Change log

0.2.0 (May 15, 2026)

• Changes

  • update versions of jax (0.9.0.1) (0.7.1), jaxlib (0.9.0.1) (0.7.1), ortools (9.15.6755), networkx (3.4.2), nlopt (2.7.1) , openpyxl (3.1.5), scipy (1.15.3), most notably numpy (2.4.4) and pandas (3.0.2) .

  • AUTODIFFED_JACOBIAN and AUTODIFFED_HESSIAN are now turned off by default.

  • method calibrate() was enriched with argument validating following the open-closed-principle.

  • _svars variables (undocumented) henceforth mobilize injected variables in lieu of their upstream counterpart.

  • recursively defined entities can henceforth also be vectorized (!!!)

  • methods _distributor() and its derivatives now rely on jax.jit’s argument static_argnames in lieu of static_argnums (internally deprecating _distributor_sargnums in favor of _distributor_sargnames, both undocumented).

  • grand masks are now computed contingently (see _compute_residuals_grand_masks()), improving the overall algorithm laziness.

  • class attribute _optim_otol_key (undocumented) was changed (from ‘f₄’) to ‘f⁴’.

  • class DSpace now exposes its Serializer instance.

  • saturation variables are now explicitly instantiated for the sake of design consistency.

  • economic attribute excises have been restricted to the positive real line, which can still be manually amended if needed.

  • attribute _code_objects was renamed as code_objects_getter() turned into a functional and instance-method _code_objects_getter() (undocumented) was turned into class method.

  • methods ~._compute_residuals_grand_mask and ~._compute_residuals_grand_masks were renamed into _compute_residuals_grand_mask() and _compute_residuals_grand_masks() respectively.

  • argument okey of method _optim_specifier() was removed.

  • method _xmains_definjecter was renamed into _definjecter() and turned into (caching) instance method.

• Breaking changes

  • constraint _spcmgs_tots_nullity_ was removed, which means that users should henceforth explicitly be willing for such a control by implying InitiallyBalancedSM via their library.

  • instance methods _minimization_coargumenter() and _hvars_getter() now takes a sequence of systems as argument instead of only one.

  • methods _distributor(), _distributors(), _distributor_jacobian(), _distributor_jacobianl(), _distributor_hessian(), _distributor_hessianl(), _distributor_hessianl_lws() and _distributor_derivatives() henceforth take positional-only argument(s).

  • loss_collapser() and its (undocumented) underliers (such as e.g. _LossCollapser) was unimplemented.

  • rename method _abounds_computer into _mabounds_computer().

  • method ___update_states___() (undocumented) becomes keywords-only.

  • method net_operating_surplus_computer of classes ImplicitEntity and BalanceOfTrade was removed in favor of net_surplus_computer and net_surplus_computer.

  • circular definitions are no longer at work (e.g. deflators_fisher, deflators_laspeyres and deflators_paasche).

  • instantiation parameter rprem of class DynamicStock was renamed as grpot.

  • class UMarket was renamed into UUMarket, while UMarket still exists but does not embark ‘unemployment uniformity’ any longer.

  • classes OImplicitLeontief and MarshallianILeo now inherits from Ordinaler instead of Investor.

• Bug Fixes:

  • _arrays_mdata (undocumented) did not contain the correct (boolean) mask type information, which prevented the correct symbol from being assigned during Symer-base enrichment.

  • the IAMAX-interfaced sparse feature of IPOPT was actually not restricting the set of Lagrangian coefficients to its very structure, e.g. inequality with no equality constraint.

• New library’s features

  • add method _factor_pow_increaser().

  • add class InitiallyBalancedSM.

  • add method _multiply(), also aliased by _safe_by_0_mul(). For the sake of consistency, _divide() is henceforth aliased by _safe_by_0_div().

  • add methods _add_ufunc() and _sum().

  • add classes LaspeyresianGFL and L1NzIntDrivenDemands.

  • add class CoInitiatedVolumes.

  • add class BalancedSM.

  • add argument ioq_shares_drates2 to compute_calibrand_0_star().

  • add class UniformGhoshAllocationPremia.

  • enrich classes ImplicitEntity and BalanceOfTrade with methods net_operating_surplus_computer() and net_operating_surplus_computer() respectively.

  • add classes ImplicitIncome and ImplicitTransitories.

  • enrich classes ImplicitEntity and BalanceOfTrade with methods other_costs_computer() and other_costs_computer() respectively.

  • enrich ArraysDealer with _weighter().

  • add classes ImplicitLeontief, AdditiveOrdinaler, OImplicitLeontief, HicksianRGAFCES, HicksianPICES, AddedOPCosts, UniformPrice.

  • class UniformUnemployment now has an instantiation argument, namely price_transmitting, as well as a new compute_calibrand_0() method.

  • class DynamicStock has an additional instantiation argument named dyear0.

  • add class Ordinaler.

• New conf’s features

  • add ENABLED_SCALING and AUTODIFFED_JACOBIAN editable under the same name (prior to module loading) via os.environ.

  • add LOSSES_FACTOR editable under the same name (prior to module loading) via os.environ.

  • add RELATIVE_OCOSTS editable under the same name (prior to module loading) via os.environ.

  • add UNCLUSTERED_USER_METHODS editable under the same name (prior to module loading) via os.environ.

  • add RELATIVE_SPCMGS editable under the same name (prior to module loading) via os.environ.

  • add NONLINEAR_SCALING editable under the same name (prior to module loading) via os.environ.

  • add LIFTED_DAG editable under the same name (prior to module loading) via os.environ.

  • add COMPILED_ARCOMPONENTS editable under the same name (prior to module loading) via os.environ.

• New core’s features

  • _ndarray_undensifier() can henceforth deal with 1-dimensional arrays as well as boolean arrays, whatever the dimensionality.

  • add post-modeling attribute _constraints_jacobian().

  • add ancillary method _iid_flattenting_identifier().

  • the output’s dimensionality of methods _distributor(), _distributor_jacobian(), _distributor_hessian(), _distributor_hessianl() and _distributor_hessianl_lws() was ‘sanitized’, which paves the way to generalization.

  • _distributor_hessianl() can now compute hessian hypermatrices of any hight indifferently (which _distributor_hessian() does intrinsically following the precited sanitization).

  • add (undocumented) ancillary methods/attributes _distributor_pkwargs(), _forced_lnrzd, etc, _batch_size.

  • add modeling method _distributor_jacobianl().

  • add modeling methods _compute_residuals_grand_masks() , _distributor_jacobianl_structurer() and _distributor_hessianl_structurer() .

  • add modeling methods _distributor_jacobianl_sp(), _distributor_hessianl_sp() and _distributor_hessianl_lws_sp().

  • add ancillary methods _jx_jit() and _soberly_computed().

  • add modeling methods _distributors(), _distributor_jacobianls() , _distributor_jacobianls_fuzzer(), _distributor_jacobianls_structurer(), _distributor_jacobianls_sp(), _distributor_hessianls(), _distributor_hessianls_lws(), _distributor_hessianls_structurer(), _distributor_hessianls_sp(), _distributor_hessianls_lws_sp(), _distributor_hessianl_sps_lws(), _distributor_lsps_lws_hessian() .

  • add method _ipopt_kwargs_maker() (undocumented) whose purpose is to explain the parameters operating when IPOPT is mobilized.

  • add modeling methods _distributor_sp_jacobianl() and _distributor_sp_hessianl().

  • enrich method minimize() with a new set of minimization methods allowing for the use of IPOPT with sparse-related features, i.e. setting argument oalgo to e.g. 'sp.mumps', 'sp.pardiso', etc.

  • add (undocumented) ancillary attribute _hessian_cpprch.

  • improve _rsaturations_computer(), i.e. stabilized, smoothed (\(C^{\infty}\)), enriched with a configurable (private) regularization offset.

  • add modeling method _distributor_ocomponents() .

  • add (undocumented) attribute _ƶvars_names and method _computers_argnames_getter().

  • add attributes _œvars_names and _ƨvars_names.

  • positivity constraints now also possess their own scaling factor.

  • add methods _systems_interpreter(), _zsystems_evaluator(), _compute_opscalers() and _set_œscalers().

  • add (undocumented) ancillary attributes _œvars_qs, __ƨvars_qs, __ƨvars_dodicts, _ƨvars_xoots, _xvars_dkind .

  • add modeling methods _distributor_decomposer(), _distributor_jacobianl_decomposer(), _distributors_fuzzer().

  • method minimize() has one new argument, salgo.

  • method _xdarray_labeller() has one new argument, force_section_frames.

  • add economic variables input_values_shares and additive_input_volumes_shares.

  • add ancillary (undocumented) attribute _sp_1mat_cq and methods _ivals_mk_cq_computer(), _ivals_mk_cq_ɗcomputer(), _coprcs_mk_cq_computer(), _ivols_mk_cq_computer().

  • improved _fancy_connecter()’s NotImplementedError error message in the unknown configuration case.

  • add method _ndarray_undtyping_booler().

  • add economic attribute ordinal_values.

  • economic attribute net_operating_surplus now has its “injecter”. Will eventually be removed in favor of an attribute located further upstream in the calculation graph.

  • add economic attribute custom_qois, aliased by transitories and user_statistics.

  • add economic attributes other_costs_bdown, other_costs, other_costs_rates

  • add ancillary attribute _dag_specs, whose content diverges from _arrays_mdata (undocumented) as UNCLUSTERED_USER_METHODS is set to '1'.

  • add method _ndarray_dtyping_negbooler().

  • input validation done at user classes’ initialization does not fail any longer.

  • pure feasibility problem, i.e. \(\text{find} \ x \ \text{s.t.} \ g(x)=0, \ h(x) \geq 0\) can now be programmed.

  • add attributes _minimization_residuals and minimization_residuals.

  • method minimize() has one new argument, rschedule.

  • add method _compose_chain() and attribute _symbolization_gain.

  • add method _type_duct_expander().

  • (dynamically augmented) user-defined classes are now picklable.

  • add method _constraints_reachability_map().

  • add ancillary attributes _identifiers_exp, _identifiers_imp, valued_econsumptions, valued_iconsumptions and valued_consumptions.

• New utils’s features

  • method otbprint() can now deal with array-like values.

  • add methods _is_static_method(), _is_class_method(), _method_input_signature().

  • add class Symer, which addition enriches the package with a symbolic component.

  • method str_sanitizer() now 1) removes "+" symbol and 2) protects greek letters from “sanitization”.

  • method _items_coercer() now deals with duplicate keys.

  • method dag_sorter() now has argument cycle_cutset.

  • try_() can henceforth return traceback.TracebackException instances as returned_v.

  • add method topological_generations_subsetter().

  • refactoring led to define method dag_maker().

  • add method dag_uncycler().

  • method namedtupler() now returns picklable instance, of their own kind though.

• New test_core’s features

  • add methods test_constraint_nth_order_linearization_reliability(), test_pub2pri_supersetting(), test_œƨvars_exhaustivity().

• New test_utils features

  • add methods test_is_static_method(), test_is_class_method(), test_method_input_signature().

  • add class TestSymer.

---

0.1.1 (August 10, 2024)

• Changes

  • update versions of jax (0.4.31) and jaxlib (0.4.31).

  • the minimum Python version is now 3.10. 3.10 will remain the minimum supported version until July 2025.

  • MarshallianGFCES has one additional instantiation argument, negated_mask.

• New conf’s features

  • add CSCALINGF_LOGBASE, REGRESSED_IGUESS and REGRESSED_SCALING editable under the same name (prior to module loading) via os.environ.

• New core’s features

  • the solver can now be provided with scaling factors, following the XUNK_SF_CNAME parameter format. Note that such possibility is henceforth provided for CUNK_0D_SYM-, RUNK_0D_SYM- and RUNK_1D_SYM-specified values.

  • add method _minimization_jacobian_computer().

  • add ancillary attribute _œvars_names.

• New library’s features

  • add classes Passive, IOnBoard, HicksianGAFCES , HicksianGFLUMRS AQConserver.

• New utils’s features

  • add method _mdefargs_collecter().

• Bug Fixes:

  • sheets named similarly to private variables were incorrectly causing _ndarray_reader() to raise a ValueError.

  • improve dtypes inference during the vectorization of the users’ classes’ instantiation arguments. Mixed types were erroneously returned as object, which behavior clearly was (numpy- and) jax-unfriendly.

---

0.1.0 (April 28, 2024)

• Breaking changes

  • library’s TanhCurvedNetPrice’s methods (still structure-dependent) now require the deflator of interest to be the first component of the passed arrays.

  • sheet names "direct_taxes", "corpo_income_taxes_rates" and "indiv_income_taxes_rates" are temporarily restricted until the circularity between taxes and transfers has been tackled.

  • economic attribute income_public_circuit_leoninv was removed until its implication could also address issues of circularity in transfers.

  • Income public/private circuits data must be specified in the transposed version of what they were until now.

  • update versions of attrs (23.1.0), anytree (2.12.1), jax (0.4.26), jaxlib (0.4.26), joblib (1.3.2), networkx (3.3), numpy (1.26.4), pandas (2.1.4), scipy (1.13.0), tabulate (0.9.0), unidecode (1.3.7), xlrd (2.0.1) and xmltodict (0.13.0).

  • iamax is now compatible with all Python versions (>3.8) and does not require OS-specific installation procedures.

  • Python 3.8 support has been dropped.

• New conf’s features

  • add EXOGENOUS_MRENTS, editable under the same name (prior to module loading) via os.environ. To be set to 1 for, e.g., backward compatibility. Indeed, data supporting markets_rents (via the sheet mapped to markets_rents) at instantiation can be seen as superfluous since the underlying information is inferrable from input_values (provided via the sheet mapped to input_values) and other variables. In such a case, this would results in 1) the removal of the variable in question from the list of those considered as exogenous and 2) the augmentation of the example folder’s XLSX file (i.e. ut(KLEM-BRICS-202006).xlsx) with the relevant set of zero growth constraints. Having EXOGENOUS_MRENTS ensures the backward compatibility allows for such an option to be chosen.

  • add JITTED_OBJECTIVE, JITTED_DERIVATIVES, ( CONSTRAINED_CALIB, AGGREGATED_CONSTRAINTS) ENABLED_DENSIFICATION, ENABLED_SERIALIZATION, COMPLEX_VALUED_UNDERLIERS, HOISTED_CONSTANT_INDEXERS, ENABLED_VECTORIZATION, BROADEST_SYSTEM, TIME_UNBASED_DATA_HASHES, AUTODIFFED_HESSIAN, AVERAGED_LOSSES, OMITTED_INEQ_CONSTRAINTS and LOSSES_EXPONENT, editable under the same name (prior to module loading) via os.environ.

  • add RELATIVE_STAXES, RELATIVE_CTAXES, RELATIVE_MRENTS, DETAILED_XCISES, RELATIVE_TRAMGS and DETAILED_TRAMGS editable under the same name (prior to module loading) via os.environ.

• New core’s features

  • _code_objects_getter() (undocumented underlier of _code_objects) now simplifies fancy indexes as possible.

  • add economic variables ordinal_prices, ordinal_volumes, with the latter resulting (for the sake of completeness) in the augmentation of some preexisting library’s class(es), such as Investor, as well as modifications within country-specific KLEM XLSX files.

  • add (economic but) ancillary variables input_values_transpose and ordinal_prices_transpose. Having such definitions poses the question of their dynamic (definition and) invocation.

  • _libobj_preparator() can now handle property getters.

  • reimplement MSystem’s __getattr__ method for undocumented attribute retrieval.

  • introduce the capability to handle simultaneously/correlated resolved trade-offs. This feature becomes essential when dealing with portfolio allocation trade-offs wherein other (orthogonal) assets specificities must be considered.

  • introduce the capability of having instantiation-only computers and rooters, whose rule of identification consists in them being suffixed by <>_crooter or <>_ccomputer.

  • add economic attribute idle_rates.

  • the solver can now be provided with initial guesses, following the XUNK_IG_CNAME parameter format. Such possibility is only provided for CUNK_0D_SYM-specified values though, triggering a NotImplementedError error otherwise.

  • inferred (e.g. via numpy.ptp) initial guesses cannot be out of bounds anymore.

  • _optim_specifier() henceforth offers (2) solver-initiation strategies.

  • add attributes _loc_block_blueprint, _valued_blocks_ilocs and _stacked_blocks, prerequisites for the drastic optimization that makes up this new version.

  • add ancillary attributes _identifiers_left_levels_range and _identifiers_right_levels_range.

  • add ancillary attributes _loc_sparsity_map, _loc_sparsity_map_flatten and _ispaces_specs.

  • following the definition of ENABLED_DENSIFICATION, add methods _ndarray_densifier() (, _2darray_raveling_densifier() undocumented), _ndarray_undensifier(), _ispaces_specs_getter_umpire(), _d2s_shapes_converter(), _s2d_coords_converter() and attribute _densification_gain.

  • following the just mentioned new methods definitions, add (undocumented) attributes/methods _sm_lixs, _sm_lmsk, _sp_sm_lixs_line, _sp_sm_ixs_line, _dot_lm_lixs_line, _dot_lm_lmsk_liner(), _dot_rm_lixs_line, _dot_om_lixs_line, _sp_dot_lm_lixs_liner(), _sp_dot_rm_lixs_liner(), _sp_dot_om_lixs_liner(), _sp_lixs_lines, _sp_lm_lixs_mapper(), _sp_rm_lixs_mapper(), _sp_dot_lm_lixs_nliner(), _sp_dot_rm_lixs_nliner(), _sp_dot_rm_lmsk_nliner(), _sp2_dot_lm_lixs_nline, _sp2_dot_rm_lixs_nline, _sp2_dot_om_lixs_line, _sp2_dot_om_lrixs_line, _sp2_dot_om_lcixs_line, _sp_qualct, _sp_1cvec_cq, _sp_1rvec_cq, _sp_lcixs_line, _sp_lrixs_line, _sp_idnmat, _sp_1mat, _sp_h1mat.

  • enrich minimize() with IPOPT via cyipopt. Due to licensing constraints, solvers are not pre-packaged. Users have to compile IPOPT with their preferred solvers, ensuring compatibility and customization based on their specific needs. Refer to IPOPT documentation for installation and compilation guidance.

  • add economic attributes income_private_circuit_rsto, disposable_real_incomes and ioq_productivities.

  • add ancillary attribute are_income_private_nodes.

  • add economic attribute gross_incomes_throughput.

  • add ancillary attribute are_externally_sourced.

  • add economic attribute spendings_dfree.

  • enrich minimize() with NLOPT via simplenlopt.

  • add economic attributes trade_margins, trade_margins_exorates, trade_margins_bases, trade_margins_rates, trade_margins_totals_bases, trade_margins_totals_rates.

  • add post-modeling methods/attributes _minimization_coargumenter() and minimization_jacobian.

• New utils’s features

  • xl_ndtrees_op_reader() now has the possibility of rendering branches instead of explicitly structured trees, via datumize=True.

  • the behavior of xl_ndtrees_op_reader() protecting original data’s structure from needless modifications has been generalized upstream at the _op_cells_unmerger()’s level.

  • _items_grouper() can now be provided with key-like getter.

  • xl_ndtrees_op_reader() can now return (implicit) multigraphs.

  • add nx_neg_subgrapher(), to be used to explicit the removable subgraph(s) associated to a list of (un)desired nodes identifiers.

  • _dspaces_kroneckerizer() uses the (sorted) last data type of its input frames as output’s.

  • add _ellipsis_tupler() and _tuples_unnester().

  • add class Serializer whose primary purpose is to save the state of an object or data structure in a way that it can be later reconstructed or restored. This is particularly important in scenarios where data needs to be e.g., stored for later use.

  • add methods _strict_subclass_superer() and _ndarrays_tupler().

• Bug Fixes:

  • In certain corner cases, xl_ndtrees_op_reader() mistakenly merged children and parent nodes. This is why the open-closed precited addition (to xl_ndtrees_op_reader()) was implemented.

  • attrssetter() henceforth considers the possibility of composed children having intersecting namespaces.

  • Ancillary variables henceforth also have their indexers reshaped, hashed and hoisted. Their unconsidered lower-than-3 dimensionality allowed them to escape such treatment.

• New library’s features

  • add TanhCMultipliedTanhCNetPrice, which makes use of TanhCurvedNetPrice and NetPriceTanhCurvedMultiplier through composition.

  • HicksianCES is now fully generalized over \(\sigma \in [0, +\infty)\), be that regarding the calibration of the distribution parameters, optimal demands, output price and output quantity. Consequently, HicksianLeo is destined to be deleted (sooner or later) since it henceforth is a HicksianCES’s particularization at sigma=0.

  • benefiting from the pre-cited generalization, classes HicksiansLeoNestedHicksianCES and HicksiansCESNestedHicksianLeo have been (‘breakingly’) removed and TwoStageHHCES added.

  • add (two HicksianCES’s children classes) ScHicksianCES and SmHicksianCES classes.

  • add (two TwoStageHHCES’s children classes) TwoStageScHHCES and TwoStageSmHHCES classes.

  • enrich ArraysDealer with _take().

  • add class Shareholders.

  • enrich ArraysDealer with _constancy().

  • add classes MarshallianFCES, MarshallianLeo and MarshallianILeo.

  • add PTermsOfTrade.

• New test_core’s features

  • add test_raw_nodes_modes_invariance() to check that mappings’s data format specified as sparse or dense surject.

• Changes

  • economic attribute operating_costs’s alias output_intmd_values has been removed.

  • given that issue np.linalg.inv support #44 has now been solved, economic variables income_private_circuit_order2, income_private_circuit_order3, income_private_circuit_order4, income_private_circuit_order5, income_public_circuit_order2, income_public_circuit_order3, income_public_circuit_order4, income_public_circuit_order5 have no raison d’être anymore and have consequently been removed.

  • following the addition of the precited economic variable ordinal_volumes, class DynamicStock’s method capacity_computer() has ontologically been ‘repaired’ by replacing its argument named additive_output_volumes with ordinal_volumes.

  • library, whose HicksianLUMRS class as been renamed as HicksianFLUMRS for the sake of name representativity.

  • disconnected variables (a.k.a _nvars’s members, undocumented) are turned read-only.

  • update version of anytree (from 2.8.0) to 2.9.0.

  • method _libobj_paths_flattener() now returns the name and the type of attributes that are found recursively, instead of just returning their name.

  • utilization_rates has been turned into idle_rates’s complement in order to fix the ontology.

  • the nature of income_public_circuit_order1 now differs from that of income_private_circuit_order1, as the former, naturally predisposed to circularities, must have it addressed.

  • library’s classes HicksiansCESNestedMarshallianCES and HicksiansCESNestedMarshallianLeo have been removed renamed as TwoStageHMCES and TwoStageHMLeo respectively .

  • ancillary attributes consumptions_types, are_autogenerated, are_intermediary, are_only_final, arent_only_final, are_final and arent_final have been moved upstream from MSystem to DSpace.

  • all user-defined classes have been turned vectorization friendly.

  • scaling is now enabled by default (as it always was for calibration).

  • improve automatic scaling (through _scale_explorer()).

  • economic attributes deflators_lfisher, deflators_fisher, deflators_fisher_1mat, deflators_fisher_1mat_transpose, purchaser_real_prices, consumer_real_prices have been extended to the complex plane, reflecting their very nature.

  • _optim_specifier()’s default scaling-factors’ logarithmic base, β0, changed from 2 to 10 .

  • all library’s methods that were using numpy.var now resort to ArraysDealer’s _constancy() method instead, which increases numerical stability.

---

0.0.11 (July 18, 2023)

• Breaking changes

  • all growth- (i.e. <>_grates) and driving-rates (i.e. <>_drates) variables that were attached to aliases have been removed.

  • economic variables output_taxes and output_taxes_rates have respectively been renamed as contrib_taxes and contrib_taxes_rates in order to fix the ontology. These changes are also reflected at the level of:

    • the example folder, with 1) data related to output taxes on domestic goods being reclassified as sales taxes, 2) relevant tabs’ names being modified accordingly and 3) a worker’s script being aligned consequently at the level of the instantiation of MSystem,

    • library, whose STxRateO2DrivenOTxRate class as been renamed as O2STxRateDrivenCTxRate and relevantly modified,

    • the calculations of the relevant downstream economic variables.

• Changes

  • greatly — not to say considerably — improve the documentation of most of the library’s classes, and especially that of section A Canonical Use Case.

  • update versions of jax (0.3.14 0.4.11 0.4.13 0.3.25 0.4.13), jaxlib (0.4.11 0.3.25 0.4.11), numpy (1.24.4), scipy (1.10.1) and openpyxl (3.1.2).

  • library’s class CurvedNetPrice has been renamed as PowCurvedNetPrice for the sake of “names representativity”. Such change is potentially a breaking one.

  • core’s variable income_private_circuit_hleoninv has been renamed as income_private_circuit_leoninv since it is not hollow anymore.

  • utilization_rates is not an alias of employment_rates anymore, but a fully specified variable that actually consists of its transpose.

  • net_operating_surplus and gross_operating_surplus have respectively been renamed as markets_rents and indirect_operating_costs so as to protect downstream future axes of generalization as well as for the short-term sake of the ontology.

  • net_operating_costs, gross_sales_revenues and all_source_taxes have respectively been renamed as operating_costs, gross_revenues and indirect_taxes_bdown so as to fix the ontology.

  • economic attribute income_circuit_lsto has been removed, as long as the need does not arise. Such change also propagates to its '<>_grates' and '<>_drates' counterparts.

  • economic attribute income_circuit (as well as all those whose name derives from it) has been renamed as income_private_circuit.

  • economic attribute budget_constraints has been renamed as disposable_incomes, of which it becomes an alias.

  • the example folder’s XLSX file (i.e. ut(KLEM-BRICS-202006).xlsx, copied when resorting to copy_example()) makes use of the feature that consists in defining implicit entities in a recursive fashion. As a result, these entities are no longer involved in tab (arbitrarily) named 'inc-circuit' (which maps to income_private_circuit when getting an instance of MSystem).

  • the method _fancy_moveupper(), added in the previous release, and to which _fancy_connecter() relies in order to sort indexes in such a way that those related to a self-definition are placed first (when served to library’s public methods) is now disabled by default, i.e. behind a boolean parameter set to False by default. This has the direct consequence of nullifying the scalar self-dependence exception.

  • injected variables’ growth and driving rates are not computed anymore.

  • _qualvars_specs (undocumented) has been moved up from MSystem to DSpace.

• Display adjustment:

  • module-level function display_klem_like_qois() of klem_mixins.py has been modified so that 1) sums are no longer done naively, i.e. by excluding implicit entities beforehand, 2) markup and margin rates are converted where necessary to match those displayed by KLEM.

• Bug Fixes:

  • consumer_prices hosted erroneous values for entities being only defined at the level of the mappings-keyed sheet. Complying with the YAGNI principle, it henceforth displays zeros instead.

• New conf’s features

  • add REFLECTED_RIGHTS and DETAILED_STAXES, both editable under the same name (prior to module loading) via os.environ.

• New core’s features

  • _libobj_preparator() can henceforth deal with library’s classes having static (and non-differentiable) arguments of instantiation.

  • add private ancillary attributes _1rvec_cq and _1mat_cq (both undocumented) with the intention of easing the precited display adjustment.

  • add ancillary attributes _ilocalizations, are_rights_reflected and frames_as_pleated, with the latter being accompanied by its (still undocumented) _iids_cq and _uiids_cq private underliers.

  • add economic attributes transfers, taxes, direct_taxes, indiv_income_taxes_rates, indiv_income_taxes, gross_incomes, corpo_income_taxes_rates, corpo_income_taxes, net_surplus, indirect_costs, indirect_costs_bdown, operating_costs_bdown, markets_rents_bdown, contrib_taxes_bdown, where the suffix <>_bdown stands for “broken down”.

  • add the public counterpart of income_private_circuit, i.e. income_public_circuit.

  • add the possibility to deal with arbitrarily deep library’s classes compositions (cf. HicksiansCESNestedMarshallianCES, which is a 2-level example of that).

  • add the possibility to rely on recursively defined entities (through the 'mappings' sheet).**

  • _fancy_moveupper() now offers the (defaulted) possibility of removing duplicate indexers.

  • the example folder copied when resorting to copy_example() henceforth contains the original XLSM file iamax was initiated against, namely KLEM BRICS jun20.xlsm, so as to authorize the (joint) opening and updating of the data file. Incidentally, text-concatenated numerical values are processed so that any non-dot decimal separator is modified.

  • considerably reduce compilation times by pruning the objective minimization graph from downstream surcalculations.

• New library’s features

  • add EmplBMagnitude class, whose vocation is to concentrate all methods related to price calculations that depend on utilization rates.

  • add TanhCurvedNetPrice class, a PowCurvedNetPrice’s counterpart.

  • add NetPriceTanhCurvedMultiplier class.

  • enhance CES with vectorization capabilities, as well as its inheriting classes in an open-closed manner, provided that they make explicit use of them.

  • add MarshallianCES, HicksiansCESNestedMarshallianCES, Leontief HicksianLeo, MarshallianLeo, HicksiansCESNestedLeontief HicksiansCESNestedHicksianLeo, HicksiansLeoNestedHicksianCES and HicksiansCESNestedMarshallianLeo classes.

  • add ArraysDealer class, whose vocation is to concentrate all methods related to arrays manipulations, etc.

  • add RatesDealer class, whose vocation is to concentrate all methods related to rates manipulation/conversion, etc.

  • add O2STxRateDrivenSTxRate, a O2STxRateDrivenCTxRate’s cousin.

  • add BalanceOfTrade class.

• New conf’s feature

  • add NULL_SYM, a symbol whose purpose is precisely to be omitted by the parser when assimilating input data. At the moment, its usefulness only appears within the “mappings” tab, a priori.

• New utils’s features

  • _op_cells_typer() can now put specific strings aside.

  • try_() henceforth indifferently deals with functions or methods.

  • add methods attrsgetter() and attrssetter(), both following operator’s philosophy.

  • _items_grouper() now deals with zip objects.

  • const_val_dropper() can now process only one axis instead of two, henceforth defaulted.

  • xl_ndtrees_op_reader() now protects original data’s structure from needless modifications.

  • add method _advanced_indices_merger().

  • xl_ndtables_pd_reader() now has an error message that indicates the problematic sheet by its name.

• New test_core’s features

  • add classes CasesKwargs, BaseTestMSystems (abstract), TestCHNSystem, TestFRASystem, TestDZASystem, parametrized via MetaTestMSystems.

---

0.0.10 (March 24, 2023)

• Breaking changes

  • economic attribute indirect_business_taxes becomes indirect_business_taxes_totals for the sake of “names representativity”. Such change also propagates to its '<>_grates' and '<>_drates' counterparts. indirect_business_taxes still exists as indirect_business_taxes_totals’s disaggregated counterpart.

  • economic attribute unemployments_rates becomes unemployment_rates for the sake of … english. As always, such change also propagates to its '<>_grates' and '<>_drates' counterparts.

• Changes

  • core’s method _dag_resolver leaves GraphML files (instead of GML) on the disk, yEd-friendly potentially.

  • many variable names in core’s documentation have been changed to improve the system’s consistency and comprehensibility.

  • library’s classes Add and NetPricesCurved have respectively been renamed as AddedVolumes and CurvedNetPrice for the sake of both “names representativity” and consistency. Such changes are potentially breaking ones.

  • library’s class ImplicitEntity has one additional (static) method regarding how the input values of its explicit underliers are aggregated.

  • economic attribute unitary_income_gross_products has been removed, as long as the need does not arise. Once again, such change propagates to …

• New core’s features

  • _fancy_connecter() now supports axis-0 vectorization-friendly indexes, which feature has been impulsed by the just-described augmentation of ImplicitEntity.

  • add method _fancy_moveupper() to which _fancy_connecter() relies in order to sort indexes in such a way that those related to a self-definition are placed first (when served to library’s public methods).

  • add ancillary attributes frames_as_denseap, _iidentifiers, _identifiers and _localizations, with the last two being respectively aliased by _identifiers_space and _localizations_space.

  • add economic attributes employment_rates (for the sake of DRY-ness), as well as spendings, spendings_lbw, spendings_lcw, purchaser_real_prices and consumer_real_prices. Such additions also propagate to their ('<>_iparts', if any,) '<>_grates' and '<>_drates' counterparts.

  • meta/economic attributes are_only_final, arent_only_final, operating_costs, gross_operating_costs_bdown and gross_revenues are now fully documented.

• New utils’s features

  • _items_grouper() now deals with generator objects.

---

0.0.9 (March 6, 2023)

• New core’s features

  • add economic attributes excisables, excisables_shares, excisables_totals, (excises,) excises_average_premia, excises_premia, excises_totals_bases, excises_totals_rates, sales_taxables, sales_taxables_shares, sales_taxables_totals, sales_taxes_totals_bases and sales_taxes_totals_rates.

• Changes

  • library’s O2STxRateDrivenCTxRate class has seen its argument change from sales_tax_rate to sales_tax_total_rate.

  • core’s economic attributes excises_exonerations and sales_taxes_exonerations have been removed.

  • re-add economic attribute excises as excises_totals’s disaggregated counterpart.

---

0.0.8 (February 22, 2023)

• New core’s feature

  • the example folder copied when resorting to copy_example() henceforth contains a ‘mixin’ module whose (only implemented) function, namely display_klem_like_qois (undocumented), displays KLEM-like quantities of interest.

• Changes

  • sales_taxes_totals’s domain expands from \(\mathbb{R}^+\) to \(\mathbb{R}\).

---

0.0.7 (February 21, 2023)

• Breaking changes

  • economic attributes io_intensities and io_shares (and their shifted versions, e.g. io_intensities_order2 or io_shares_order2) respectively become ioq_intensities and ioq_shares for the sake of “names representativity”.

• New core’s features

  • add economic attributes iov_intensities and iov_shares as well as their shifted versions.

• New library’s features

  • add O2STxRateDrivenCTxRate class; which addition is reflected within the entities tab of the example folder’s XLSX file, copied when resorting to copy_example().

• Changes

  • contrib_taxes_rates’s domain expands from \(\mathbb{R}^+\) to \(\mathbb{R}\).

---

0.0.6 (February 17, 2023)

• Breaking changes

  • economic attributes sales_taxes and excises respectively become sales_taxes_totals and excises_totals for the sake of “names representativity”. Such change also propagates to their '<>_grates' and '<>_drates' counterparts. sales_taxes still exists as sales_taxes_totals’s disaggregated counterpart.

  • HicksianFCES is renamed as HicksianAFCES for the sake “names representativity” given it was implying accounting-based output prices. HicksianFCES still exists, logically as mobilizing no accounting equation based cost structure.

• New core’s features

  • method _rpremia_computer() is now fully documented.

  • meta/economic attributes are_autogenerated and added_values are now fully documented.

  • add the ability for the solver to explore the complex plane of economic attributes. At the moment, only attribute producer_gross_prices is implemented, reachable via the producer_gross_prices_iparts sheet.

  • add the possibility to discriminately set arrays’ data type instead of doing so globally.

  • add economic/ancillary attributes consumer_prices_iparts, input_volumes_iparts, producer_gross_prices_iparts, producer_net_prices_iparts, purchaser_prices_iparts.

  • add post-optimization attribute grand_saturations.

• Changes

  • all _null_staticmethod (undocumented) are now purely anonymous.

---

0.0.5 (February 5, 2023)

• Deprecations

  • variables suffixed with '<>_nrates', aka “noise variables”, are deprecated. Use their '<>_drates' counterparts instead.

• Breaking changes

  • economic attribute known as consumed_volumes has been renamed as aggregated_input_volumes for the sake of “names representativity”, although still existing as something else, cf. consumed_volumes.

• Changes

  • behavioral settings’ scalar parameters (such as elasticities’) are henceforth considered as well when deviating from the example folder’s XLSX file that is copied when resorting to copy_example().

  • _grand_saturation has been replaced for _grand_saturations (undocumented) with the intention of storing each optimization’s reached saturation. grand_saturation reflects this change by referencing year-dependent values.

  • _libobj_conformity_ensurer() no longer triggers “will work in-place” warnings for calibrating methods, since they are assumed to work so.

• New library’s features

  • reimplement the modeling of crisis-like shocks, such as those on labor productivity. Cf. compute_calibrand_0_star().

  • add HicksianACES, which (inherits from Accountant and HicksianCES and consequently) generates output prices that simply follow the canonical accounting equation.

• New core’s features

  • all constraints are henceforth met relatively, be them involving 0s. Such feature has strong implications in terms of generalization. The possibility of using absolute constraints is preserved though.

  • new static/class-methods: _ndarray_zeros_canceller().

  • generalize the approach regarding how illegal sheets’ names are user-notified.

  • offer the possibility to change entities’ parameters at any step/recursion of the modeling run.

  • library’s classes are henceforth allowed to “side-effect” in a (proactively) XLA-reordering-proof fashion. Such new feature could have justified an increment of the version’s minor number.

• New utils’s features

  • otbprint() now have the possibility to put JSON-unrepresentable objects aside.

---

0.0.4 (January 24, 2023)

• New core’s features

  • smooth objective functions (by avoiding the use of any positivity constraint), which improves performance.

---

0.0.3 (December 22, 2022)

• Bug fixes

  • in previous versions, specifying a year (cf DSpace’s instantiation arguments) that was not belonging to _localizations_space ran the program into a (pandas’s induced) RecursionError. This is henceforth controlled and prevented by explicitly soliciting the user’s intervention, incidentally also unit-tested.

• Changes

  • the example folder’s XLSX file (i.e. ut(KLEM-BRICS-202006).xlsx) is now editable in such a way that pointing to other countries versions is almost 100% automatic. So to speak, the only thing left to do manually is the modification of the behavioral settings. Note that 1) calculation is set to manual mode and 2) pressing F9 is likely to make all targeted data #N/A, in which case you can either specify the source directory of the KLEM file of interest or, even simpler, open it. Do not forget to press F9 again.

---

0.0.2b (December 19, 2022)

• Changes

  • year is now interfaced via/with a setter.

  • ___init_aarrays___() (undocumented) henceforth provides users with scalar DSpace’s property attributes on the basis of their annotation.

  • no dotfile example anymore.

  • all ancillary/meta variables (such as e.g. identity_matrix, consumptions_types, are_autogenerated, …) are made available from library.

  • improve the documentation of some static and class methods, cf. e.g. _loss_computer().

• Breaking changes

  • variables’ suffixes related to noise components have been changed from '<>_noises' to '<>_nrates'. For example, consumer_prices_noises henceforth is consumer_prices_nrates.

  • core’s economic attribute consumer_deflators has been removed.

• New core’s features

  • add ancillary attributes _localizations_space_deltas, _localization_delta, _year, _dlocalization, _dγ, delta_year, dyear.

  • add economic attributes additive_output_volumes, input_values_lbw, input_values_lcw, deflators_llaspeyres, deflators_laspeyres, deflators_lpaasche, deflators_paasche, deflators_sq_lfisher, deflators_lfisher, deflators_fisher.

  • extend MSystem with a new family of variables, whose names end with _drates (standing for ‘driving rates’). This new variables set is analogous to that of _grates (growth rates) or _nrates (noise rates).

  • new static/class-methods: _ndarray_taker(), _ndarrays_taker(), _ndarray_slicing_vals_replacer().

• New library’s features

  • add DynamicStock class.

• New utils’s features

  • method() can now omit arguments via their (integer-specified) position.

0.0.2 (November 3, 2022)

• Changes

  • Complete the set of noise variables.

---

0.0.1 (September 13, 2022)

• Initiation