Code center > Suggestions

A few suggestions(Endy)

(1/4) > >>

Endy:
Here are just a few ideas I've had in the past weeks; not sure how they'll be received. :unsure:

The first idea had is that it would be possible to have functions, of a sort, in DB. The idea is that functions would be defined and entered into the dna at the start.

ie.
f1 0 floor 1 ceil mult dec
f2 dup div mult inc
cond
start
.fixpos *.fixed f2
stop

This would be how you could write it, but for the bots all they would see and be able to seperatly is the actual code.

cond
start
.fixpos dup div mult inc
stop

This way no mutations in the functions could occur, but the code they mean would be capable of mutating.(not the normal method, I know, but most programmers don't have to worry about mutating code :) )

Another idea was a set of negative sysvars equal to the positive ones(abs num before storing). This way the bots could still use a negative store to produce a behavior.

The last idea I had was a "flip" command that would flip the top two stacked numbers. The only problem is how to have it activated for a specific peice of dna, I was thinking of it only flipping based on a number before/after it. Should allow an effect similar to rnd for evobots, allowing them greater versatility from a single strand of dna.

Alright that's all for now, hope you like the ideas, comments are appreciated. :D

Endy B)

Botsareus:

--- Quote ---Another idea was a set of negative sysvars equal to the positive ones(abs num before storing). This way the bots could still use a negative store to produce a behavior.
--- End quote ---

Well I wrote a "Random Painter" program that uses modulation (modulation: a more complex method then simple abs()) for out of range numbers everywhere there is an error, the result was a generally chaotic generation of numbers. My point is the DNA code must be very specific about what it wants to do; the compiler should not "repair" the code for the robot. I think it’s a bad idea.




--- Quote ---The first idea had is that it would be possible to have functions
--- End quote ---

Good Idea , I think functions will result in better mutations , and will make it easyer for the user to program robots.



--- Quote ---This would be how you could write it, but for the bots all they would see and be able to seperatly is the actual code.
--- End quote ---

Bad Idea, functions are a good replacement for "cloning" specific math in the DNA. I believe that things like cloning is very important, currently we only have genes that can be "cloned". If we let the DNA define its own costume sysvars , functions , and finally make use of the new "and" "or" operators as it mutates we should defiantly get more effective results.

But generaly Endy , it sounds like the only thing you are missing is a good dna editor. I think Num is working on a DNA debugger it should be close.


--- Quote ---The last idea I had was a "flip" command that would flip the top two stacked numbers.
--- End quote ---
I like this, it will work for well stuff like "a / b" or "b / a"

shvarz:
Don't care about the first - can be done if enough people want it.

Second idea (functions) is good, as far as these functions don't make it into the actual DNA code - Nums had the same idea a couple of days (weeks?) back, but he wanted bots to be able to do that too.

Third idea (flip command) is nice.  Don't see what you would use it for, but it sounds like fun.

PurpleYouko:
I think the flip would be more use as a type of mutation. Maybe a bot could simply flip two values in  its DNA.

Numsgil:
Well I have a reverse mutation I'm working on.  If you set a length of two, you basically have a swap.

Navigation

[0] Message Index

[#] Next page