Messages

46

DNA - General / Re: Newbie question about coding behaviors into DNA.

« on: April 20, 2011, 03:15:02 AM »

Welcome to the wonderful world of emergent behaviour, where simple rules give complex and unexpected results. It's part of the game...
When programming my bot, Gimmick I learned that swarming is actually not good behaviour in this sim. That is to say, a species avoiding it own defeats a swarming variety in a death match. Solitary rules. Reason being that the odds on a species level the amount of food consumed increases if they look for different prey.

47

Biology / Re: human mutation rate

« on: April 06, 2011, 04:36:17 AM »

Even at the 1% (protein) code and 100 mutations level the chances of  being hit by a mutation is 1-(1-0.01)^100=63%. For 2% and 200 the number is 98%. I still say that mutations cannot have an equal occurrence across the genome.

48

Biology / Re: human mutation rate

« on: April 02, 2011, 03:24:01 AM »

A mutation in a functional part of the genome (which is certainly bigger than 5%)

If the definition of functional gene means functional for the phenotype, the 5 % is a low estimate. (A large part of DNA are selfish genes, genes with no benefit to the organism but the genes are good in inserting copies of themselves.) Higher estimates of % lead to higher numbers of broken genes.

Chances are very high that it will remain neutral or have a very small effect (positive or negative) on fitness.

Common assumption is that most mutations are detrimental.

49

Suggestions / Re: minor interface redesigns

« on: March 27, 2011, 03:26:11 AM »

With my revision of 2.45 i was thinking about redesigning the interface:

* You can only select robots when the simulation is paused.

Why?

* If you select entire organism, you can drag the entire organism.

Yes

* If simulation is running and you are zoomed in, the screen location changes with the robot, i.e. the camera follows the robot around.

YES

* You only need to click once to select the robot, you do not need to double click.

YES

* You can save dna to clipboard from a drop down menu.

Yes

*Instead of hash, the dna will contain a simplified stream of mutations accumulated, then the program can redisplay it as 'old mutations' in a new simulation (the cycles info will not be saved)

The hash doesn't work. Fix it or drop it. If I restart I don't need old mutations

*The first line in the robot file will contain it's dna size, and adjust the default mutation rates accordingly.

No? Not quite sure what you want to achieve. When it comes to mutation rates I want to be able to make the odds much smaller by default

* You can save and load mutation rates from a drop down menu, the default file extension will be bla.mrates.txt

Why?

* Instead of un-hiding or hiding the dna view in robot info , (old way: to hide you press "close details") The textbox 'undocks' if you hover over the view mutations or view dna buttons, which I will probably replace with picture boxes. I got the idea from the way visual studio .net shows and hides un-pinned tool bars.

Ok

* If someone is running multiple instance of the program, they need to know what chart (if any) belongs to what instance. Therefore the current randomization seed will be displayed on each chart as well as the mdi-form.

I dont usually run multiple instances with graphs but seems like an idea.

50

Darwinbots3 / Re: Chromosomes and Sexual Rproduction

« on: March 22, 2011, 01:19:05 AM »

crossover of DNA within a chromosome pair?

51

Biology / Re: Evolution of DNA

« on: March 15, 2011, 11:59:11 PM »

I'm not sure how factual the idea of the scripts are, because I haven't heard of that idea outside of that site, but leaving aside reality for the moment I think it's a reasonable way to program super structures like multicellular animals, etc. 

http://en.wikipedia.org/wiki/Hox_gene This to me is an example

I might experiment with some of the ideas for DB3.  Maybe "slow" DNA that bot authors can use to program a bot's lifecycle (larva stage: feed; adult stage: combat) in a way that is amenable to mutations and evolution.

I think the way to go would be codule - codule stack - stack of codule stacks.

52

Biology / Re: Evolution of DNA

« on: March 15, 2011, 09:37:06 AM »

I read it. It is good. I have no background in biochemistry but the site is very good in giving very much information without making my head spin. I'm sure the namegiving helps. It really opens (at least to me) the black box of what kind of processes are needed at a chemical level to support life

I think the scripting part is off though. Specifically on the low level repetition of building blocks. Basically the idea represented is that making the cell wall consists of repeating an exact sequence of attach molecule x. It would mean the cell wall of each cell of an organism (even a species!) would consist of exactly the same number of molecules. Maintenance, for instance cell wall repair would be a bitch. At which repeat of insert molecule x should the script restart to repair a cell. I think it works more along the lines of as long as chemical balance is off (because the cell wall is incomplete or damaged) make and place molecule x. Chemical equilibrium. The author has no problem with the concept at other instances, so why he deviates here is a mystery to me.
On the other hand some scripts will work that way. I have a definite number of ribs, and exactly two eyes

53

Off Topic / Re: age

« on: March 15, 2011, 07:04:42 AM »

36

54

Biology / Re: human mutation rate

« on: February 05, 2011, 08:46:15 AM »

I just did the quick math: assuming 150 mutations per generation and 95% of DNA non functional the probability of not having a single mutation in functional DNA is 1-.95^150 (assuming random spread of mutations through the DNA. This is a one in 2200 chance.

Maybe there's also some intelligent mutation-ing going on.  Maybe some parts of the genome are allowed to mutate more than other parts.

I just realised I omitted DNA correction mechanisms. They may have enormous impact on the probability of mutations occurring in functional vs in non functional DNA. This would invalidate my random spread mutations assumption.

55

Biology / Re: human mutation rate

« on: February 05, 2011, 04:01:49 AM »

Keep in mind how many embryos that are fertilized never result in a viable pregnancy. Not just miscarriages, but "pregnancies" of only a week or less that we never really know about.  So there's a bit of a survivor's bias.

Sure, this begins with the sperm's rat race to fertilize. But estimates of spontaneous abortions are only (only is used in terms of the 5-10 mutations in functional code) in the 10-25% range. With half the women pregnant within 3-4 months these numbers don't seem to add up. I have to admit I haven't done any math and probabilities are hard to estimate beyond the simple draw

Maybe there's also some intelligent mutation-ing going on.  Maybe some parts of the genome are allowed to mutate more than other parts.

For me this means either the code is non functional and already discounted in the 95% mutations in junk or it is redundant, with other code partly or entirely absorbing loss of functionality

56

Biology / human mutation rate

« on: February 04, 2011, 07:20:09 AM »

I came across this article claiming people have about 100 to 200 mutations per generation. To me this seems high. Even if only 5% of the genome is functional this means on average the genome breaks in 5-10 places
http://news.bbc.co.uk/2/hi/8227442.stm

57

Evolution and Internet Sharing Sims / Re: My 2.44.1 first-bot results so far...

« on: January 27, 2011, 02:31:00 AM »

It also makes a certain amount of sense: you can either evolve conspec recognition, or you can just evolve so that you don't die by choosing a "lane" and sticking to it.

I hadn't thought of that, but now I do it is even more subtle than conspec: it benefits relatives but nothing else

58

Evolution and Internet Sharing Sims / Re: My 2.44.1 first-bot results so far...

« on: January 26, 2011, 01:53:46 AM »

2.) It's children have less body weight, but sometimes it reproduces with two children. So it is like 25 , 25 , 50 where 50 is kept by the parent.
3.) It uses -6 shoot instead of negative -1

fairly normal mutations

1.) It turns to the 100 degree direction as soon as it reproduces by reproducing to the side, either the left or the right at random. (This is the coolest because of the minor velocity changes it does to achive it)
The result is a 'wall' of robots sweeping the screen.

this is the fun part

59

Darwinbots Program Source Code / Re: Re-writing DB in JAVA

« on: January 07, 2011, 05:53:30 AM »

OT (but that hasn't ever stopped me or anyone else on this forum): eta for db3?

60

Bot Tavern / Re: How will leagues work when we update the program to use chloroplasts?

« on: December 06, 2010, 03:30:43 AM »

Perfect balancing is indicated by best chloro & best regular should have even odds at winning. You could use it to set various constants