Author Topic: Mutation Rates  (Read 3237 times)

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Mutation Rates
« on: July 22, 2005, 04:56:31 PM »
As I started work on the new mutation code, I ran across an interesting problem.

Which mutations are per base pair and which aren't.  That is, what mutations are independant of genome length.  Assuming the presence of Junk DNA in genomes really doesn't effect the final organism directly, we could assume that it's a protection mechanism for DNA that actively codes.  That is, a hide a seek game for mutations.

Such a strategy is only effective if at least some mutations aren't per base pair, but are per organism.

Basically I'm wondering if point mutations, that occur at any time in an organism's life, aren't independant of genome length.

Long story short: should the new point mutations that occur at any time in a bot's life be independant of genome length (that is, should the odds be expressed as 1 in X per base pair per cycle, or 1 in X per cycle?)

Offline Botsareus

  • Society makes it all backwards - there is a good reason for that
  • Bot God
  • *****
  • Posts: 4483
    • View Profile
Mutation Rates
« Reply #1 on: July 23, 2005, 02:18:10 PM »
Well, If we remove the genome length checking on mutation "N" as an example Then the robot with long dna is more resestent to mutation "N". Therefor, if you need some kind of mutation "N" to follow this rule, then go ahead make it.

My Qustion is: What will be the advantage of having this particuler mutations happen less in robots with longer dna?

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Mutation Rates
« Reply #2 on: July 23, 2005, 03:07:25 PM »
It's really just a think out loud excersise.  Shvarz seems to pick mutation rates such that he can control the mutation rates per organism instead of per base pair, so this would be a way of doing that.

I don't suppose it inherantly is better or worse, just different.

Offline shvarz

  • Moderator
  • Bot God
  • *****
  • Posts: 1341
    • View Profile
Mutation Rates
« Reply #3 on: July 26, 2005, 02:13:23 AM »
Sorry for late reply, was on vacation.

I can't really think of any type of mutation that would be "per organism".  They all seem to work on "per base" basis.

But even though in theory simulating "per base" mutation rate seems to be more accurate, I'd rather prefer a "per reproduction" frequency.  Here is why: we know now that bots accumulate DNA very quickly.  So a mutation rate that was set at N/command in the beginning of the sim gives 1 in X mutant offsprings, but after some time X becomes smaller and smaller.   I've seen this happen in the "First Bots" sim - almost every offspring has a unique genome.  I'd rather have a mutation rate that stays more or less constant regrdless of the amount of DNA.  This way it would be much easier for me.
"Never underestimate the power of stupid things in big numbers" - Serious Sam