Author Topic: Started a new evo sim today  (Read 3267 times)

Offline Testlund

  • Bot God
  • *****
  • Posts: 1574
    • View Profile
Started a new evo sim today
« on: June 15, 2006, 11:19:08 AM »
I started a new ex nihlo evo sim today, the same as I've done earlier in this topic, but now with the new 2.42.6 version. I'm using the same species as before:

Veggie:

cond
.nrg
0
0
start
0
0
0
0
0
0
stop
end

Bot:

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

The problem is that it doesn't seem to work to continue a sim from an older version of DB to a new version. It screws up somehow. I had this great sim from version 2.42.3, I think, that I couldn't continue to use in later versions. The mutations stopped working. I also tried taking a bot that developed in a previous version and use it in a later version but I couldn't get it to survive. So what I wonder is if it will be like this again when the next version comes out. It feels discouraging that I have to start a new sim for every new version. Maybe it's because there have been so many fixes in the latest versions so the bots doesn't fit in anymore. Anyway, I'm going to give this a try.
The internet is corrupt and controlled by criminally minded people.

Offline EricL

  • Administrator
  • Bot God
  • *****
  • Posts: 2266
    • View Profile
Started a new evo sim today
« Reply #1 on: June 15, 2006, 12:16:43 PM »
I've said this many times before.  Obviously I have not explained this well enough.  Let me try again.

In *ALL* versions prior to 2.42.4 (including 2.36.7 I'm prettty sure) there is a serious bug with saved sim and dbo bot files which only manifests itself in long run evo sims.  If a sim had run long enough for any bot to accumulate more than 2^15 bytes worth of mutation details (the string descriptions of what mutations have occurred in it's ancestry, this happens around 150 mutations) then any saved sim or bot will be corrupted.  This bug had been in the code since long before I had ever heard of DB.

Any bot saved in binary format (as a dbo) would also be corrupted as saving a bot in this fashion (as opposed to just saving the DNA, which is fine) includes the mutation history and thus the corruption in pre 2.42.4 versions).

This is why nobody ran long run evo sims prior to 2.42.4.  (Okay, this is not strictly true - you can run long runs evo sims prior to 2.42.4, but any saves get corrupted and fail to load.)

In 2.42.4, I fixed this problem for all new sims.  Any newly created sim in 2.42.4 or beyond with newly created bots will save and load just fine with essentiall infinite limits on mutation details.

For already corrupted sims and bots, I did my best to load the sim or bot, working around the mutation details corruption.  But a side effect of working around the corruption in 2.42.4 is to mess up the mutation settings for every bot which had exceeded the 2^15 byte limit.  So if you load a pre 2.42.4 bot in a 2.42.5 sim and then save such a sim or bot, the messed up mutation settings get carried forward in the new saved sims.

In 2.42.6, I did more work here (just for you Testlund - well, okay, not really) so that I now default the mutation settings if a pre-2.42.4 sim or bot is loaded.  So now, even loading pre 2.42.4 sims should work even better than in 2.42.4 or 2.42.5.

So, let me re-interate:  this issue was a one time thing.  If you have a corrupted sim or binary bot that ever lived in a pre 2.42.4 version, things should load just fine in 2.42.6, but the mutation setttings will be set to the defaults.  (If you don;t trust me and want to be totally safe, save any bots in old sims as textual DNA and load them into a new sim created post 2.42.4).

That said, as a rule, saved sims and bots from one version should always work fine in the next.  Any failure for them to do so is a bug.
« Last Edit: June 15, 2006, 12:18:20 PM by EricL »
Many beers....

Offline Testlund

  • Bot God
  • *****
  • Posts: 1574
    • View Profile
Started a new evo sim today
« Reply #2 on: June 15, 2006, 02:36:52 PM »
Ok, that sounds comforting.    I thought it whould be worth starting a new sim with the 2.42.6 version, seeing as it is much improved over the others. In earlier versions I wasn't sure if the bots evolved the way they should.

There can't be much more to fix here, can it? Unless a bunch of new bugs show up.
The internet is corrupt and controlled by criminally minded people.

Offline Testlund

  • Bot God
  • *****
  • Posts: 1574
    • View Profile
Started a new evo sim today
« Reply #3 on: June 16, 2006, 11:33:56 PM »
Check this out. I have two veggies here not related to each other that have developed quite similar DNA. They started out with the simple genes mensioned above (mostly 0's). One is green and has developed these genes out of 23 point mutations:

 0 ~ ~ 0 angle angle 1 & *.up 0 *.up add rnd *-11 1 | <
 ~=
 18 0 -14 <
 !=
 angle -6 *.up inc
 store
 store
 20 <
 7 and
 store
 angle 0 <
 *.shoot floor *.mass 3 -1 sub *.up 15 xor
 4 ++ 1 0 store
 0 7 -2 *-24 289 0 and
 dec
 *.up add .up dec
 xor
 -4 sgn ceil

The other is dark purple and has developed these genes out of 31 point mutations:

0 add 17 - ++ 14 1 and
 0 1 1 sqr else
 and
 %=
 0 0 0 0 0 =
 >
 start
 -3 start
 1 and
 14 0 angle store
 ^ <
 -8 -5 <
 0 ~ ~ 0 angle angle 1 & *.up 0 *.up add rnd *-11 1 | <
 ~=
 18 0 -14 <
 !=
 angle -6 *.up inc
 store
 store
 20 <
 7 and
 store
 angle 0 <
 *.shoot floor *.mass 3 -1 sub *.up 15 xor
 4 ++ 1 0 store
 0 7 -2 *-24 289 0 and
 dec
 *.up add .up dec
 xor
 -4 sgn

I wonder how so much DNA could have entered the cells with no 'insert' mutation, only point mutations wich is only supposed to change, not add anything. Also it's odd that the genes have become so similar.  
The internet is corrupt and controlled by criminally minded people.

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Started a new evo sim today
« Reply #4 on: June 16, 2006, 11:40:01 PM »
Looks like a virus.  The odd thing is that it doesn't seem to form a gene, which viruses tend to do (read: always do unless maybe you have no genes at all).  More disturbing is the lack of "start", which would be (should be) necessary for the DNA to actually do anything.

Offline Testlund

  • Bot God
  • *****
  • Posts: 1574
    • View Profile
Started a new evo sim today
« Reply #5 on: June 22, 2006, 06:01:26 PM »
I've started a new evosim tonight, this time with two species with only one zero in the genes, one species with autotroph. I have set point mutation to 100000 and insertion to 100. I was thinking that insertion means that a random value or command is added at a random location in the genes, wich whould build up a longer sequence after awhile. Is that what insertion does?
The internet is corrupt and controlled by criminally minded people.

Offline Testlund

  • Bot God
  • *****
  • Posts: 1574
    • View Profile
Started a new evo sim today
« Reply #6 on: June 22, 2006, 06:39:09 PM »
Looks like I have missunderstood what insertion means. I tried a sim with 20 bots with insertion set to 1. After 50000 cycles nothing had been inserted. I'll start a new sim with my previous zerobots mensioned in the beginning of this topic instead.
The internet is corrupt and controlled by criminally minded people.

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Started a new evo sim today
« Reply #7 on: June 23, 2006, 12:23:04 AM »
No, you understand Insertion correctly.  Except that insertion can only occur during reproduction.  Most mutations are like that.  Delta and Point Mutations are the sole exceptions.