add method to make geospace as a solara component

[wang-boyu]

This PR:

• removes GeoJupyterViz and LeafletViz, hence partly addresses Integrate Jupyterviz #199 (comment) by removing code duplicated from Mesa
• adds a method make_geospace_leaflet for the new SolaraViz api in Mesa, as previously discussed in:
  • Implement new SolaraViz API mesa#2263 (comment), and
  • Implement new SolaraViz API mesa#2263 (comment)

As a result, instead of

import mesa_geo.visualization as mgv

mgv.GeoJupyterViz(
    GeoSIR,
    model_params,
    measures= [["infected", "susceptible", "recovered", "dead"], ["safe", "hotspot"]],
    name="GeoSIR",
    agent_portrayal=SIR_draw,
    zoom=12,
    scroll_wheel_zoom=False
)

we can now do

# importing from Mesa now, not Mesa-Geo
from mesa.visualization import SolaraViz

model = GeoSIR()
SolaraViz(
    model,
    name="GeoSIR",
    components=[
        mgv.make_geospace_leaflet(SIR_draw, zoom=12, scroll_wheel_zoom=False),
        mesa.visualization.make_plot_measure(["infected", "susceptible", "recovered", "dead"]),
        mesa.visualization.make_plot_measure(["safe", "hotspot"]),
    ]
)

A few other things to consider (could be separate PRs, or to be addressed in this one):

• fix test cases
• update intro tutorial (in separate PR)
• have a default geospace component, similar to add default SolaraViz mesa#2280 for Mesa (in separate PR)
• wait for a new name for SolaraViz (doesn't really matter; only need to update tutorial)
  • discussed in Implement new SolaraViz API mesa#2263 (comment), and
  • Implement new SolaraViz API mesa#2263 (comment))
• wait for SolaraViz to be released
• keep GeoJupyterViz with deprecation warnings

@tpike3 would you like to test this out and see if it's working as expected?

[Corvince]

Great work! Looks pretty good from a brief look, but as said earlier can't test it atm

[EwoutH]

I will make time for full review Monday, but the fact that this replaces 428 lines of code with only 84 is always a good sign!

[EwoutH]

Looks like an awesome, almost drop-in replacement!

Could we add aliases the old names at some place to keep those working (maybe with a deprecation warning to use the new name)?

Is there anything in particular you would like me to review or test in detail?

[wang-boyu]

Thanks for your comments!

Could we add aliases the old names at some place to keep those working (maybe with a deprecation warning to use the new name)?

Certainly! I've added another item into the todo list. Will bring back GeoJupyterViz together with some deprecation warnings.

Is there anything in particular you would like me to review or test in detail?

Mostly about how make_geospace_leaflet is implemented. Would also be great if you could verify if it's working for the intro tutorial (btw mesa needs to be installed from main branch), to make sure that it's not just working on my laptop : )

Thanks agin for looking into this.

[wang-boyu]

Rebased to main branch

[codecov]

Codecov Report

Attention: Patch coverage is 77.35849% with 24 lines in your changes missing coverage. Please review.

Project coverage is 73.48%. Comparing base (ce501b6) to head (70fd9f5).
Report is 68 commits behind head on main.

Files with missing lines	Patch %	Lines
...a_geo/visualization/components/geospace_leaflet.py	76.04%	17 Missing and 6 partials ⚠️
mesa_geo/geospace.py	83.33%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #246      +/-   ##
==========================================
- Coverage   78.21%   73.48%   -4.73%     
==========================================
  Files          10       10              
  Lines         693      856     +163     
  Branches      151      182      +31     
==========================================
+ Hits          542      629      +87     
- Misses        127      194      +67     
- Partials       24       33       +9     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[EwoutH]

Nice! Test pass, that's a good thing.

Let's us know if you need any input!

[wang-boyu]

Brought back GeoJupyterViz and rebased to main branch. I'll leave the update to intro tutorial and deprecation warnings to future PRs.

Ready for merging now.

[EwoutH]

Any specific input/testing/review you need?

[wang-boyu]

Thanks @EwoutH. I experienced some slowness while trying the new solara api (also reported here projectmesa/mesa#2289 (comment)), for both this intro tutorial as well as Mesa's viz tutorial. Would be great if you could confirm this (or it could be just my machine).

But I think this is outside the scope of this PR, since GeoJupyterViz isn't really touched here.

[rht]

@tpike3 this needs your review.

[rht]

This PR has been stuck for a while, and given that @tpike3 hasn't responded, I am going to unblock by approving.

[rht]

@wang-boyu test GIS examples are still failing.

[EwoutH]

I resolved almost all errors and warnings in the GIS examples (only one left is projectmesa/mesa-examples#209).

[EwoutH]

I would love to do a Mesa-geo pre-release this week, since now all the gis examples work with the latest Mesa 3.0 pre-release. It would be really nice to have this in!

[wang-boyu]

Thanks @rht for approving and @EwoutH for fixing the tests and intro tutorial. I would suggest to go ahead and merge this, since the existing GeoJupyterViz is not touched.

[EwoutH]

Let’s give @tpike3 a few more hours, most of the time he’s online between 10 and 12 CET.

[tpike3]

Sorry for the delay --- very cool wrapping this in the SolaraViz!

Only concern is the build is failing

I lost the thread of the conversation is the intro tutorial going to be updated in a different pull request?

[EwoutH]

Thanks for reviewing, did you try it yourself?

Only concern is the build is failing

That's not related to this PR, and we already fixed that in #251

[EwoutH]

Okay so I tested this by building a viz app for geo_sir

import sys
import os

# Add the directory containing mesa-geo to the Python path
sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), '..', '..', '..')))

from mesa.visualization import SolaraViz, make_plot_measure
from agents import PersonAgent
from model import GeoSir

# Import the make_geospace_leaflet function directly
from mesa_geo.visualization import make_geospace_leaflet

def SIR_draw(agent):
    """
    Portrayal Method for canvas
    """
    portrayal = {}
    if isinstance(agent, PersonAgent):
        portrayal["marker_type"] = "Circle"
        portrayal["radius"] = 2
    if agent.atype in ["hotspot", "infected"]:
        portrayal["color"] = "Red"
    elif agent.atype in ["safe", "susceptible"]:
        portrayal["color"] = "Green"
    elif agent.atype in ["recovered"]:
        portrayal["color"] = "Blue"
    elif agent.atype in ["dead"]:
        portrayal["color"] = "Black"
    return portrayal

model_params = {
    "pop_size": {
        "type": "SliderInt",
        "value": 30,
        "min": 10,
        "max": 100,
        "step": 10,
        "label": "Population size",
    },
    "init_infected": {
        "type": "SliderFloat",
        "value": 0.2,
        "min": 0.0,
        "max": 1.0,
        "step": 0.05,
        "label": "Fraction initial infection",
    },
    "exposure_distance": {
        "type": "SliderInt",
        "value": 500,
        "min": 100,
        "max": 1000,
        "step": 100,
        "label": "Exposure distance",
    },
}

model1 = GeoSir()

page = SolaraViz(
    model1,
    model_params=model_params,
    name="Basic agent-based SIR model",
    components=[
        make_geospace_leaflet(SIR_draw, zoom=12, scroll_wheel_zoom=False),
        make_plot_measure(["infected", "susceptible", "recovered", "dead"]),
        make_plot_measure(["safe", "hotspot"]),
    ]
)

Running with:

solara run "docs/examples/projectmesa mesa-examples main gis-geo_sir/app.py"

(yeah I named my folder terribly)

It runs, which is great:

However, when I press reset I and then run, it directly crashes on the first step:

Traceback (most recent call last):
  File "C:\Users\Ewout\.virtualenvs\Py312\Lib\site-packages\reacton\core.py", line 1707, in _render
    root_element = el.component.f(*el.args, **el.kwargs)
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\Ewout\.virtualenvs\Py312\Lib\site-packages\mesa\visualization\solara_viz.py", line 243, in ModelController
    solara.lab.use_task(step, dependencies=[playing.value], prefer_threaded=False)
  File "C:\Users\Ewout\.virtualenvs\Py312\Lib\site-packages\solara\tasks.py", line 846, in use_task
    return wrapper(f)
           ^^^^^^^^^^
  File "C:\Users\Ewout\.virtualenvs\Py312\Lib\site-packages\solara\tasks.py", line 840, in wrapper
    raise task_instance.exception
  File "C:\Users\Ewout\.virtualenvs\Py312\Lib\site-packages\solara\tasks.py", line 259, in runner
    self._last_value = value = await self.function(*args, **kwargs)
                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\Ewout\.virtualenvs\Py312\Lib\site-packages\mesa\visualization\solara_viz.py", line 241, in step
    do_step()
  File "C:\Users\Ewout\.virtualenvs\Py312\Lib\site-packages\mesa\visualization\solara_viz.py", line 247, in do_step
    model.value.step()
  File "C:\Users\Ewout\.virtualenvs\Py312\Lib\site-packages\mesa\visualization\solara_viz.py", line 152, in step
    original_step()
  File "C:\Users\Ewout\.virtualenvs\Py312\Lib\site-packages\mesa\model.py", line 73, in _wrapped_step
    self._user_step(*args, **kwargs)
  File "C:\Users\Ewout\Documents\GitHub\mesa-geo\docs\examples\projectmesa mesa-examples main gis-geo_sir\model.py", line 108, in step
    self.agents_by_type[PersonAgent].shuffle_do("step")
  File "C:\Users\Ewout\.virtualenvs\Py312\Lib\site-packages\mesa\agent.py", line 306, in shuffle_do
    getattr(agent, method)(*args, **kwargs)
  File "C:\Users\Ewout\Documents\GitHub\mesa-geo\docs\examples\projectmesa mesa-examples main gis-geo_sir\agents.py", line 55, in step
    for neighbor in neighbors:
                    ^^^^^^^^^
  File "C:\Users\Ewout\Documents\GitHub\mesa-geo\mesa_geo\geospace.py", line 205, in get_neighbors_within_distance
    yield from self._agent_layer.get_neighbors_within_distance(
  File "C:\Users\Ewout\Documents\GitHub\mesa-geo\mesa_geo\geospace.py", line 397, in get_neighbors_within_distance
    possible_neighbors = self._get_rtree_intersections(geometry)
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\Ewout\Documents\GitHub\mesa-geo\mesa_geo\geospace.py", line 287, in _get_rtree_intersections
    self._id_to_agent[i] for i in self._idx.intersection(geometry.bounds)
    ~~~~~~~~~~~~~~~~~^^^
KeyError: 2287846026144

[tpike3]

Okay so I tested this by building a viz app for geo_sir

...

self._id_to_agent[i] for i in self._idx.intersection(geometry.bounds)
~~~~~~~~~~~~~~~~~^^^

KeyError: 2287846026144

OK thats because mesa-geo assigns a very large number to plygon agents, which should now be replaced just the auto created id. I am looking for the line(s)

[tpike3]

@EwoutH I think the problem is here --mesa-geo agents property

This is using agents as a property and may be causing a collision with agents in Mesa now.

Mesa-geo creates a rtree index for spatial indexing, so I think the agent_ids are coming into conflict.

[wang-boyu]

Is it because of this:

mesa-geo/mesa_geo/geospace.py

Line 322 in 4e22c5c

index_data = ((id(agent), agent.geometry.bounds, None) for agent in agents)

We used the agent object id as key, and the agent object may be deleted and recreated during model reset, resulting in a different object id. Changing this to agent.unique_id may help? Not very sure about it.

[wang-boyu]

Replaced id(agent) with agent.unique_id in geospace. Would you like to try again @EwoutH ? Thanks for testing this btw.

[EwoutH]

Awesome, resetting the model now works!

[EwoutH]

I'm good with merging. It works, run well, and it's great that it now uses SolaraViz directly, it's nice to have a single interface.

It might even help with projectmesa/mesa#2330 ;)

Thanks for the effort!

[wang-boyu]

Thank you @Corvince @EwoutH @rht @tpike3. I'll merge this now.

[EwoutH]

Congrats!

Are there any old viz components that should be removed or deprecated?

[wang-boyu]

No I would prefer to keep the current GeoJupyterViz for now until the new API is ready and we figure out what to do with GeoJupyterViz. Would you like to make a prerelease now? I can start updating the gis examples to use the solara viz.