Darwinbots Forum
General => Off Topic => Topic started by: Numsgil on April 13, 2006, 11:13:18 PM
-
This is something of a long standing whim of mine. I'd like to try and write a science article about some research in Darwinbots and get it published in one of the Alife journals.
First step is, of course, deciding on something to write about. While I could write about the simulation itself I think it would be more worthwhile to do some experimenting and present the results from within the framework of Darwinbots.
So what do people think would be a good relatively long term experiment?
-
It probably needs to be something related to evolutionary processes.
-
Nums, this is something I would love to be a part of. I am actually considerring going back to school and finishing my doctorate in Alife. Yea, I have a long way to go.... Haven't made that decision yet, but it's on the table.
Here's some thinking off the top:
When I think about the unique attributes of DB relative to Avida or Tierra, one thing that comes to mind is competition. Most of the ALife papers I've read (I actually have the proceedings from Artifical Life IX on my desk at the moment) and most of the research done in these other systems utilize an external fitness routine, selecting for sorting ability or some such. I think there's a general paper just on the topic of using competition as a fitness criteria which could discuss limiting energy in the sim, mutation rates, etc.
Another half formed idea is the seperation of genotype and phenotype. Other systems often have organisms interacting directly on each others DNA, kind of Core Wars on steriods. DB has a pretty sophisticed phenotype encironemnt where most interaction is between phenotypes and the environment they inhabit via ties, shots and other artifacts withdirect interaction with the genome playing a background role. This strikes me as paper worthy.
And lest I forget to sneak my broken record pitch in, I think that if we added in-genome locale-relative mutation probabilites, then there is a paper on Genotypic Plastisity (as opposed to phenotypic Plastisity) - the importance of this mechanism for increasing the probability of varation in places in the genome where doing so is advantageous and vice versa...
-E
-
I like those ideas, they sound like good topics. I'm all for a colaboration. Eventually I'd like to get a masters and doctorate (I'm in my last semester of B.S. at the moment) in either ALife or simulations programming.
As to locale-relative mutations, I was thinking we could tack those onto codules, since codules provide a very useful way to segment the code.
I like the first idea because alot of work has already been done on this in the forum, and it's relatively easy to understand so we could make it a more community project.
-
Another idea: Species Recognition and the Evolution of Canablism in Asexually Reproducing Organisms
Locale-relative mutations in codules - I like it, I assumed we would do that. I do think we also need bp-level as well though, even if it's only a small set of descrete settings which impact point mutation probability at that loci on a logrthymic scale.... I may try to prototype this in 2.4...
-
I was thinking Canabilism too. It's one of the few traits that has existed in all incarnations of Darwinbots. Big Berthas (non reproducing bots that just get bigger and bigger) and other traits seem to have died off as the program became more complex.
-
What would be really cool is if we could show the evolution of a more sophisitcated meachanism for conspec recognition from a less sophsiticated one because of the advantage of doing so. Or perhaps the evolution of a different one - using myshoot instead of myeye.... so you could prey on your cousins without them fighting back...
-
For conspec recognition to not only not break but get better I imagine you'd need a way to provide periodic outside predation.
-
Good idea, but this is a lot of work. I'd start by reading some current literature and seeing what topics are hot and what kinds of questions are being asked. Then you'd be in the position to formulate a specific question that can be addressed in DBs.
This seems to be a good paper by the way: http://www.ncbi.nlm.nih.gov/entrez/query.f...t_uids=16393455 (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=16393455)
Too bad I don't have free access to this particular journal. You may send an e-mail to the author and ask for a pdf copy. It used to be a common practice with paper reprints, I'm not sure if that's allowed for pdfs.
This paper looks like a similr thing to what we would want to do, describe the program and give some examples: http://www.ncbi.nlm.nih.gov/entrez/query.f...t_uids=16386355 (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=16386355)
I don't have access to it either
-
I might have access at school. I'll see if I can pull them up.
-
Sounds like a good idea.
-
School doesn't give me access to it. Shame too.
I'd really like access to all their old articles. The only thing in academia I hate is when they make you give money to read their articles
-
Yeah, that sucks, especially considering that most of research is done on public money. It is not the scientists who are doing it though, it is the journal-publishing companies, Elsevier being the worst of them all.
Support PLoS!!!
-
http://www.cse.msu.edu/~ofria/home/pubs/pa...KinCheaters.pdf (http://www.cse.msu.edu/~ofria/home/pubs/papers/KinCheaters.pdf).
Any article of cannis could use that as a source.
http://www.cse.msu.edu/~ofria/home/pubs/pa...Intro-ALife.pdf (http://www.cse.msu.edu/~ofria/home/pubs/papers/AvidaIntro-ALife.pdf) The Avida paper.
-
http://www.cse.msu.edu/~ofria/home/pubs/papers/ (http://www.cse.msu.edu/~ofria/home/pubs/papers/) In fact this whole dirctory rocks my socks