Self-organizing is a generative project I launched since 2017 September.
The whole system is based on 4 basic behaviours – growth and death abstracted from DLA system, as well as repelling and attracting abstracted from Differential Growth. Behaviors are applied to particles according to density. It’s a hybrid system of basic behaviors, which produce diversified form and structure from one single model.
For more detail please check: 【算法】基础行为的自组织|Self-Organizing | 20170924-20171227
Behaviors – System
It’s a density system, developing itsself to fully occupy the space. The whole system is based on 2 types of particles playing special roles: Seed and Body. Accordingly 2 pairs of basic behaviors are implemented.
As system develops, points are changing between 2 types according to density.
The growth and death part, behave like the DLA system (outer edges and branches are more likely to grow). The repelling and attracting part, behave more like Differential Growth (particles are keeping a distance from each other and bulges will increase).
When particles get too close to each other, they repel to maintain the density. Meanwhile, particles with highest growth weight（Low density particles） will keep “grabbing” others closer, to increase differentiation, making branches more like branches. I call this part of behavior “Dynamic Process” – it’s an addtional part above the aggregation system, giving the system dynamic variables, thresholds become more malleable than traditional DLA-like systems.
These behaviors, are opposite thresholds. If we cancel any single one from pairs, the system would become imbalanced and single-polarized, and totally lost ability of self-organizing. It is the neutral and dynamic state that develops the system, from which intelligence evolves.
System – Prototypes
If we set higher weight to one specific behaviour, the system will show more feature of this prototype – growth with high dynamic, showed features of differential growth and those with low dynamic, showed rigid features of DLA system. Growth with high differentiation, showed a trend of more branches, and those with low differentiation, showed a trend of more bodies.
Based on the above, the system is a mixture of several prototypes, from which we extract individual behaviors. By controlling weight of behaviors, we can control the weight of prototypes, adapt the system to specific circumstances. As a “created” mix system, it is defined by the ratio and weight we set, for different goals.
System – Derivatives
One of the essential feature of intelligence, is reacting to its environment. By changing variables of environment, time and other factors, the system shows intelligence and responds to surrounding, adapting themselve to field and environment – such as multi-colony growth, obstacle growth and field growth.This means the possibility to utilize variables of complex field and society in practical application.
Environment as Variable
Time as variable – Aggregation
Experiments at earlier stages are mainly based on 2D plane, I transferred density information given by accumulated growth(time as a variable) into height – by that, artificial landscape is generated.
Self Organizing 3D
To improve the performance and application, after months, I started to work on a 3D version, a full-version algorithm with simpler methods for solving extremely complex problems in a 3D level, the density detection is developed as a directional-density based on proximity.
System – City
Behaviors of system can be seen as keys of gene, the properties of system are defined by properties of behaviors.
If we take behaviors from other prototypes like React – Diffusion system, each point of this system can hold more complex information. Properties of any kind of behavior, can be directly combined with existing system.
Human society is a multi-dimensional system of basic behaviors, and basic components hold far-more complicated informations – just like chemicals of swarm.
Architecture is a explicit form of social hierarchy, it could also be a self-organization – we can find a better solution based on balanced powers – a bottom-up solution. In my schoolwork BIT-EROSION, I started my very first experiment on iterative spatial structure.