Author Topic: question about costs / evo  (Read 3800 times)

Offline Nitus

  • Bot Neophyte
  • *
  • Posts: 29
    • View Profile
question about costs / evo
« on: July 27, 2008, 07:12:21 PM »
I've been having some difficulties in my evo sims in recent versions, specifically the bots expend very little energy doing anything, and absorb energy from feeding at a furious rate.
 
This leads to a situation where there's no evolutionary pressure.
 
In old versions of darwinbots, you could control the costs of movement, of executing a condition, and etc - unfortunately, when I try to bring up the "set costs" menu, darwinbots crashes with a "you need msstdfmt.dll on this machine."
 
I tried downloading it and placing it in the darwinbots directory, and in windows/system32, to no effect.
 
Is there some way to manually configure the costs? Or rather, what lines in the settings file contain those variables?
 
I'd really like to be able to make the cost/consuption ratio more balanced - bots speeding around with little to no expenditure just isn't conducive to evo, and playing with friction and viscosity just isn't balancing it out enough.

Offline Nitus

  • Bot Neophyte
  • *
  • Posts: 29
    • View Profile
question about costs / evo
« Reply #1 on: July 27, 2008, 07:57:07 PM »
Ok, I registered msstdfmt.dll and made it to the menu, but the changes I make there don't seem to be taking.
 
As a test, I jacked up the cost for rotation and toggled "zero momentum" mode, and the bots aren't expending energy despite rotating and changing direction of rotation. The changes I make to the cost settings don't seem to have any effect on gameplay, for whatever reason. I'm using XP pro, if this is realted to the "regional" issue or whatever..

 
I do have a few questions about some of the controls. What is "number" and "*number" and "logic" and what are the differences between the various commands?
 
edit The regional settings aren't the problem, now I've read the "installation issues" page on the wiki. I've disabled "brownian motion. The bots just simply aren't expending energy, despite the cost settings.
« Last Edit: July 27, 2008, 08:27:19 PM by Nitus »

Offline Testlund

  • Bot God
  • *****
  • Posts: 1574
    • View Profile
question about costs / evo
« Reply #2 on: July 28, 2008, 01:08:10 PM »
If your bots are autotrophs, maybe you're giving them too much energy/cycle? What number do you have in the field for 'Veggie Energy' on the 'General' tab? I think the default value is 5000 which is ridiculous. Set it to 1 and select 'Kilobody point'. Set morphological costs on the costs tab to 1 on everything but body and dna upkeep as a starting point. Then you can tweak it later.
The internet is corrupt and controlled by criminally minded people.

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
question about costs / evo
« Reply #3 on: July 28, 2008, 01:09:34 PM »
What happens when you change something in that form, close it, close the options panel, and reopen it and that form again?  Are the numbers you entered still there?

Offline Nitus

  • Bot Neophyte
  • *
  • Posts: 29
    • View Profile
question about costs / evo
« Reply #4 on: July 28, 2008, 03:13:54 PM »
Quote
If your bots are autotrophs, maybe you're giving them too much energy/cycle?

I do have veggie energy/cycle set to 1, but this is happening wih animals; my veggies don't do much but get eaten.


Quote
What happens when you change something in that form, close it, close the options panel, and reopen it and that form again?  Are the numbers you entered still there?

Yes, they're still there. They just aren't having any effect.
 
I started with no default settings file - is there perhaps some value somehwere that acts as a multiplier to costs, that I might have left blank?
 
Edit Ah, heh, such as "Cost Multiplier". It was set to zero. ;p
 
Soo . . . what's "number" "*number" and "logic" - and what's the difference between basic, advanced, bitwise, and etc?
« Last Edit: July 28, 2008, 04:14:23 PM by Nitus »

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
question about costs / evo
« Reply #5 on: July 28, 2008, 04:38:59 PM »
They let you specify different costs for different sorts of DNA commands.  A *number is something in the DNA that looks like *.this  Basic commands are things like add, sub, mult and div.  Advanced commands are things like pyth, sqrt, etc.

The original thinking was to allow more complex commands, like pyth and sqrt, to cost more than more basic commands, to favor things like add that people are good at figuring out what it does.  Anymore I think the whole thing is overly complex, so it's safe to ignore it if you want to

Offline Nitus

  • Bot Neophyte
  • *
  • Posts: 29
    • View Profile
question about costs / evo
« Reply #6 on: July 29, 2008, 03:11:19 AM »
Quote from: Numsgil
They let you specify different costs for different sorts of DNA commands.  A *number is something in the DNA that looks like *.this  Basic commands are things like add, sub, mult and div.  Advanced commands are things like pyth, sqrt, etc.

The original thinking was to allow more complex commands, like pyth and sqrt, to cost more than more basic commands, to favor things like add that people are good at figuring out what it does.  Anymore I think the whole thing is overly complex, so it's safe to ignore it if you want to

Oh, ok. Well, I'm sure it'll someday be on the wiki.
 
I have to admit I'm impressed with the work that's been done since the last time I updated. I was skeptical at first, because in the earlier releases [at least as far as plebs or "end users" go] made me feel that some of the intent behind darwinbots had been lost in the eagerness to facilitate someone's bot-writing. But I can see already that I should have checked in earlier and updated, because it's clear now that much of what I thought was being lost was simply in utero, waiting to be implemented.
 
Since the last time I checked in, you guys have taken my doubts and turned them into admiration - kudos to whoever it is that works on this thing.
« Last Edit: July 29, 2008, 03:16:48 AM by Nitus »

Offline Testlund

  • Bot God
  • *****
  • Posts: 1574
    • View Profile
question about costs / evo
« Reply #7 on: July 29, 2008, 04:03:03 AM »
I think the consensus here is that command costs shouldn't be used because it selects against complexity. Maybe Eric will remove that part of the cost tab in the future. Some may think that ANY cost selects against complexity, but I think morphological costs are just natural. In the sim I'm running right now I started out by having CostX set to 0 until population increased it. Currently my sim is at 23M cycles and CostX at 0.11. All my heterotrophs have died and my autotrophs are doing nada except replicating. A very boring sim.  
The internet is corrupt and controlled by criminally minded people.