Revisions to Market parameter structure

[mnwhite]

This PR is meant to address issues #766 and #989 ; the main change is a regularization of get_economy_data. Rather than each AgentType subclass having their own method that says how to get information from the associated Market instance, there is now a unified AgentType.get_market_params method. AgentType subclasses that (usually) use an associate Market should specify a market_vars class attribute naming the parameters that should be pulled down from the Market. The user also doesn't need to write a trivial loop over AgentType instances in the Market; instead, they can just call Market.give_agent_params(), which also runs agent.construct() by default because the agents usually need information from the economy to finish building themselves.

The example notebook has been lightly edited to work with the changes, and most of the tests pass. However, two market-level solution results changed slightly. I am issuing the PR now so I can look carefully at what I changed and see if I made a mistake somewhere.

Incidentally, this PR also makes a tiny extension to the special get_it_from constructor. If the thing that it is told to get_it_from is just a single float, int, bool, complex, or string, then it is interpreted as literally just taking that value itself. In most cases, get_it_from is used to "unpack" attributes from an object or entries from a dictionary, but now it also means "it's this number by a different name".

Tomorrow morning I'll add some flexibility to allow "parameter-like" arguments to calc_dynamics, which is in a similar vein to these changes.

[mnwhite]

The two changed test targets are because the seeds of distributions have changed. With the revisions here, the "order of operations" for building distributions has changed, so the agents' IncShkDstn ends up with a different seed(s) than before.

[mnwhite]

This is ready for review, but the CHANGELOG-only PR should be merged into main first, then pulled into this branch (and updated).

[mnwhite]

CHANGELOG has been updated and this is now properly ready for review.

[Copilot]

Pull request overview

This PR addresses issues #766 and #989 by regularizing how AgentType subclasses obtain parameters from their associated Market instances. The changes replace custom get_economy_data() methods with a unified get_market_params() approach that uses a market_vars class attribute to specify which parameters should be fetched from the Market.

Changes:

• Introduced AgentType.get_market_params() method and market_vars class attribute to replace custom get_economy_data() methods across agent types
• Added Market.give_agent_params() convenience method to distribute market-level parameters to all agents and run their construct() methods
• Extended the get_it_from constructor to handle scalar values directly, not just objects and dictionaries
• Modified Market.update_dynamics() to allow calc_dynamics to accept additional arguments beyond those in track_vars by looking for matching Market attributes
• Refactored income shock distribution constructors into standalone functions in IncomeProcesses.py
• Updated test assertion values that changed slightly due to different ordering of distribution construction

Reviewed changes

Copilot reviewed 11 out of 12 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
HARK/core.py	Added get_market_params() method to AgentType and give_agent_params() method to Market; enhanced update_dynamics() to fetch non-tracked arguments from Market attributes
HARK/utilities.py	Extended get_it_from constructor to return scalar values directly when the parent is a primitive type
HARK/ConsumptionSaving/ConsAggShockModel.py	Major refactoring: removed get_economy_data() and add_AggShkDstn() methods; added market_vars attributes; moved income shock combination logic to standalone constructor functions; renamed MrkvNow_init to MrkvInit for consistency
HARK/ConsumptionSaving/ConsIndShockModel.py	Updated docstring parameter names from aNrmInitMean/Std to kLogInitMean/Std and pLvlInitMean/Std to pLogInitMean/Std
HARK/Calibration/Income/IncomeProcesses.py	Removed duplicate assertion; added combine_ind_and_agg_income_shocks() and combine_markov_ind_and_agg_income_shocks() utility functions
tests/ConsumptionSaving/test_ConsAggShockModel.py	Updated tests to use give_agent_params() instead of manual loops calling get_economy_data(); updated expected test values that changed slightly due to constructor ordering changes
tests/ConsumptionSaving/test_SmallOpenEconomy.py	Replaced manual parameter assignment and get_economy_data() loop with consolidated dictionary approach and give_agent_params() call
examples/Journeys/Journey-PhD.ipynb	Updated to use new give_agent_params() method; incidentally changed Python version to 3.12.9
examples/Gentle-Intro/Market-Intro.ipynb	Added documentation for max_loops parameter; clarified that calc_dynamics can accept additional Market attribute arguments; enhanced explanation of intractability in Krusell & Smith model; incidentally changed Python version to 3.12.9
examples/Gentle-Intro/Constructors-Intro.ipynb	Added new section explaining the special case for AgentType subclasses in Market contexts that don't run construct() at instantiation
docs/CHANGELOG.md	Documented breaking changes and new features

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.