Update multiagent vs. union performance docs for Julia 1.11 (#1091)

* Update multiagent vs. union performance docs for Julia 1.11

* Update performance_tips.md

* Update performance_tips.md

* Update performance_tips.md

* Update multiagent_vs_union.jl

* Update performance_tips.md

* Update performance_tips.md

* Update Project.toml

* Update Project.toml

---------

Co-authored-by: George Datseris <[email protected]>

Author: Tortar

Project.toml

@@ -1,7 +1,7 @@
 name = "Agents"
 uuid = "46ada45e-f475-11e8-01d0-f70cc89e6671"
 authors = ["George Datseris", "Tim DuBois", "Aayush Sabharwal", "Ali Vahdati", "Adriano Meligrana"]
-version = "6.1.11"
+version = "6.1.12"
 
 [deps]
 CSV = "336ed68f-0bac-5ca0-87d4-7b16caf5d00b"

docs/src/performance_tips.md

@@ -134,12 +134,7 @@ t = joinpath(dirname(dirname(x)), "test", "performance", "variable_agent_types_s
 include(t)
 ```
 
-We see that Unions of up to three different Agent types do not suffer much.
-Hence, if you have less than four agent types in your model, using different types is still a valid option.
-For more agent types however we recommend using the [`@multiagent`](@ref) macro.
-
-Finally, we also have a more realistic benchmark of the two approaches at `test/performance/multiagent_vs_union.jl` where the
-result of running the model with the two methodologies are
+Finally, we also have a more realistic benchmark of the two approaches at [`test/performance/multiagent_vs_union.jl`](https://github.com/JuliaDynamics/Agents.jl/blob/main/test/performance/multiagent_vs_union.jl) where each type has a different set of behaviours, unlike in the previous benchmark. The result of running the model with the two methodologies are
 
 ```@example performance_2
 using Agents
@@ -148,7 +143,8 @@ t = joinpath(dirname(dirname(x)), "test", "performance", "multiagent_vs_union.jl
 include(t)
 ```
 
-In reality, we benchmarked the models also in Julia>=1.11 and from that version on a `Union` is considerably
-more performant. Though, there is still a general 1.5-2x advantage in many cases in favour of [`@multiagent`](@ref),
-so we suggest to use [`@multiagent`](@ref) only when the speed of the multi-agent simulation is really critical.
+As you can see, [`@multiagent`](@ref) has the edge over a `Union`: there is a general 1.5-2x advantage in many cases
+in its favour. This is true for Julia>=1.11, where we then suggest to go with [`@multiagent`](@ref) only if the speed of the
+simulation is critical. However, keep in mind that on Julia<=1.10 the difference is much bigger: [`@multiagent`](@ref)
+is almost one order of magnitude faster than a `Union`.
 

test/performance/multiagent_vs_union.jl

@@ -62,11 +62,15 @@ function agent_step!(agent::GridAgentFour, model1)
     agent.one += sum(a.one for a in nearby_agents(agent, model1))
 end
 function agent_step!(agent::GridAgentFive, model1)
-    targets = filter!(a->a.one > 1.0, collect(types, nearby_agents(agent, model1, 3)))
+    targets = Iterators.filter(a->a.one > 1.0, nearby_agents(agent, model1, 3))
     if !isempty(targets)
-        idx = argmax(map(t->euclidean_distance(agent, t, model1), targets))
-        farthest = targets[idx]
-        walk!(agent, sign.(farthest.pos .- agent.pos), model1)
+        farthest = 0.0
+        a = first(targets)
+        for t in targets
+            d = euclidean_distance(agent, t, model1)
+            d > farthest && (a, farthest = t, d)
+        end
+        walk!(agent, sign.(a.pos .- agent.pos), model1)
     end
 end
 function agent_step!(agent::GridAgentSix, model1)
@@ -93,31 +97,35 @@ end
 
 ################### DEFINITION 2 ###############
 
-@inline agent_step!(agent, model2) = agent_step!(agent, model2, variant(agent))
+agent_step!(agent, model2) = agent_step!(agent, model2, variant(agent))
 
-@inline agent_step!(agent, model2, ::GridAgentOne) = randomwalk!(agent, model2)
-@inline function agent_step!(agent, model2, ::GridAgentTwo)
+agent_step!(agent, model2, ::GridAgentOne) = randomwalk!(agent, model2)
+function agent_step!(agent, model2, ::GridAgentTwo)
     agent.one += rand(abmrng(model2))
     agent.two = rand(abmrng(model2), Bool)
 end
-@inline function agent_step!(agent, model2, ::GridAgentThree)
+function agent_step!(agent, model2, ::GridAgentThree)
     if any(a-> variant(a) isa GridAgentTwo, nearby_agents(agent, model2))
         agent.two = true
         randomwalk!(agent, model2)
     end
 end
-@inline function agent_step!(agent, model2, ::GridAgentFour)
+function agent_step!(agent, model2, ::GridAgentFour)
     agent.one += sum(a.one for a in nearby_agents(agent, model2))
 end
-@inline function agent_step!(agent, model2, ::GridAgentFive)
-    targets = filter!(a->a.one > 1.0, collect(GridAgentAll, nearby_agents(agent, model2, 3)))
+function agent_step!(agent, model2, ::GridAgentFive)
+    targets = Iterators.filter(a->a.one > 1.0, nearby_agents(agent, model2, 3))
     if !isempty(targets)
-        idx = argmax(map(t->euclidean_distance(agent, t, model2), targets))
-        farthest = targets[idx]
-        walk!(agent, sign.(farthest.pos .- agent.pos), model2)
+        farthest = 0.0
+        a = first(targets)
+        for t in targets
+            d = euclidean_distance(agent, t, model2)
+            d > farthest && (a, farthest = t, d)
+        end
+        walk!(agent, sign.(a.pos .- agent.pos), model2)
     end
 end
-@inline function agent_step!(agent, model2, ::GridAgentSix)
+function agent_step!(agent, model2, ::GridAgentSix)
     agent.eight += sum(rand(abmrng(model2), (0, 1)) for a in nearby_agents(agent, model2))
 end