White Moon (nano-hill)

On 23 Nov, 14:27, Fluffy <flu... RemoveThis @65536.org> wrote:
> Hi,
>
> White Moon is a nano-warrior which already survived more than 1600
> successful submissions to nano@SAL.
>
> ;redcode-nano verbose
> ;date Mon Jan 15 06:29:25 UTC 2007
> ;name White Moon
> ;author Fluffy
> ;strategy Watch the Moon!
> ;assert CORESIZE == 80
>
>         ORG start
>
> start   mov.i   < 44, $ 70
>         spl.x   # 40, } 66
>         mov.i   @ 21, < 79
>         mov.i   > 47, { 78
>         djn.i   $ 78, @ 79
>
>         END
>
> - Fluffy

Fluffy,

is this an evolved warrior?
If yes, the following idea could be real: evolved warriors are
generally stronger than hand-written ones.
Has anybody had the same sensation?

In the spirit of publishing new warriors, here is my new paper/clear.
I've experimented with different kinds of paper/clears, but this one
is the most effective (I've found).

;redcode-nano
;name Pacler Deux
;author Roy van Rijn
;assert 1

step equ 58

spl # 0 ,> 37
mov.i } -1 , }1
spl < step ,> 33
mov.i > 26 , < 74
djn.f -1 , < 45

end

And here is its predecessor with the 'new' type of paper (little twist
in copying, not more effective...):

;redcode-nano
;name pclr*pclr
;author Roy van Rijn
;strategy New type of paper/clear
;assert 1

step equ 19

spl #0 , }step
mov.i }-1 ,>-1
spl <-2 ,>-2
mov.i {54 , { 51
djn.f -1 , < 17

end

They are both optimized against the infinano, so I expect them to do
better on the infonano then on the current nano hill....

Roy

On Nov 25, 8:27 am, Cristian210273 <cristian210... RemoveThis @yahoo.it> wrote:
Hi,

> is this an evolved warrior?

yes, you can say so, but I have used something like a directed search
to get this warrior. It took about a day to create it.

> If yes, the following idea could be real: evolved warriors are
> generally stronger than hand-written ones.
> Has anybody had the same sensation?

For nano this might be true at the moment, but from time to time
there are hand-written warriors which rule the hill for a while.

For all larger hills it is really difficult to tell because of huge
computing power that you need to create a good warrior.

- Fluffy

>
> yes, you can say so, but I have used something like a directed search
> to get this warrior. It took about a day to create it.
>
What do you mean by "direct search"? Perhaps an exhaustive search
in the solution space (i.e. the generation of all the possible
permutations)?

> > If yes, the following idea could be real: evolved warriors are
> > generally stronger than hand-written ones.
> > Has anybody had the same sensation?
>
> For nano this might be true at the moment, but from time to time
> there are hand-written warriors which rule the hill for a while.
>
Yes, but what I noticed is that evolved warrior has a life longer than
hand-written ones (you have to think to the uGP evolved warriors...).

Bye!

On Nov 23 2008, 7:27 am, Fluffy <flu... RemoveThis @65536.org> wrote:
> start   mov.i   < 44, $ 70
>         spl.x   # 40, } 66
>         mov.i   @ 21, < 79
>         mov.i   > 47, { 78
>         djn.i   $ 78, @ 79

Seeing this makes me think a good protective measure would be to put a
DAT -5,-5 after your code. You don't want DAT 0,0 because those
decrement attacks can use it as a pointer to you. Avoiding low-number
operands in your working code might also be useful.

P. Kline