fix: update alliance_formation example to use AgentSet.to_list() by souro26 · Pull Request #3235 · mesa/mesa

souro26 · 2026-01-31T13:02:06Z

Summary

Updates the alliance_formation example model to use the new AgentSet.to_list() method instead of deprecated direct indexing.

Bug / Issue

The alliance_formation model uses deprecated AgentSet.__getitem__ (e.g., agents[0]), which was deprecated in #3208 and will be removed in Mesa 4.0.
This triggers PendingDeprecationWarning when running the example, and since this is a reference example that users often copy, it should demonstrate current best practices.

Implementation

Replaced direct indexing:

agent_0 = agents[0]
agent_1 = agents[1]

With the recommended pattern from the migration guide (#3218):

agent_list = agents.to_list()
agent_0 = agent_list[0]
agent_1 = agent_list[1]

Testing

Ran the alliance_formation example test:

pytest tests/examples/test_examples.py::test_alliance_formation_model -v

Additional Notes

Follow-up to #3208 and #3218.
This change eliminates deprecation warnings in the alliance_formation example model.

github-actions · 2026-01-31T13:16:32Z

Performance benchmarks:

Model	Size	Init time [95% CI]	Run time [95% CI]
BoltzmannWealth	small	🔵 -1.4% [-1.6%, -1.1%]	🔵 -1.2% [-1.4%, -1.1%]
BoltzmannWealth	large	🔵 +1.5% [+0.9%, +2.1%]	🔵 -0.3% [-0.8%, +0.2%]
Schelling	small	🔵 +0.6% [+0.4%, +0.8%]	🔵 +0.7% [+0.7%, +0.8%]
Schelling	large	🔵 +0.7% [+0.2%, +1.2%]	🔴 +3.5% [+3.1%, +3.9%]
WolfSheep	small	🔵 -1.4% [-1.6%, -1.3%]	🔵 -0.5% [-0.7%, -0.3%]
WolfSheep	large	🔵 -0.7% [-1.5%, +0.1%]	🔵 -0.9% [-1.4%, -0.5%]
BoidFlockers	small	🔵 -0.1% [-0.3%, +0.2%]	🔵 +1.7% [+1.5%, +1.8%]
BoidFlockers	large	🔵 +0.1% [-0.3%, +0.5%]	🔵 +1.6% [+1.2%, +2.0%]

quaquel · 2026-01-31T16:01:22Z

mesa/examples/advanced/alliance_formation/model.py

-        agent_1 = agents[1]
+        agent_list = agents.to_list()
+        agent_0 = agent_list[0]
+        agent_1 = agent_list[1]


I suggest fixing this find_combinations instead.

souro26 · 2026-02-01T05:55:47Z

Thanks for the quick review @quaquel.

I double-checked find_combinations, and it doesn’t use deprecated indexing. The deprecated access was in calculate_shapley_value, which is now resolved by converting the AgentSet once via to_list(). I avoided changing find_combinations since converting the AgentSet there would alter its iteration semantics. Please let me know if you’d prefer a different approach.

quaquel · 2026-02-01T06:49:38Z

That is not entirely true. Look at the typing of find_combinations. group is an AgentSet. This should be changed to a list, or, looking at the code, an iterable. Next, evaluate_combination explicitly puts the group (which is typed as a tuple) into an agent set. This line should be removed. Also, the typing of evaluation_func should be updated across all of meta_agents.py.

- Update find_combinations to operate on iterables/tuples - Remove intermediate AgentSet creation in evaluate_combination - Update evaluation_func typing across meta_agents - Adjust alliance_formation example to convert tuple to AgentSet internally

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

souro26 · 2026-02-01T18:15:42Z

Thanks for the clarification — I’ve updated the implementation accordingly.

Changes made:

Updated evaluation_func typing across meta_agents.py to operate on tuple[Agent, ...]
Removed internal AgentSet construction from evaluate_combination
Updated find_combinations to accept AgentSet | Iterable, converting once for iteration
Updated the alliance_formation example to explicitly convert the tuple to AgentSet only where AgentSet-specific methods are required

I’ve also noticed a couple of now-redundant lines in calculate_shapley_value (duplicate agent extraction / AgentSet usage) and can clean those up if you’d like me to push a small follow-up commit.

All relevant tests are passing:

tests/experimental/test_meta_agents.py
tests/examples/test_examples.py::test_alliance_formation_model

I'm happy to adjust further based on your preference.

quaquel · 2026-02-01T18:28:12Z

mesa/experimental/meta_agents/meta_agent.py

 def find_combinations(
    model,
-    group: AgentSet,
+    group: AgentSet | list | Iterable,


Suggested change

group: AgentSet | list | Iterable,

group: Iterable,

iterable is enough here. AgentSet and list are already iterable so no need to add them here.

quaquel · 2026-02-01T18:28:42Z

mesa/experimental/meta_agents/meta_agent.py

    Args:
        model: The model instance.
-        group (AgentSet): The set of agents to find combinations in.
+        group (AgentSet | list | Iterable): The set of agents to find combinations in.


please remove all typing from docstring. Should not be there if typing is used.

quaquel · 2026-02-01T18:29:00Z

mesa/experimental/meta_agents/meta_agent.py

+    if isinstance(group, AgentSet):
+        agent_list = group.to_list()
+    else:
+        agent_list = list(group) if not isinstance(group, list) else group
+


not needed at all.

quaquel · 2026-02-01T18:29:43Z

mesa/examples/advanced/alliance_formation/model.py

+            group_set = AgentSet(group, random=self.random)
+            agent_ids = sorted(group_set.get("unique_id"))  # by default is bilateral


you don't need an agentset here. You can just use operator.itemgetter instead as key in sorted.

quaquel · 2026-02-01T18:30:08Z

I’ve also noticed a couple of now-redundant lines in calculate_shapley_value (duplicate agent extraction / AgentSet usage) and can clean those up if you’d like me to push a small follow-up commit.

Please do.

- Remove unnecessary AgentSet conversions - Accept Iterable directly in find_combinations - Simplify calculate_shapley_value agent handling - Clean up redundant logic per review feedback

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

quaquel · 2026-02-01T20:05:55Z

mesa/examples/advanced/alliance_formation/model.py

-        agent_0 = agents[0]
-        agent_1 = agents[1]
+        agent_0, agent_1 = agents
+        agent_set = AgentSet(agents, random=self.random)


this should not be needed at all.

quaquel · 2026-02-01T20:06:29Z

mesa/examples/advanced/alliance_formation/model.py

+        positions = agent_set.get("position")
        new_position = 1 - (max(positions) - min(positions))
-        potential_utility = agents.agg("power", sum) * 1.2 * new_position
+        potential_utility = agent_set.agg("power", sum) * 1.2 * new_position


You only have 2 agents, so why use an aggregation here? Just write it up explicitly for 2 agents.

souro26 · 2026-02-01T20:11:14Z

Thanks for the feedback, it makes sense I'll simplify it to use the attributes directly since it's just 2 agents.

…/github.com/souro26/mesa into fix-alliance-formation-agentset-deprecation :wq:wq# the commit.

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

fix: update alliance_formation to use AgentSet.to_list()

017af1e

quaquel reviewed Jan 31, 2026

View reviewed changes

quaquel added the bug Release notes label label Feb 1, 2026

souro26 and others added 2 commits February 1, 2026 23:11

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

bc8bcaf

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

quaquel reviewed Feb 1, 2026

View reviewed changes

souro26 added 2 commits February 2, 2026 01:20

refactor: simplify meta_agents iteration and example usage

31257e0

- Remove unnecessary AgentSet conversions - Accept Iterable directly in find_combinations - Simplify calculate_shapley_value agent handling - Clean up redundant logic per review feedback

resolve merge conflict

95d97c7

souro26 force-pushed the fix-alliance-formation-agentset-deprecation branch from 9de6e4c to 95d97c7 Compare February 1, 2026 20:00

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

a0ec713

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

quaquel reviewed Feb 1, 2026

View reviewed changes

souro26 and others added 3 commits February 2, 2026 01:50

Simplify calculate_shapley_value for explicit two-agent case

32f7d13

Merge branch 'fix-alliance-formation-agentset-deprecation' of https:/…

ee6880a

…/github.com/souro26/mesa into fix-alliance-formation-agentset-deprecation :wq:wq# the commit.

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

09287ce

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

quaquel approved these changes Feb 1, 2026

View reviewed changes

quaquel merged commit 2f2656b into mesa:main Feb 1, 2026
11 of 12 checks passed

This was referenced Feb 4, 2026

Deprecated AgentSet indexing in meta_agents.extract_class function #3240

Closed

fix: update meta_agents extract_class to use to_list() #3241

Merged

		group_set = AgentSet(group, random=self.random)
		agent_ids = sorted(group_set.get("unique_id")) # by default is bilateral

Uh oh!

Conversation

souro26 commented Jan 31, 2026

Summary

Bug / Issue

Implementation

Testing

Additional Notes

Uh oh!

github-actions bot commented Jan 31, 2026

Uh oh!

quaquel Jan 31, 2026

Choose a reason for hiding this comment

Uh oh!

souro26 commented Feb 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

quaquel commented Feb 1, 2026

Uh oh!

souro26 commented Feb 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

quaquel Feb 1, 2026

Choose a reason for hiding this comment

Uh oh!

quaquel Feb 1, 2026

Choose a reason for hiding this comment

Uh oh!

quaquel Feb 1, 2026

Choose a reason for hiding this comment

Uh oh!

quaquel Feb 1, 2026

Choose a reason for hiding this comment

Uh oh!

quaquel commented Feb 1, 2026

Uh oh!

quaquel Feb 1, 2026

Choose a reason for hiding this comment

Uh oh!

quaquel Feb 1, 2026

Choose a reason for hiding this comment

Uh oh!

souro26 commented Feb 1, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

souro26 commented Feb 1, 2026 •

edited

Loading

souro26 commented Feb 1, 2026 •

edited

Loading