Self-inspecting replication

1. Initial replicators

Self-inspecting replicators are machines which are able to scan its own structure and rebuild it. Thus the building information need not to be stored otherwise. We will conduct experiments with three types of self-inspecting replicators possessing different size and topological complexity.

  • The first replicator type is rather minimal and consists of six cells in a ring structure. It has a scanner cell reading its structure and a constructor cell build the offspring. Moreover, an attacking cell provides the cluster with energy consumed from its surroundings and a propulsion cell drives the cluster around.
  • The second variant also has a ring structure but with more cells featuring the same function. However, many of its cells are not mandatory.
  • The last replicator possesses a much more complex spatial structure related to a deformed hexagon.

2. Evolution experiments

2. 1. Setting up

We will run a simulation for each replicator type. The idea is to start with a small universe filled with energy and some few initial replicators. More precisely:

  • universe size: 1000 x 1000 units
  • 5000 randomly distributed rectangular blocks of size 8 x 4 as "nutrients" 
  • 20 replicators

The simulation parameters are chosen to represent a "friendly" universe, e.g. low mutation rates, low radiation, low energy cost for cell functions, etc. The simulations at initial configuration can be download here:

 

2.2. Simulations

2.2.1. Small replicators

At the beginning the number of replicators growth exponentially until all resources are exhausted. After than, a equilibrium is established where the number of around 6000 replicators stays constant. The video on the left hand side below shows how the replicators consumes the last free resources. After that, they need to consume each others. However, due to mutation, minor optimizations took place and led to better adapted individuals. One can observe the rising of dense colonies at some time. This phenomenon is shown in the picture on the right hand side taken at 114k time steps. Each shining dot represents a different individual/replicator.

The colonies aim to concentrate energy in a hostile universe and seems to be only stable when the universe is relatively dense because of the surrounding pressure of other materials. The high density inside a colony provides easy energy consumption for the replicators. Since they are more or less perfectly adapted to its environment, no further significant adaption took place. A more detailed look reveals that the replicators have lost the ability the move on their own because there is no necessity anymore.

Below you find two videos: The left shows how the replicators capture all available resources in the small universe. The right video gives a detailed glance of the processes inside a colony.

After this initial phase we gradually increase the universe size to 4000 x 1000 units and gradually increase the simulation parameter cell function properties -> weapon -> energy cost to 1.4. These changes, on the one hand, lowers the surrounding pressure and lead to dispersion of the colonies. On the other hand, energy consumption becomes more difficult. Suddenly, there is a necessity for adaption. The replicators develop moving capabilities in order to consume resource more actively. After few million time steps and letting the above parameter fixed, a new equilibrium merge. They consume each others and generate offspring. But the number of replicators stays constant. Such behavior can be observed in the video on the left hand side below taken after 22 million time steps of evolution.

We then further increase the above parameter up to 2.4. This leads to a disturbance of the equilibrium since the replicator lose energy too fast. However they found a way to survive. Each time when resources are locally concentrated they take the chance and consume and replicate as fast as they can. After that, they diverge and most of them are dying. The destroyed/malfunctioned replicators are drifting through space. When many of them come close they will be consumed by others. Thus a more complicated equilibrium emerges where material waves collide periodically and are rapidly consumed by few existing replicators. The number of replicators grow very fast for a short time. Then, most of them are dying and forming a new material wave. One can observe this phenomenon on the video at the right below hand side.

The replicator resulting from evolution after 70 million time steps is depicted below. In the video it is shown how one will spread in a "friendly" universe. An interesting behavior one can observe in the simulation is that many such replicators actual do not replicate whereas only a few generate much offspring. This might be advantageous if many of them compete for limited resources.

2.2.2. Large replicators

The evolution of this larger type of replicator follows a slightly different pattern. On the one hand, its larger structure offers more potential for optimization and thus adaption to new environments. On the other hand, the replication process takes more time and energy. As a consequence, the development of periodic material waves occurred in a much weaker form because the replicator were not fast enough the capture the material.

Below you can find the evolution product after 82 million time steps and a video showing its procreation. The replicator moves in very abrupt and sometimes suicidal patterns. This aggressive behavior seems to be advantageous in universes with rare resources. Its structure has evolved but is still relatively close to the initial configuration. One can also observe that its structure and function does not change much in later times. In comparison to the beginning much more tokens are rotating on its cells leading to faster activities.

2.2.3. Complex replicators

The largest variant of initial replicators develop smaller structures on quite short time scales. As for the large variant above, there is more freedom for optimization and one can observe more complex behavior patterns in comparison to the small initial replicator version. This fact influences also the overall structures of the universe. There is a higher variety of clusters of different sizes. One can also observe no periodic material waves. Instead there are more chaotic movements and local agglomeration of replicator machines. In the video on the left hand side below this phenomenon can be seen.

Such conditions might be more suitable for evolution towards higher complexity. In the simulation, after 26 million time steps a completely new kind of structure had emerged. It attracts one's attention because of its massive internal processing. It appears to be like a growing crystalline structure with no active movements or energy consumption. Procreation is done by breaking the structure apart due to physical impacts. The parts are able to grow again. A magnification of this surprising product of evolution is shown on the right hand side below.

That exotic creature lives in coexistence with the other evolved replicators. An example of such replicator is shown below together with a video showing its spreading.

4. Conclusion

All three types of replicators undergo most changes in the first million time steps due to mutation and natural selection. Higher complexity in the initial configuration offers more possibilities for optimization. However, as soon as the environmental conditions (simulation parameters and universe dimensions) are fixed, evolutionary adaptation is slowing down. Silent mutations also seemed to occur rarely.

All types of initial replicators evolve to variants which are much faster in procreation and consumption of resources. The structural most complex initial variants also develop more complex behavior patterns.

A surprising product was the arising of crystal-like structures which are able to grow but have no active movement, energy consumption and replication functions. They reproduces themselves by breaking apart by outside influences. The parts are than mostly still functioning.