Contents |
Investigating agent interaction & encapsulation; taking molecular biology of the cell, and symbiogenesis (endosymbiotic theory) as a computational paradigm. In hope of an Artificial Life system.
Haha that title is full of stupid big werds and stuff :) HOORAY!
Concept
To create a computer simulated world in 2D, possibly 3D (though that is likely to just dilute interaction between agents and the world). The world will be populated with cell like agents that will be "evolved" in a similar incremental way as a genetic algorithm. Agents survive by getting energy from light, finding molecules from the environment and ultimately storing energy in molecules to explore new parts of the simulation as old environments stale and become uninhabitable.
I hope to see something analogous to a shallow bit of ocean with lots of plankton like Agents buzzing around, keeping to the light and resources. With a simulated membrane and molecule system, I hope to make it possible that a more predatory approach for an agent’s survival might ultimately be possible.
Light as Energy:
Light and interaction with certain molecules inside an Agent gives it energy, once an agents energy is zero it can either lay dormant without any processing until it is possibly enveloped by another Agent or it can be completely removed from the simulation. This is dependent on the age of the agent. If it lived a long life then it's genome remains to be consumed by another agent through a process similar to symbiogenesis.
An agent can lose energy by either using the molecules that give it energy to do something else more complicated or it might just move out of range of a light source.
Molecules as Messengers:
Each agent moves molecules across its membrane or onto its membrane, from outside-in or vice versa. Molecules on the membrane dictate what can be transported in and out, and if a virus or other agent can manipulate a given agents life. To bring in molecules that are in high abundance inside the Agent takes energy to do the inverse is also true. A wandering virus only enters an Agent if its membrane holds the required sequence of molecules at the time the virus stumbles upon it, this interaction can keep happening if the virus is on the same vector as the Agent.
Overall I would like to see co-dependent relationships/interactions form between agents. Possibly see them spiral around each other as they share molecules required for inverted directions of movement etc. Or even forming colonies close together, sharing energy molecules when they receive a certain molecule from a nearby Agent. The key to this kind of behaviour I feel is to not add too many complexities to the underlying simulation rules. Conway’s game of Life has exceedingly simple fixed rules yet complex behaviours arises, if you add on too many extra clauses the emergent complexity tails off and just becomes random unstable noise.
Molecules as Energy Stores:
Membranes Encapsulate Agents:
Entities
Agents:
The free roaming cell like agents I want to see develop as the simulation executes. They will have a "membrane" and "genome" made up from numerical strings that directly hash to molecules, which indirectly have a one to one relationship with the simulated world actions: such as moving, constructing larger molecules, or breaking molecules down.
Molecules:
Molecules are nothing more than a numerical string with a life span. After a certain amount of time exposed to various intensities of simulated "light" they will break down in to fixed product molecules. However inside an agent: interactions with other molecules inside the membrane are considered and might well change the resultant products or stability of the molecule.
Viruses:
Viruses are given a fixed random velocity inside the game world on creation. They look at an Agent’s membrane-numeric-string and decide through a fixed hash if they can paste in their own numeric string into "crossover points" inside the agents genome. The virus only does this when in proximity to an Agent; they are in control of the crossover as an agent might be in a forced dormant state by yielding its processing. After the interaction the virus continues, however the viruses "genome" mutates randomly with time and exposure to light.
Molecule Wells:
Molecule Wells are sources of molecule objects and spew out random molecules between random thresholds for each well on creation; i.e. some wells might be able to spew out longer more complicated molecules that have more products before becoming stable. Molecule Wells have a fixed random life time and on death a new one will pop up within another location that is in a realistic reach of the majority of agents.
Light Sources:
Light Sources in the simulated world are fixed in location but have random intensities through time. The intensity radiates as a circle (possibly a sphere depending on if I need to go 3D) decreasing at an inverse square rate, similar to here on Earth.
Links to stuff
COMPUTAR
http://en.wikipedia.org/wiki/Conway%27s_Game_of_Life http://en.wikipedia.org/wiki/Artificial_life http://en.wikipedia.org/wiki/Artificial_chemistry http://en.wikipedia.org/wiki/Game_theory
Self Organisation:
http://en.wikipedia.org/wiki/Stigmergy http://en.wikipedia.org/wiki/Quorum_sensing
(These ideas tie in nicely with cellular automata and proximity rules, but along with the concept of membrane interaction and gene “expression” Quorum sensing might be a better concept to use.)
Genetics:
http://en.wikipedia.org/wiki/Operon http://en.wikipedia.org/wiki/Lac_operon
(It might be better to think of Operons instead of individual genes, this simplifies the problem greatly, whilst still maintaining a close resemblance to a biological system.)
http://en.wikipedia.org/wiki/Gene_regulation http://en.wikipedia.org/wiki/Gene_expression
(Might be nice to keep any negative/positive feedback operations in my program resembling this.)
Microbiology:
http://en.wikipedia.org/wiki/Retrovirus (Possibly an effective mechanism for cross over without being particularly destructive. With the possibility of specific useful “genes” becoming prevalent within a population of agents.) http://en.wikipedia.org/wiki/Endomembrane_system (Specific to agent encapsulation and interaction.)
Symbiogenesis:
http://en.wikipedia.org/wiki/Symbiogenesis http://en.wikipedia.org/wiki/Mixotricha_paradoxa http://en.wikipedia.org/wiki/Endosymbionts
(All this stuffs is a nice way to evolve more complex agents, by having them capture each other.)