I want to add an agenset of turtles to the variable TurtlesICanSee of a certain turtle that depends on that turtles properties. For instance, in one application I want to add only the turtle itself to TurtlesICanSee, in another application I want to add the two turtles (if there are any) with adjacent who-numbers (the turtle's own who-number + or - 1).
If I can figure out the first application by using who-numbers I think I can extend that to second application. However, I cannot figure out the first.
I tried
ask turtles [
set TheTurtlesICanSee turtles with [who = ([who] of self)]
]
but this fills the TheTurtlesICanSee of each turtle with every turtle.
I think I understand why; NetLogo thinks that I want every turtle x that has the same who-number as itself (x), i.e. every turtle. But I don't. For every turtle x I want every turtle y that has the same who-number as x.
Can anyone help me with this? Note that the solution that I need to the first application is one that can be generalized to the second. So not any way of adding a turtle to one of its own variables will do. I need a form of self-reference involving who (or a good argument against doing it this way I guess, but preferably the former).
Your code needs only a slight alteration to work, as follows:
ask turtles [ set TheTurtlesICanSee turtles with [who = [who] of myself] ]
Note the substitution of myself for self; http://ccl.northwestern.edu/netlogo/docs/dictionary.html#myself has an explanation of the difference.
But actually there's no need to involve who numbers. It's almost never necessary to use who numbers in NetLogo; there's almost always a simpler, more direct solution. A simpler solution is:
ask turtles [ set TheTurtlesICanSee turtles with [self = myself] ]
But actually it isn't necessary to use with at all. We can use turtle-set to build the desired agentset directly:
ask turtles [ set TheTurtlesICanSee (turtle-set self) ]
This is the solution I would recommend, for clarity and brevity, but also because it will run faster, since it doesn't involve iterating over the set of all turtles, as the with-based solutions do.
Related
I'm trying to design a model for the spread of infection from person to environment.
Turtles have a hand contamination variable that shows the percentage of their hands that are contaminated. I'd like to give this number to patches that they're passing, but I'm getting an error saying it's a turtle-only variable.
Is it possible to give a hand contamination number to the patch?
This is part of my code:
turtles-own [hand contamination]
patches-own [p-contamination]
ask patches [set p-contamination hand-contamination]
A patch can't ever refer to turtle variables directly: What if there is more than one turtle there...which one? What if there are none?
However, a turtle can access the variables of the patch it is standing on. So you would probably do this from the turtle's point of view: I think this also makes sense, logically, since it is the turtle visiting the patch, and contaminating it.
;; turtles contaminate the patch they are standing on
ask turtles [ set p-contamination hand-contamination]
Note that if there is more than one turtle on a patch, they will overwrite each other's values. So, you may need to add the amount, or otherwise blend the two values, rather than replace it.
If there are more turtles than patches, or you really want the patch to be the thing that in control, the patch can look for turtles and acess their variables with OF:
ask patches
[ let visitors turtles-here
if any? visitors
[ set p-contamination ..some expression..
So, there it depends on your needs, and you have to decide what that value is.
There is only ever at most one turtle:
[ Contamination ] of one-of visitors
Even if many turtles, pick one at random:
[ contamination ] of one-of visitors
Use the value of the most-contaminated visitor:
(max (sentence [ contamination ] of visitors))
Average the values of contamination
(mean (sentence [ contamination ] of visitors))
...or some other expression that you choose
Again, this is all overwriting the patch variable. If you need to take the patchs' current values for that variable, you need to decide how:
If already contaminated, should it:
leave value alone
add turtle value to current value of P-Contamination
save the max of the two values
save the mean of the two values
blend them in some other way
The final part of my design involves me recording down anytime a car breed drives into or in netlogo terms, is on the same patch or X and Y coordinate as the people breed as they navigate across the edge of the screen. Had this been java I could've done something like
if Car.xPostion == Person.xPostion
(Do something...)
But unfortunately I do not know how to do the same in NetLogo, all I've been able to do so far is just ask the two breeds by giving every turtle a boolean variable called movable and setting them to true and the rest to false, is there anyway I can check the two coordinates of two different turtles on Netlogo? This is all I 've been able to do so far.
to record-accidents
ask turtles with [movable? = true]
[
]
If you tried something like your java approach, it would fail because turtle positions are continuous and floating numbers are nearly always not equal.
If I have understood your question correctly, you have given a boolean variable called movable? set to true for cars and false for all other breeds. You don't need to do this, turtles know their own breed so you can do ask cars.
To answer your specific question, there are several ways to approach it depending on the perspective (sort of, which agent is in charge).
You could identify patches where there are accidents:, which is the answer to your question in the title (about identifying patches with two breeds).
let accident-locations patches with [any? people-here and any? cars-here]
if any? accident-locations
[ ask accident-locations
[ <do something>
But you can also take a turtle perspective. You could start from pedestrians who have been hit. This takes advantage of the fact that turtles can automatically access the patch variables (like turtles-here) for the patch where they are located:
let hit people with [any? cars-here]
if any? hit
[ ask hit...
or from cars:
let hitters cars with [any? people-here]
if any? hitters
[ ask hitters...
I am brand new to Netlogo and am coding a world with caching animals. They will go to their caches (I set them as blue patches) if their energy value falls below 10. They can find these caches based on a random 'memory' value given to them which is used as an in-radius value. That way, they will face and go towards a cache within their in-radius memory if they are about to die. I am starting to incorporate a home-base system where the turtle remains in a smaller area with their own caches. Does anyone know how I can make a patch belong to an individual turtle? This would allow turtles to have their specific caches in their territory. I have looked into using links and breeds, but links are only between turtles and making the individual breeds for the 50+ turtles at a time seems ineffective and complex. I apologize I don't have any code attempting to have a cache (patch) belong to a turtle, I don't know where to start. Any ideas?
If you want a turtle to remember anything (patches or income or anything else), then you need to assign a variable in a turtles-own statement and then set the value appropriately. Here's some example code fragments. They won't work, and you actual code would likely look a lot different because you will need some design about the conditions under which a cache will be assigned, but they show you what a variable solution looks like.
turtles-own
[ my-caches
]
set my-caches (patch-set my-caches patch-here) ; code when a turtle finds a new cache site
If you want a patch that belongs to a turtle to make that patch unavailable to other turtles, then also set up a patch variable to store its owner.
patches-own
[ my-owner
]
ask turtles
[ if [my-owner] of patch-here = nobody [set my-caches (patch-set my-caches patch-here)]
ask patch-here [set my-owner myself]
]
I suggest you do several NetLogo tutorials, then look at some library models (and understand them) before starting your own model. You need to understand basic concepts like turtles/patches, variables, ticks before trying to build a model.
I have a programme that sets up a number of different breeds of turtles. Each breed needs to have a leader turtle and a follower turtle. I have assigned these as global variables as they come up a lot in the code further down.
I need to assign these variables to turtles in the breeds and then create a link from the leader to the follower. There are a lot of conditions in the interface that determine how many and which breeds are created so i cannot assign by turtle number.
I am receiving an error (not all of the time) 'turtle cannot link with itself' which i presume occurs when they overwrite the first set command and assign the same turtle to the two variables. Does anybody know a condition i can put in that will allow it to set up everytime without the error. ( I have tried if statements, is-turtle?, one-of other, other)
breed [flinks flink] ;; linked turtles that will turn away from sources
globals [
flink-leader
flink-followers]
to set-up
clear-all
setup-turtles
reset-ticks
end
to setup-turtles
create-flinks 2 [
set flink-leader one-of flinks
set flink-followers one-of other flinks
ask flink-followers [create-link-with flink-leader]
ask flink-followers [set color pink]
ask flink-leader [
setxy 10 4]
ask flink-followers [
setxy 19.5 4]
]
end
to go
fd 1
end
There would be many different ways to approach this. Here is one that doesn't stray too far from the code you have provided:
to setup-turtles
create-flinks 2
set flink-leader one-of flinks
ask flink-leader [
set flink-followers one-of other flinks
setxy 10 4
]
ask flink-followers [
create-link-with flink-leader
setxy 19.5 4
set color pink
]
end
Note that your intuition about using other to make sure that the follower(s) is/are different from the leader was correct.
To understand what was going on, you need to grasp the notion of "context" in NetLogo. Some primitives, like ask, of and create-turtles, are "context switching": one of their argument is a code block (the part between [ and ]) that runs in the context of a particular turtle. Other primitives depend on the context in which the code is running: the primitive named other, for example, will report all the agents from a given agentset, except the one in the context of which the block is running.
In your version, you wrapped most of the code inside a code block provided for create-flinks. That meant that the code block was run once for each turtle that was created. So your calls to set flink-leader, set flink-followers and so on were all run twice, each time in a different turtle context. Can you see how that was messing things up?
Keeping track of the different context in NetLogo can be challenging at first (the frequent confusion between self/myself being a case in point), but once you get it, it should become easy and natural.
One last point as an addendum. You say:
i cannot assign by turtle number
Good! Never¹ assign anything by turtle number! It leads to brittle, error prone, more complex, less general, unnetlogoish code. If you think you need to use turtle numbers anywhere in your code, come ask another question here. Someone will most likely suggest a better way to do it.
¹ Well, almost never.
I'm new in NetLogo and this may be a too obvious question, but I don't see how to test if what I am doing is right.
I'm making a selection of an agent of breed1 (turtles) based on its distance to breed2 (crocodiles). I would like the crocodile to choose one turtle randomly from the ones around it, but with a higher probability to be chosen the closer the turtle is. Thus, I am using rnd extension and distance command.
My question is if the distance command is referring to the right agent (i.e. distance between the crocodile and the turtles):
ask crocodiles [
let potential_preys turtles in-radius max_distance
let selected_prey rnd:weighted-one-of potential_preys [ (1 - ( distance ? / max_distance ) ) ]
ask selected_prey [
scape
]
]
Before I get to your question, there is another problem I noticed with your code.
I had never realized this before, but NetLogo's semantics can make it tricky to model actual turtles! (At least when other breeds are involved.)
What I mean by this is that turtles refer to all turtles in the model, regardless of their breed. It means that, in your case, crocodiles are included in turtles, so when you say:
let potential_preys turtles in-radius max_distance
...crocodiles can be included in the potential preys!
Getting around this is easy enough, though: just choose another name for the breed that represent actual turtles in your model, e.g.:
breed [ tortoises tortoise ]
And then you can write:
let potential_preys tortoises in-radius max_distance
And now, for your question regarding distance, I think what you want is the distance to myself, where myself would be the crocodile that is selecting its prey. The myself primitive refers to the agent in the "outer" context of the block where you use it, i.e., the "calling" agent.
This gives you something like:
let selected_prey rnd:weighted-one-of potential_preys [
1 - (distance myself / max_distance)
]
Haha, I didn't thought about the turtles detail, indeed... ^^
Anyway it was an example, not the actual name of my breeds, so no problem, but thanks for noticing it!
Regarding the question itself, I also think myself will do, so I'll keep it like this, but now with higher confidence :D
Thanks Nicolas!