What are specific implementation problems or bottlenecks that need to be resolved to implement an agent-based model or ABM simulation software such as NetLogo?
TL DR version:
What are the problems associated with implementing ABM software, in particular NetLogo, so that it can use multiple processors?
What are the attempts to solve these problems?
What is needed to parallelize NetLogo?
How else will NetLogo run in parallel?
Longer version:
NetLogo is currently performing simulations by sequentially passing one agent at a time. Although it is possible to run several models on the same machine (in separate NetLogo instances) or even in different configurations of a specific model (in one NetLogo instance) using BehaviorSpace [6], one model works on one CPU [2, 3].
There is also a pseudo-parallel mode of behavior for handlers using the ask-concurrent command [4, 5], but this does not entail the parallel execution of several processors.
NetLogo, Haskell ( HLogo ) [1]. , , " , HLogo , NetLogo, , , . , STM, HLogo ."
, - . , NetLogo /parallelism [ ] [2, 3], , , , .
?
, :