Author Topic: 2.36.2 mutations , good stuff first  (Read 7365 times)

Offline Botsareus

  • Society makes it all backwards - there is a good reason for that
  • Bot God
  • *****
  • Posts: 4483
    • View Profile
2.36.2 mutations , good stuff first
« on: April 07, 2005, 12:29:33 PM »
[you]Good Stuff:[/you]
the new sliders do get saved correctly in the settings
the new sliders look cool
random memory locations do mutate in the "actions part of the gene"

Finaly, the top and the buttom portions of the dna condition mutate, I think I even seen robots generate there own random memory adress for the conditions part of the gene.

GOOD JOB PY!



[you]Bad Stuff:[/you]

the new sliders do not mutate

'Insert a new value' resets to "0" when you load up different settings or when the robot mutates its rates.

Robot does not save its mutation rates just like hash when you save the robots dna into a text file.
Please make this seporate from hash because I usualy remove the hash to run robots of the same kind agenst each other.

[you]Not sure about , doing the long-term checks now:[/you]

*30 'self generated double memory cond
*20
>

,

*20
store ' store based  on memory pointer
*20 inc 'memory of pointer inc
*20 dec
etc.

And I found a major bug doing long term simulations (thank god it has nothing to do with mutations)
« Last Edit: April 07, 2005, 12:47:56 PM by Botsareus »

Offline PurpleYouko

  • Bot God
  • *****
  • Posts: 2556
    • View Profile
2.36.2 mutations , good stuff first
« Reply #1 on: April 07, 2005, 01:15:32 PM »
Quote
the new sliders do not mutate

No they don't. The reason is that the sliders are at present universal controls. They work equally for all robots kind of like a physics control fro the DNA.

My long term aim is to store them in the mutarray for each bot but that is going to be a lot more work since the present array is now full and it is a pain in the ass to make it bigger. I will get around to it eventually though.

Quote
'Insert a new value' resets to "0" when you load up different settings or when the robot mutates its rates.
I can understand why it would reset to zero when loading a settings file. That is because the mutarray address used already exists in the old save files but was never set to non-zero in old settings files. You should be able to save a settings file then reload it and get the values back. Let me know if this doesn't work.
Shouldn't be effected by mutating rates. :unsure:
I will check into that.

Quote
Robot does not save its mutation rates just like hash when you save the robots dna into a text file.
Please make this seporate from hash because I usualy remove the hash to run robots of the same kind agenst each other.
Haven't started on that yet. It won't be stored in the hash. It will be in a completely separate section of the DNA file and will be user definable from a text editor just like DNA is now. If the values are there then they will be loaded into the mutations settings else the default values (5000) will be set.

Quote
*30 'self generated double memory cond
*20
>

,

*20
store ' store based on memory pointer
*20 inc 'memory of pointer inc
*20 dec
etc.
All these are possible. I have seen them all happen during testing.

Quote
And I found a major bug doing long term simulations (thank god it has nothing to do with mutations)

At least my new stuff works then  :D

Have you reported the bug yet?
There are 10 kinds of people in the world
Those who understand binary.
and those who don't

:D PY :D

Offline Light

  • Bot Destroyer
  • ***
  • Posts: 245
    • View Profile
2.36.2 mutations , good stuff first
« Reply #2 on: April 07, 2005, 02:12:56 PM »
quick point when you make 32000 shell in one go, it doesn't cost you anything

Offline Botsareus

  • Society makes it all backwards - there is a good reason for that
  • Bot God
  • *****
  • Posts: 4483
    • View Profile
2.36.2 mutations , good stuff first
« Reply #3 on: April 07, 2005, 03:31:04 PM »
quicker point , who gives about shells in here , we are talking about mutations , start a new tread or this never gets done.

Offline PurpleYouko

  • Bot God
  • *****
  • Posts: 2556
    • View Profile
2.36.2 mutations , good stuff first
« Reply #4 on: April 07, 2005, 04:00:27 PM »
Do try to keep on topic Light, there's a good chap!
There are 10 kinds of people in the world
Those who understand binary.
and those who don't

:D PY :D

Offline Botsareus

  • Society makes it all backwards - there is a good reason for that
  • Bot God
  • *****
  • Posts: 4483
    • View Profile
2.36.2 mutations , good stuff first
« Reply #5 on: April 07, 2005, 04:09:46 PM »
OK PY , save a settings file , load a settings file , o look: Insert a new value is 0.

Its true PY , the mutation rates have the same effect. They set the value to 0.

Offline PurpleYouko

  • Bot God
  • *****
  • Posts: 2556
    • View Profile
2.36.2 mutations , good stuff first
« Reply #6 on: April 07, 2005, 04:39:38 PM »
Yes I just saw that myself. The strange thing is that this was a bug that I noticed and fixed earlier before I sent the code to Nums. I wonder if he forgot to add one of the modules.
There are 10 kinds of people in the world
Those who understand binary.
and those who don't

:D PY :D

Offline Botsareus

  • Society makes it all backwards - there is a good reason for that
  • Bot God
  • *****
  • Posts: 4483
    • View Profile
2.36.2 mutations , good stuff first
« Reply #7 on: April 07, 2005, 05:07:16 PM »
ok I just seen a robot with the top and the button in cond bouth memorys:

*57
*90
>

good job.


When you finaly fix the 'insert a new value' thing make sure it can insert *20 *62 stuff as well

*20
store ' store based on memory pointer
*20 inc 'memory of pointer inc
*20 dec
etc.

*20
30
add
*10
sub
42
add
etc.

Looks like 'insert a new value' is the only thing in there with the power to do that.


*****
Here is a new one I noticed resently:

There is now way to delete stuff like "add" "inc" "mult" without deleting the entire gene.

Offline shvarz

  • Bot God
  • *****
  • Posts: 1341
    • View Profile
2.36.2 mutations , good stuff first
« Reply #8 on: April 07, 2005, 06:05:22 PM »
Hmm, my previous sim finally crashed (after about 10 million cycles), so I decided to try  the version with PY's mutation code.  Almost exactly the same sim, but I am getting Overflow errors almost immediately.

Any ideas?
"Never underestimate the power of stupid things in big numbers" - Serious Sam

Offline shvarz

  • Bot God
  • *****
  • Posts: 1341
    • View Profile
2.36.2 mutations , good stuff first
« Reply #9 on: April 07, 2005, 06:10:18 PM »
OK, I figured it should have something to do with mutation rates.  So I checked mutation rates panel, and did not find anything suspicious.  Only one thing: my veggies supposed to have 0 for all mutation rates and they did.  But there was a space before 0.  When I click "Disable all mutations" I see the 0s jumping to the left of the field.

Could that be it?
"Never underestimate the power of stupid things in big numbers" - Serious Sam

Offline shvarz

  • Bot God
  • *****
  • Posts: 1341
    • View Profile
2.36.2 mutations , good stuff first
« Reply #10 on: April 07, 2005, 06:28:28 PM »
Yep, I think that was the problem.  When loading default settings the mutation rates for veggies had spaces before 0 and once I removed them, the simulation seems to run fine.
"Never underestimate the power of stupid things in big numbers" - Serious Sam

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
2.36.2 mutations , good stuff first
« Reply #11 on: April 07, 2005, 07:01:43 PM »
PY, I went through line by line and added your code for everything but the options panel.  I didn't want to replace old files because I had some minor changes I didn't want lost.

I forgot to mention that.  Some minor bugs probably are comming from there.  I'll look into shell and slime again too.

Offline shvarz

  • Bot God
  • *****
  • Posts: 1341
    • View Profile
2.36.2 mutations , good stuff first
« Reply #12 on: April 07, 2005, 07:08:49 PM »
I just noticed another thing that might explain some of my crashes, although this time it is my fault.  I copied DNA from a bot in a sim and pasted it into an empty text file.  What I forgot to do is to remove those numbers that program assigns to each gene.  As a result gene numbers are now duplicated in DNA.  Can that cause a problem?  It would not be an obvious one, cause my sims run fine, but maybe when a program tries to mutate this portions or something like that....
"Never underestimate the power of stupid things in big numbers" - Serious Sam

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
2.36.2 mutations , good stuff first
« Reply #13 on: April 07, 2005, 07:17:47 PM »
We should probably just add comment lines to the gene numbers when they're displayed so when you do that it can't cause problems.

Offline Botsareus

  • Society makes it all backwards - there is a good reason for that
  • Bot God
  • *****
  • Posts: 4483
    • View Profile
2.36.2 mutations , good stuff first
« Reply #14 on: April 07, 2005, 07:35:53 PM »
And dont forget that bouth reguler values have to be inserted and memory pointers , memory locations, etc. bouth before stuff like "store" and before stuff like "inc" "mult".
-done using 'Insert new value' , that currently does not work

And dont forget that there is NO WAY for the mutator to delete "inc" "mult" "div" etc. without deleting the whole gene....
-done using 'delete instruction' witch currently is not in the dna editor
« Last Edit: April 07, 2005, 07:39:48 PM by Botsareus »