Added a main() function to bank_reserves `batch_run.py` and sugarscape_g1mt `run.py` scripts to facilitate testing and script execution.

The `main()` function encapsulates the primary script logic, allowing for easier modular testing and execution. By defining script operations within `main()`, we can directly invoke this function in testing environments without relying on command-line execution. This practice enhances code readability, maintainability, and testability, providing a clear entry point for the script's functionality.

Stop testing stable mesa builds, which can be enabled again when Mesa 3.0 is released. This allows us to start updating Mesa example to break with 2.x conventions, towards development of Mesa 3.0.

Mesa 2.x examples can be found on the mesa-2.x branch.

updates:
- [github.com/astral-sh/ruff-pre-commit: v0.5.0 → v0.5.6](astral-sh/ruff-pre-commit@v0.5.0...v0.5.6)
- [github.com/asottile/pyupgrade: v3.16.0 → v3.17.0](asottile/pyupgrade@v3.16.0...v3.17.0)

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

This allows seeing which tests are collected, passed, etc.

See https://docs.pytest.org/en/7.1.x/how-to/output.html#verbosity

See https://docs.pytest.org/en/7.1.x/how-to/output.html#producing-a-detailed-summary-report

This updates all the examples to use the renamed and stabilized SolaraViz.

Resolves 4 warnings by using the AgentSet select(), shuffle() and do() functionality.

Also removes the now unused scheduler.

Don't pass the position parameter, but just use place_agent. Resolves the duplicate position warning.

Don't pass the position parameter, but just use place_agent. Resolves the duplicate position warning.

This resolves all warnings outputted by this model.

For the model step, the behavior of the old RandomActivationByType scheduler when using step(shuffle_types=True, shuffle_agents=True) is replicated. Conceptually, it can be argued that this should be modelled differently.

The verbose prints are also removed.

Don't pass the position parameter, but just use place_agent. Resolves the duplicate position warning.

Uses the  -Werror option of pytest to treat warnings as errors.

See https://docs.python.org/3/using/cmdline.html#cmdoption-W

Use model.get_agents_of_type instead of model.agents.select() to select all agents of a certain type. It should be faster, since it's a direct dictionary call.

Replace all usages of model.schedule.agents with the use of the model.agents AgentSet in all examples.

Replace the old RandomActivation scheduler with AgentSet functionality. self.agents.shuffle.do("step") is equivalent and self.schedule.step() with an RandomActivation scheduler, and model behavior should not change.

The SimultaneousActivation scheduler used a "step" and "advanced" function in the Agent. These have been renamed for each model to be more descriptive (like "determine_state" and "assume_state"), and then the model step was replace by something like:
        self.agents.do("determine_state")
        self.agents.do("assume_state")

Docstring was also updated accordingly.

Schedulers don't exist anymore, so we don't need to add agents anymore.

Agents are automatically added to model.agents on Agent initialisation.

These models use complex schedulers which warrant an separate PR.

It doesn't have a step function and there is only one, so it shouldn't be an Agent, just an Python object.

…lity (projectmesa#170)

This commit reverts PR projectmesa#161 projectmesa#161

That PR assumed that time advancement would be done automatically, like proposed in projectmesa/mesa#2223

We encountered some underlying issues with time, which we couldn't resolve in time.

* adding first ant

* Ant path building

* Implementing stigmergy and ACO

* TSP benchmark

* refactor 01

* app fix

* simple fix, isort, clean

* lint

* readme

* adding default args to pass unit tests

* addressing PR comments

* comment fix 2

* PR comments iter

---------

Co-authored-by: Zak Jost <[email protected]>

Use relative imports in all models. This way models can be ran from different locations, which will help with testing them.

Co-Authored-By: rht <[email protected]>

- Rename workflow
- Run on each push (which allows developing on test branches in this repo)
- Run only on changes to the example models, the test script or the CI config
- Allow multiple commits after each other (to allow continuous trial and error

When inheriting from a Mesa Model you always need to initialize it by calling super().__init__(). This commit fixes that in two gis models.

Fix the data path import errors in the geo_schelling, geo_schelling_points and geo_sir models.

This commit addresses several of the remaining issues across the GIS examples:

- Corrects data file paths in the population, rainfall, and urban growth models to use relative paths from the script directory.
- Fixes the urban growth model by passing the model object to City and UrbanCell classes, as required by mesa-geo 0.8.0.
- Updates the population model to pass the model object when creating agents, ensuring they can access Model variables and properties.
- Corrects the use of `/vsigzip/` paths for GDAL compatibility.

The `/vsigzip/` situation is super weird. It isn't part of the file path, but it's a "marker" a GDAL convention. See the PR message, mesa-geo#235 and https://gdal.org/user/virtual_file_systems.html#vsigzip-gzipped-file for more details.

Co-authored-by: Ewout ter Hoeven <[email protected]>

…sa#181)

the datacollector allways includes grasspatches even if grass is False. This is breaking (the proposed changes to agent storage in Model.

The underlying problem is that Model.get_agents_of_type now raises a KeyError if type does not exist. I think this is desirable behavior becuase errrors should not be passed over in silence. But this requires fixing the wolf sheep example as done here.

…ectmesa#162)

Remove the remove temporary _advance_time() method from schelling_experimental. This isn't needed anymore, because in Mesa 3.0 time will be automatically increased.

A special implementation of Conway's Game of Life, using only the PropertyLayer, without needing a Grid or even any Agents.

Add minimal Solara visualisation, to control model parameters.

- Add two metrics to the model
- Collect those with the datacollector
- Plot them as measures in SolaraViz

The fast GoL model needs it.

Otherwise we can't use this experimental feature.

`steps` is now increased automatically, it's not needed anymore!

…onality

This reinstates PR projectmesa#161 after the previous revert.

Confirm that model.steps is correctly increased to  10, and not some other value

Use pass the new rio_opener argument with gzip.open in RasterLayer.from_file(). This prevents having to do the weird /vsigzip/ stuff.

Note that urban_growth uses the rasterio opener directly, while the other two examples pass it via the mesa_geo.raster_layers.RasterLayer.open_file() method.

Fixes the population and rainfall models by passing model to RasterLayer.from_file().

> The RasterLayer creates Cells, Cells inherit from Agents, Agents need model input, but RasterLayer doesn't have access to model.

updates:
- [github.com/astral-sh/ruff-pre-commit: v0.5.6 → v0.6.3](astral-sh/ruff-pre-commit@v0.5.6...v0.6.3)

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

[pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

update

…rojectmesa#190)

The Model method `get_agents_of_type()` is replaced by the `agents_by_type` property, which directly returns the dict.

Instead of using:
```Python
model.get_agents_of_type(Wolf)
```
You should now use:
```Python
model.agents_by_type[Wolf]
```

Note that examples on the main branch work on the latest Mesa-Geo development branch, and examples Mesa-Geo 0.8.x are available on the mesa-2.x branch.

Add seeding RL examples folder (projectmesa#178)

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Also use the new, standardized order of count-do-collect.

All examples can be updated to no longer pass a unique id, nor use model.next_id.

This only fixes the examples, not the gis-examples or rl examples.

Co-authored-by: Ewout ter Hoeven <[email protected]>

Replace shuffle().do() in 18 models with the performance optimized shuffle_do() method.

…mesa#202)

This PR completes the migration from schedulers to AgentSet functionality across the mesa-examples repository for all regular (non-`gis`/-`rl`) examples. Key changes include:

- Replaced `RandomActivation`, `SimultaneousActivation`, and `RandomActivationByType` schedulers with appropriate AgentSet methods
- Updated `Model.step()` implementations to use AgentSet activation
- Removed references to `schedule.steps`, `schedule.agents`, and `schedule.agents_by_type`
- Updated agent addition/removal logic to work with AgentSets
- Adjusted data collection and visualization code to use `Model.steps` and `Model.agents`

For more details on migrating from schedulers to AgentSets, see the migration guide: https://mesa.readthedocs.io/en/latest/migration_guide.html#time-and-schedulers