Refactor agent allocation #284
Conversation
|
Hey. The PR is ready for review? Cause right now it's only a "draft PR" |
| circles = [] | ||
|
|
||
| while len(agents_for_type) > 0: | ||
| circle_size = circles_size_distribution.rvs() |
There was a problem hiding this comment.
Using numpy choice is much better
| adults = [agent for agent in agents_for_type if agent.age > 18] | ||
| non_adults = [agent for agent in agents_for_type if agent.age <= 18] | ||
|
|
||
| adult_num = min(round(adult_type_distribution.rvs()), len(adults)) |
There was a problem hiding this comment.
With the current implementation, the adult distribution is only available from data_holder a randint. Should I change it to a tuple of arrays, to be used with numpy choice?
| # if there is place left in the circle, fill it with agents: | ||
| while circle.agent_count < size: | ||
| agent = agents_for_type.pop() | ||
| assert agent not in circle.agents |
There was a problem hiding this comment.
Social circle already checks this. So this is not needed.
FOR ALL the asserts. instead, why not use hust add_many instead of add_agent each time?
There was a problem hiding this comment.
Yes, the assert was used to debug a specific issue. I haven't found a simpler way to select adults multiple adults, and remove them at once from adults and agents_for_type, so the loop would still be required
There was a problem hiding this comment.
Why not? You know how many are needed ( size - circle.agent_count) , you can take a cut out of agent_of_type use add_many.
Also pop sounds like a bad use, because you modify the outer list, a result which seems undesirable.
There was a problem hiding this comment.
The outer list should be modified because those agents were allocated to a circle of the current connection type, and should not be selected for similar circles again. I guess this can be accomplished more efficiently through list slicing, instead of this original implementation
There was a problem hiding this comment.
@denzo9 Check if it does so, what in actuall might happen with just slicing is that it will change your pointer
There was a problem hiding this comment.
I pushed a solution that seems to work, using slicing and the del keyword
| # if there is place left in the circle, fill it with agents: | ||
| while circle.agent_count < size: | ||
| agent = agents_for_type.pop() | ||
| assert agent not in circle.agents |
There was a problem hiding this comment.
Why not? You know how many are needed ( size - circle.agent_count) , you can take a cut out of agent_of_type use add_many.
Also pop sounds like a bad use, because you modify the outer list, a result which seems undesirable.
No description provided.