Code center > Suggestions

Bringing back .sexrepro

(1/6) > >>

Botsareus:
You can make .sexrerpo possible by grouping simular commands under cromozones.  Robots need to see each other to reproduce (I might add restrictions later like the robots must share 50% dna to reproduce)

I ran into some problems; I will wait after gene structure gets done.

Peter:
Okay, *warning* I have a feeling I am going to say something stupid.

Bot1 and bot2 together create bot3.

So you get the even genes from bot1 and the uneven from bot2

If bot1 has 10 and bot2 10

Then bot3 gets

from bot1

2
4
6
8
10
from bot2

1
3
5
7
9

The genes get then sorted like 1,2,3,4,5,6,7,8,9,10, it is simple.

if bot1 has 4 and bot2 has 7

from bot1
2
4

from bot7
1
3
5
7

The sorting goes like 1,2,3,4,5,7

In this case you probably get from different species, some non really living ones aka non funtioning 100% there could happen some interesting things. Different species can't sexrepro in real live eather. Have ever seen a cat-sheep, well have you.

With the same specie you get probably good children, becouse the genes are mostly in the right order.


My idea is pretty simple and there has been some talking about this(in times I wasn't around here), that I think something is completely wrong with my idea.

But okay what is wrong with my idea, I want to know.

Botsareus:
 corrected (I realized you can revice the system, by running the same dna for multi outputs. I know you people still don't have a clue what I am talking about, but I promise to post examples as soon as I write them out. The general idea is currently in my head...)

"some non really living ones aka non funtioning 100% " I know how to fix that little part. It means the dna will run * 2 or * 3 times slower but I think its worth it... I have yet to write my thoughts out on paper before I type them... stay tooned , thats all I want to say. , thanks...

* 2 or *3 slower I mean (don't ask yet)

a robot with 20 commands will run 40 to 60 commands. Stay tooned.


--- Quote ---(I might add restrictions later like the robots must share 50% dna to reproduce)
--- End quote ---
I leave this up to the bots. But, onless its e-grid mode the bots cannot reproduce with the plants. And the simularity limit does not go above 90%

Testlund:

--- Quote from: Botsareus ---(I might add restrictions later like the robots must share 50% dna to reproduce)
--- End quote ---

Why not at least 99%? All mammals share 99% of the genes but can't reproduce with other mammal species. With only 50% it would be like a human trying to mate with a mushroom!

EricL:

--- Quote from: Testlund ---All mammals share 99% of the genes but can't reproduce with other mammal species. With only 50% it would be like a human trying to mate with a mushroom!
--- End quote ---
This is not strictly correct.  All mammals do not have 99% of their genes in common.  In general, the percentage is quite a bit lower.

If you just compare DNA without looking at genes or considerring what is under selective constraint, the difference is large.  Chimps and human DNA differ by about 2% accross the entire genome.   Gorillas and humans by about 3% and so on until you get to the most distance mammals from humans (where the common ancestor is more than 100 million years ago) such as the platapus where it's like 50% I think.  Humans for example have "lost" 100's of olfactory genes which dogs and many other mammels still posses.  In our genomes, they have become pusedogenes - fossil genes - and have mutated greatly since they are no longer under selective constraint tho you can still trace their ancestory.

If you just look at working genes which can largely be assumed to be under conserving selecton pressure, it is true that a large percentage of genes in mammals are shared tho it is still quite a bit lower than 99% and there are many polymorphisms along the diffent evolutionary braches even for shared genes.  That is, the genes may be homologous and perfrom the same function, but they may differ greatly in their actual coding and fucntion.

Regarding .sexrepro, as I have stated elsewhere in depth, I am very strongly opposed to hard coding into the simulator any concept of sexual reproduction including any concept of assumed genetric structure, chromozones, recombination strategies, mate selection, etc.  So, while others are of course welcome to implement whatever they like, I will not be making .sexrepro work.  Period.

What I am totally willing to do is to enhance the base mechanisms for sharing DNA between willing bots so that sexual reproduction can be better hand coded or evolved in the bot DNA itself.  If someone would like to propose enhancements to this end, I would be very willing to entertain and potentially implement them.   Such suggestions could be in the form of better DNA range selection mechanisms beyond .thisgene, enhancements to viruses, asexual reproduction with a specified subset of DNA, sperm shots, etc.  I view these are low level mechanisms which can be used for sexual reproduction but also potentially for other purposes and in particular, do not put constraints upon DNA structure, recombination method or mate selection.

Let me say it again so I am clear.   Any proposal which requires a specific DNA structure, recombination algorithm or mate selection method (such as simple proximinty) for sexual reproduction is a nonstarter in my book and will not be implemented by me.

Navigation

[0] Message Index

[#] Next page