Bots and Simulations > Simulation Emporium

Q: shortest zerobot reproduction time

(1/4) > >>

ikke:
Out of curiosity: what is the shortest time anyone has ever evolved a selfreplication from a zero bot?
At the first attempt I had 100 new bots after 150 k cycles. After that poor dynamic cost set up killed them all  
Now (my second attempt) I am at 1.7 m cycles and nothing

EricL:
Depends on several factors.  Longer starting zerro-bot genomes should shorten the time it takes to evolve a replicator.  Also, a low waste threshold may result in replication due to altzheimers before the DNA to replicate has evolved.  

In the sims I've run with relativly short genomes (30-100 base pairs) and no waste threshold, I generally see replicators in less than 10M cycles.

Numsgil:
From my experience, if you allow viruses you'll first end up with primitive viruses that create new viruses by storing running through all the memory locations and incrementing them.  A side effect of this will be reproduction.

bacillus:
The problem is with intentional reproduction is that it requires several parts put together in the right order. Feeding is easy enough, you just need .shoot dec, but reproduction needs the condition which needs to evolve first, a logical value between about 20 and 80, not to mention that without a conspec the new bot will eat its parent or be eaten, and all this has to survive further mutations. At best, a low mutation rate would ensure that replicators survive for a maximum amount of time, even though they will take a long time to appear. How do you apply evolutionary pressure to force reproduction?

EricL:

--- Quote from: bacillus ---How do you apply evolutionary pressure to force reproduction?
--- End quote ---
Its the other way around.  How can you have evolutionary pressure without reproduction?  You can't.

Zerobots eventually hit on some way to periodicially reproduce by unconditionally writing to .repro or .mrepro.  Point mutations cause this, the only mutation type that occurs outside of reproduction.   Some bots end up with DNA that attempts reproduction every cycle.  They rely on the body limit to determine when reproduction actually happens.  Others do it periodically e.g. every 1000 cycles as they walk through all the sysvars incrementing them using .robage or similar.  You don't need conditional logic to evolve replicators.

In fact, reproduction is actually much easier to evolve then feeding via .shoot.  It's quite a sophisiticated thing for there to be sufficient selection pressure to favor .shoot.  I've never seen it occur in zerobot sims.   There has to be significant benifit for it to be favorred and selected for and without aiming, which requires vision, which is very sophisitcated, random shooting doesn't tend to convey enough benifit to be selected for in my experience.  At least not yet.   What is way more common is for bots to vary some other aspect such as their speed to maximize the chances of hitting nrg shots from shepards.  You also see reproduction itself used as an indirect means of feeding.  If you reproduce when you have the nrg/body to do so, then more of your phenotypes will happen to be where the food is, such as near a shepard or the act of reproducing will slow you down when you hit an nrg shot, increasing the probability you or your offspring will hang around where the food is.  I've seen early replicator zerobots rely on this mindless coincidence between reproduction and feeding many times.

Conspec recognition is only important if you hunt or reproduce sexually or otheriwse have some something to do with others after you're born, which as above, all requires sophisticated DNA I've never seen evolve.  There is no need to recognize your conspecs if you have no way to harm them or don't need them to reproduce.

Navigation

[0] Message Index

[#] Next page