Single Bot Sim?

[Lyndon]

Er I haven't really posted all that much but earlier this year I had a fair bit of fun playing with Darwinbots and just picked it up again to have something to do when my computer just sits there.
Anyway I have been trying an evosim using a zerobot but sadly haven't really got the patience or processing power for anything good (1.4GHZ laptop...), my current proper evosim consits of 50 zerobots with 100 base pair and has been running about 4 hours, no discernible change in behaviour but they are developing some interesting mutations that might allow for reproduction and locomotion, I have made them veggies as having nrg shooting bots bogs down my cycles a second, put gravity on so they fall to the bottom and added Brownian motion, any suggestions as to when they will learn to reproduce?

On to the topic I wanted to discuss, I have left my original evosim for the moment and tried something odd, I have create a sim governed by the same rules as the original except it has only 1 bot in it, my cycles a second are between 650 and 1060 and have had over 35 million so far. I was just wondering whether others have tried single bot sims and whether they are worth something since they are quicker than normal ones.
So far my little bot has 83 mutations, 3 genes and a few store commands. None of the genes are activate though.

[Numsgil]

When you're in the first stages of sim, 1000 bots at 1 cycles/sec and 1 bot at 1000 cycles/sec are roughly the same thing, so go with whichever one you prefer.  You get the same number of mutated base pairs per second either way.

Reproduction usually happens in the first few million cycles of the sims I've run.  Be patient

[ikke]

To me they are not the same. Parallel and sequential mutations are not the same. Furthermore one bot will never outcompete, or be outcompeted.


One of the things I looked at was testing clones in a sim to see how long it would take for one to outcompete the other. I don't know the numbers by hart, you can find them in one of my previous posts but the conclusion was that for the population sizes tested (up to 400) one clone would replace the other due to random factors before any meaningfull evolution could take place

[Numsgil]

Competition requires action.  If your zerobots don't yet do anything, then there's no natural selection and it's all the same.  Once you get a zerobot that can reproduce, it'll create its own competition.  So for the purposes of zerobots, 1000 bots or 1 bot, it's essentially the same.

[Testlund]

My guess is that you'll get better results if the mutations are spread out over many bots, rather than just one bot getting a all the random mutations. If it is spread out the bots will mutate less, each getting it's own set of mutations where a few might get the right ones. With just one bot I guess you wouldn't be able to keep any good mutations. They'll get removed as soon as they appear so to speak. Just my theory as I haven't tried it, but it would be interesting to know how it goes. I've toyed with this idea myself actually.

Also I would recommend you don't have any copy mutations on as that might destroy the mutation that causes replication, at least in the beginning.

[ikke]

Competition requires action.  If your zerobots don't yet do anything, then there's no natural selection and it's all the same.  Once you get a zerobot that can reproduce, it'll create its own competition.  So for the purposes of zerobots, 1000 bots or 1 bot, it's essentially the same.

Even without competition, my intuition tells me it is not the same. I don't have the (insight in the) math, but to me a simple example is the following: in sequential mutations a negative mutation has to be undone and positive mutations have to occur, whereas in parallel mutatin the negatives don't have to be undone.

[Lyndon]

My zerobot did reproduce (3 times) but it lost its reproduction gene, surely the new bots will cause some sort of competition and since they are already thoroughly mutated they should themselves find a way to reproduce more quickly than the first.

[Peter]

Competition requires action.  If your zerobots don't yet do anything, then there's no natural selection and it's all the same.  Once you get a zerobot that can reproduce, it'll create its own competition.  So for the purposes of zerobots, 1000 bots or 1 bot, it's essentially the same.

Even without competition, my intuition tells me it is not the same. I don't have the (insight in the) math, but to me a simple example is the following: in sequential mutations a negative mutation has to be undone and positive mutations have to occur, whereas in parallel mutatin the negatives don't have to be undone.

In zerobot sims it is often just waiting till one reproduces. That way it is still the very same.

My zerobot did reproduce (3 times) but it lost its reproduction gene, surely the new bots will cause some sort of competition and since they are already thoroughly mutated they should themselves find a way to reproduce more quickly than the first.

He lost it, did you have high mutation rates.

[Lyndon]

Yes high rates.

[ikke]

If you have too high rates your bot will mutate more often than it reproduces. This effectively allows your species genome to degrade, by accumulating random mutations. Lower it. My sim has about 400 bots and a default mutation rate. Evolution is about patience. If you have the reproducing genome you may want to restart a sim with more specimen of the same species.