Core authors.

1.1.2 - 2022-10-12 16:25:33

Fixed

• Bug in the country template microsimulation class failing to load data.

1.1.1 - 2022-10-12 15:58:34

Added

• Behaviour to handle default datasets.

1.1.0 - 2022-10-12 15:33:59

Added

• Microsimulation features
• Documentation for the CLI interface

1.0.6 - 2022-10-09 23:00:35

Fixed

• Python 3.9 used in all GitHub actions.

1.0.5 - 2022-10-09 22:55:14

Added

• Test that documentation builds

1.0.4 - 2022-10-09 22:33:24

Added

• Type hints to most functions and classes.
• Jupyter book documentation for all modules.

1.0.3 - 2022-10-02 16:59:49

Fixed

• Build action always runs.

1.0.2 - 2022-10-02 16:45:22

Fixed

• Push action doesn't attempt to publish nonexistent documentation.

1.0.1 - 2022-10-02 16:38:34

Fixed

• Version update action.

1.0.0 - 2022-10-02 16:35:17

Changed

• Non-essential (for Python simulations) code removed.

[0.1.0] - 2022-10-02 16:02:00

Added

• OpenFisca-Core forked.