Extending Wolf Sheep example to include PropertyLayer

[mgemaakbar]

Hello everyone,

This discussion is a follow-up of @EwoutH 's comment on my pull request . Sorry for the late response.

Before implementing anything, I would like to confirm if my understanding of the "soil fertility" idea is correct. Currently the countdown and grass_regrowth_time attribute in GrassPatch class are used to schedule initial grass growth and grass growth after the grass was eaten, which is shown here at the code.

And what I understand at the moment is: this "soil fertility" idea is basically just moving countdown and/or grass_regrowth_time values to a PropertyLayer which @EwoutH referred here as the "soil layer"? Is my understanding correct?

If that is correct, then after the soil PropertyLayer is introduced -- does this mean that each cell can have different grass_regrowth_time?

For example, grass_regrowth_time of cell at (1,2) can be different from grass_regrowth_time of cell (1,3) depending on what is stored at the 'soil layer' PropertyLayer.

Also what I understand is: currently all grass cells have the same (only one) grass_regrowth_time, which is shown here in the GrassPatch constructor.

Please let me know what you think, thank you in advance

[EwoutH]

And what I understand at the moment is: this "soil fertility" idea is basically just moving countdown and/or grass_regrowth_time values to a PropertyLayer which @EwoutH referred here as the "soil layer"? Is my understanding correct?

Yes, I think this would be quite interesting. Especially if soil fertility is some spatial distribution + some noise.

If that is correct, then after the soil PropertyLayer is introduced -- does this mean that each cell can have different grass_regrowth_time?

For example, grass_regrowth_time of cell at (1,2) can be different from grass_regrowth_time of cell (1,3) depending on what is stored at the 'soil layer' PropertyLayer.

Yes, we would lookup the grass_regrowth_time directly from the PropertyLayer.

[mgemaakbar]

Hi, what do you mean by "spatial distribution"? Do you mean that, determining the grass_regrowth_time of the cells shouldn't be completely random (by simply just using random number generator)?

I think I have an idea how to implement this PropertyLayer demonstration on Wolf Sheep (I can create a draft PR if you'd like to see and give me feedback). I can make the grass_regrowth_time to be completely random, but I am still not sure I understand about the "spatial distribution" part.

[mgemaakbar]

Is this a correct example of how "spatial distribution" is when it's visualized?

[EwoutH]

Yeah I think something like that would be great!

[mgemaakbar]

Great. Then I will update the PR soon.