Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Agent size change based on size of figure #1820

Closed
wants to merge 5 commits into from
Closed

Agent size change based on size of figure #1820

Changes from 4 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
452cca1
caclulate agent size based on Number of Agents
ankitk50 Sep 25, 2023
b8e6c09
cap min size of agents to 5
ankitk50 Sep 25, 2023
5ff171a
update size of plots based on number of users
ankitk50 Sep 26, 2023
fd5e9a5
update algo to scale agents
ankitk50 Sep 30, 2023
0e1b2de
change aspect ratio to fill fraction
ankitk50 Oct 5, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
38 changes: 32 additions & 6 deletions mesa/experimental/jupyter_viz.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -51,6 +51,23 @@ def make_model():
set_current_step(0)
return model

def calculate_space_size():
if model_parameters["N"] > 50:
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Once again, the grid size is independent of the number of agents. Having few agents and having many agents should result in the same marker size and figure size.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, agents less than 50 the figure size of default used by matplotlib. I start to scale the it more num of agents.

Ideal would be scale the image size based on screen size, but this doesn't seem feasible. Any other ideas?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I start to scale the it more num of agents.

Number of agents shouldn't have effect on image size for the grid.

fixed_size = int(model_parameters["N"] / 5)
model_parameters.update({"width": fixed_size, "height": fixed_size})
plt.rcParams["figure.figsize"] = (fixed_size, fixed_size)

def calculate_agent_size():
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The Matplotlib marker size shouldn't depend on the number of agents. It should depend only on the grid size and the computer's screen size.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It should be more abstract, like screen width, because in a Jupyter notebook, the width should be even smaller.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

hmm, let me think how this can be implemented. Currently the size of each agent is parsed via agent_portrayal and hence it's easier(also less compute expensive) to manipulate.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think it would also be important to show all agents in a smaller window size, hence the implementation can work there. I try to come up with some scaling to fit the screen width.

aspect_ratio_high = (
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Definition of aspect ratio is width/height: https://en.wikipedia.org/wiki/Aspect_ratio_(image). fill_fraction is a more descriptive term.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

sure, makes more sense, thanks!

0.5 # maintain this aspect ration b/w agent size and space size
)
aspect_ratio_low = 0.1 #
print()
size_of_agent = aspect_ratio_high * (
model_parameters["width"] * model_parameters["height"]
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(Must be GitHub hiccup. I remembered already commenting about this) the size is length^1, but the width * height is length^2, so it scales quadratically instead. It makes more sense to scale linearly over min(width, height).

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

yes, the idea is to have a constant ratio b/w size of agents (for e.g. area of a circle) and size of the figure (area of a square). Hence the scaling is still linear i.e. the size of agent = aspect_ratio * size of figure.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I just realized that the s param in Matplotlib scatter is length^2: https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.scatter.html. It's markersize**2 without a multiplication by pi. We should document that size refers to Matplotlib size, and it is an area.

)
return size_of_agent

reset_counter = solara.use_reactive(0)
model = solara.use_memo(
make_model, dependencies=[*list(model_parameters.values()), reset_counter.value]
Expand All @@ -61,20 +78,28 @@ def handle_change_model_params(name: str, value: any):

# 3. Set up UI
solara.Markdown(name)

# 4. Calculate space size and scale agents

calculate_space_size()
ag_size = calculate_agent_size()
print(ag_size)

UserInputs(user_params, on_change=handle_change_model_params)
ModelController(model, play_interval, current_step, set_current_step, reset_counter)

with solara.GridFixed(columns=2):
# 4. Space
with solara.Row():
# 5. Space
if space_drawer == "default":
# draw with the default implementation
make_space(model, agent_portrayal)
make_space(model, agent_portrayal, ag_size=ag_size)
elif space_drawer:
# if specified, draw agent space with an alternate renderer
space_drawer(model, agent_portrayal)
# otherwise, do nothing (do not draw space)

# 5. Plots
# 6. Plots
with solara.Row():
for measure in measures:
if callable(measure):
# Is a custom object
Expand Down Expand Up @@ -234,7 +259,7 @@ def change_handler(value, name=name):
raise ValueError(f"{input_type} is not a supported input type")


def make_space(model, agent_portrayal):
def make_space(model, agent_portrayal, ag_size=50):
def portray(g):
x = []
y = []
Expand All @@ -253,7 +278,8 @@ def portray(g):
x.append(i)
y.append(j)
if "size" in data:
s.append(data["size"])
print(ag_size)
s.append(ag_size)
if "color" in data:
c.append(data["color"])
out = {"x": x, "y": y}
Expand Down