RegisterRegister 
 
Archival threads
Goto page Previous  1, 2, 3, ... 9, 10, 11  Next
 
Post new topic   Reply to topic    Mirek's software Forum Index -> MCell Feature discussion

Rock Reply with quote
Guest





PostPosted: Mon Dec 27, 1999 5:06 am    Post subject: Chemistry and interactivity
 
2) CHEMISTRY, GENES AND MEMES --

Chemistry could stand to be introduced as a more variable set of cell states than just 2 state (on/off), or 4 state (2 + flip-/keep- present state), or the more common generations rule,
there are 10, 20 and 30 state cells in Turing machines, etc.; what if the cells went through the states as they progressed in their generations? This would undoubtedly lead to a longer lived pattern, generally, though not necessarily hardier.

The genes are easily alterable by most of the freeware currently extant, but only in as much as the first circumferance of neighboring cells are concerned (except when considering 1-d life, where some programs allow 5 neighbors or more). What if we allowed cells to be affected non-locally, by 2 unit distant cells or further?

For example, although the patterns and rules might not sound so scrambled, imagine the actual results of cells in a 23/3 set being affected by the 8 neighbors found by a chess
knight's moves? Or in a 16 neighbor rule by the outside cells in a 5x5 block?

Cell-count totalistic rules and configuration-dependancy are all good and well, but could there be something missing, a method of reproduction not being hit?

Then again there's the question of genetic fidelity. How accurate must the transfer of pattern be in order for life to remain viable? Certainly in the case of gliders it's necessary,
but in all the others, they either die out, go static, or oscillate in place (much like mortality, petrification, or isolated closed systems). Could mutation rates (ether a given
constant for the universe, or tailorable to any given species) breathe new life into an otherwise foredoomed situation?

Mutation rules could be arranged so that the whole universe has a set level of fidelity or for certain areas to constantly or variably be "irradiated" to a given degree, or even for mutation to occur only under predefined circumstances.

Perhaps there's a way to have evolution run its course without too much loss of valuable trial and error. Maybe two patterns could result when collisions occur: one ontogenetic
(duplicating) of the intended result and one phylognetic (close copy). Add a genetic memory to the cells and the succesful patterns will avoid the unsuccesful attempts, put the memory into the system itself and a genral bias results yielding an accelerating improvement in overall fitness of design.
Back to top

Rock Reply with quote
Guest





PostPosted: Mon Dec 27, 1999 5:10 am    Post subject: Memes and interactivity
 
Memes. Can life be said to have memes? Not really, no. But perhaps their activity could be altered when in the presence of a given species (or any other pattern, even another one of the same species) within a certain distance. This could define active mating of gliders, aggressive territoriality for safe collision-induced destruction (predation/defense) of interloping life patterns... The patterns might be able to have a mutation keyed into their cycles to allow for specific motility with reasoned path choosing for herding away (or the death of) the outsider, or for escaping the advance of another, even evading their detection through hiding, possibly even a simple version of camouflage. Then, of course, they might be keyed to react with different methods, such as firing off gliders, tentacular extensions, or cell walls, or being "poisonous" (by operating under optimally different rules when being attacked/collided with).

What if a gun spat out different offspring from the same baup the cells in accordance with the rules set for the pattern, rather than the increase normally used in life. This then means that one could set a given pattern of choice into the system rather than simply wait to see what develops from a given starting design. But then, once set into motion the pattern will most likely dissipate into thin air. This means that one must have a rule utility to allow the life form's future patterns to be explored under a varying rule set in order to assure that it will return to this or another pattern within a set range of generations.

That's where the complications then arise. It could be allowed to explore the possible sets at random, or it could be tailored with specific results in mind; if so, then the pattern becomes more than a mere set of cells and becomes a truly complicated proposition: under which rules shall it run for how many generations, how much larger does its routine (genome) become when coded, is there a more optimal set of rules/patterns, is it really worth it, etc.?


3) INTERACTIVITY --

How do the patterns interact if one is working in a non-preferancially ruled medium? Perhaps it would be best to have the two non-interactive? Maybe a third rule set could be used as default, or an average of the two? Perhaps the most interesting (and possibly most realistic) answer would be to have each pattern react to the other according to the rules it runs on, i.e.: a 23/3 and a 23/36 pattern run into each other. The 23/3 cells die off and reproduce in reaction to the 23/36 cells' presence in accordance with 23/3 rules and vice versa, the resulting cells of course being treated by the rules of the parent's set; when a cell is required to be on according to both sets (which could be represented by a color or set of colors not used for the existing rule sets) it could be treated as two superimposed cells- when one no longer needs it on it reverts to the other's rules only, then turns off when no longer needed by either rule set.

Co-evolving CA using non-uniform rules are apparently not an untried concept. The question arises: what of rules competing with each other from cell to cell, or remaining static/cyclical/random over time w.r.t. a given cell or set? These do have simulations. When cells' rules start to affect those of the cells next to them, or the meta-rules do this, where does it lead? How about when the rules are written to allow a cell state not just to react to the state of the
cells nearby, but to actually attempt to takeover the running of the other cells' states, not rewriting but circumventing their rules? But what of patterns affecting the rules? What would occur then, when a pattern may arise and be allowed to try its hand at rewriting the rules which govern it? What might be seen then?
Back to top

Rock Reply with quote
Guest





PostPosted: Mon Dec 27, 1999 5:12 am    Post subject: Fundamental rules and summary
 
4) FUNDAMENTAL RULES --

Are the rules themselves inviolate? Perhaps it's not so outlandish to consider a set of meta-rules regarding the evolution or cycle of rules through which the rules may run over the course of several generations or at random; it could be a more complex situation where a regression occurs of meta-meta rules as dictated by the user. Such a situation sounds as if it would lead only to chaos. Indeed it would if precautions weren't taken to either design the rules changes according to the specific state of affairs in the grid at any given generation, or to allow the patterns to react accordingly and metamorphose in to a more appropriate pattern in accordance with the new situation (could be deadly to any patterns caught with their pants down).

It could be arranged as a slow tweaking of the rules for optimum life density, or a particular end pattern, or even to be more aesthetically pleasing according to preset preferences. Something like a semi-guided search program for genetic engineering.

What if the rules were worded in such a way as to lack parity w.r.t. time reversal? Or if the same rules (etc.) produced different results, having only averagable not reproducable laws (quantum life?)?

On a lower order of programming, what would the results be if instead of updating the universe as a snapshot entirety, it were updated from a central location, a set of points such as the corners, from cells at random, from the edge inward... basically anything other than the accepted instantaneous flash of all-points evolution?


5) SUMMARY --

Many different modifications have been discussed here, among them some which have been incorporated into Mirek's life versions and other life programs. Some are too complicated or unrelated to the actual point of life to be of any value to serious users, only to those who use it as a game of diversion. This does not negate their value or interest, but does leave them little chance of seeing incorporation into any programs.

Perhaps the most realistic of these will be encoded, such as interacting life-rule lifeforms, or regions where different rules apply in the same life universe (this has been done at least once that I know of). Conceivably the idea of cells mutating the rules governing their own birth and death could be done. Would any of this advance the useful applications seen to develop from life? I don't know, but I think that many of these applications would be fascinating to both the serious student of life development and the "mere" tinkerer.
Back to top

Rock Reply with quote
Guest





PostPosted: Mon Dec 27, 1999 5:40 am    Post subject: RE: My favourite CA rule
 
Brians brain seems to be a pretty popular one, and it certainly gets my vote along with the Hex rules and weighted life. I know weighted life isn't exactly a set of rules so much as a way to create and manipulate rules, but I love it anyway, it allows so much flexibility.
Back to top

Kamil Przybyla Reply with quote
Guest





PostPosted: Tue Dec 28, 1999 7:18 am    Post subject: Sorry
 
I'd like to apologize everybody, I've been out of a scene for about a month. I'm returning to research CA. Now I'm downloading 2.51 of MCell.

Lionel - I'm curious how are you going to make fuzzy logic gates with CA (I mean StarWars/MirekGro). Also if somebody knows how to change cross puffer to emit non-runnered crosses please let me know. Also, what I'm researching is mainly Generations CA.
Back to top

mirwoj Reply with quote
Admin


Joined: 15 Aug 2002
Posts: 45

PostPosted: Tue Dec 28, 1999 8:36 am    Post subject: RE: Sorry
 
Hi Kamil.

Just curious, how is your StarWars Lexicon?
_________________
--
Mirek
Back to top
View user's profile Send private message Visit poster's website Yahoo Messenger

mirwoj Reply with quote
Admin


Joined: 15 Aug 2002
Posts: 45

PostPosted: Tue Dec 28, 1999 8:51 am    Post subject: RE: What would you like to see in MCell 3.0?
 
Hi Rock,

it was really looong Smile
I'll go slowly through it. From what I've already seen, some topics are already possible. Answers soon.

Thanks.
_________________
--
Mirek
Back to top
View user's profile Send private message Visit poster's website Yahoo Messenger

mirwoj Reply with quote
Admin


Joined: 15 Aug 2002
Posts: 45

PostPosted: Tue Dec 28, 1999 9:01 am    Post subject: RE: My favourite CA rule
 
> Brians brain seems to be a pretty popular one

And fascinating. I still remember those sleepless nights...

> (...) weighted life (...) but I love it anyway

I suggest you take also a look at Rules Tables. Weighted Life is a great generalization of Life, Vote, and Generations. But Rules Tables open quite new possibilities impossible to code in totalistic rules. Take the "Busy Brain" as an example. Just a small modification to Brian's Brain, and what a change. And Cars. Quite another way of designing CA.
_________________
--
Mirek
Back to top
View user's profile Send private message Visit poster's website Yahoo Messenger

mirwoj Reply with quote
Admin


Joined: 15 Aug 2002
Posts: 45

PostPosted: Wed Dec 29, 1999 3:22 am    Post subject: Oscillators in Brian's Brain
 
Hi,

I posted this already to Usenet, but I think it's worth reposting here. Michael Sweney constructed the first oscillator working in... Brian's Brain!

I played with BB for a long time and was nearly sure this
dynamic rule where no cells can survive can't make stable
patterns. I include the Michael's oscillator at the end
of the message.

I'm sure even this simple oscillator will allow interesting
constructions, maybe reflectors?


Here's the oscillator in semigraphics, where
'.' is 0, '*' is 1 and 'o' is 2.

Phase 1:
..o.
o**.
.**o
.o..

Phase 2:
.*..
.oo*
*oo.
..*.

Phase 3:
.o*.
*..o
o..*
.*o.

and the pattern in MCell format:

#MCell 2.50
#GAME Generations
#RULE /2/3
#BOARD 200x200
#D Brain's Brain rule
#D Oscillators
#D Michael Sweney December 1999
#L 4$5.B$3.BAA$4.AAB$4.B
_________________
--
Mirek
Back to top
View user's profile Send private message Visit poster's website Yahoo Messenger

Kamil Przybyla Reply with quote
Guest





PostPosted: Wed Dec 29, 1999 9:37 am    Post subject: RE: Oscillators in Brian's Brain
 
How about stabilizing this pattern?:

#MCell 2.0
#GAME Generations
#RULE /2/3
#BOARD 400x400
#WRAP 1
#L $4.ABA$4.B.B$3.AB..AB$8.A$7.B$7.AB$8.A$7.B$7.AB$8.A$7.B$7.AB$8.A$7.B$
#L 7.AB$8.A
Back to top

Rock Reply with quote
Guest





PostPosted: Wed Dec 29, 1999 10:36 am    Post subject: RE: My favourite CA rule
 
This is true, it's been a bit since the last time I toyed with MCell. Smile Excellent work, by the way. Of course, that's the driving force behind self organising chaotic systems. Even looking at Conway's original rules, just modifying a birth or death count slightly can result in vastly different behaviors.

In a way it rather reminds me of factoring. When one puts 10^4 to an exponent (say, put it to the 5th power), the result can be quite a bit larger if we start with (3*10^4)^5 than (10^4)^5. It's only changing the original number by a factor of 3, but the result is much greater a total difference, although the ratio would remain the same.

I've been fascinated by Life since I first heard about it in a book by Piers Anthony, but back then I had no computer and had to work it out by pen and paper. Even then it was difficult because the rules weren't spelled out. I had to work them out from the examples graphically demonstrated. I've still got the original notes. I remember finding a couple of bones that I meant to pick with the description of the rules and the rules as demonstrated in the sequences.... I wonder if I was right? Perhaps I'll dig the notes out and see what the problem was.
Back to top

Rock Reply with quote
Guest





PostPosted: Wed Dec 29, 1999 11:13 am    Post subject: Thanks. :-)
 
I appreciate your taking the time, thank you. With any luck it won't be too boring or descend into too ridiculous a flight of fancy (that might sound vaguely sarcastic, but I actually mean it in all seriousness).

Sorry about the length, it was an attempt at being thorough and a lot of the text is to clarify into English what might otherwise be either too abstract or confined by the lack of explanation resulting in gibberish and bits of meaning jumping out only to a mathematician, biologist, or whatever the particular line of study a given analogy might be in.
Back to top

Michael Sweney Reply with quote
Guest





PostPosted: Thu Dec 30, 1999 9:38 am    Post subject: RE: Oscillators in Brian's Brain
 
I have found two oscillators in Brian's Brain, the first is
fairly small, the second much larger. I am posting the pattern
file to Mirek.
Back to top

Jason Reply with quote
Guest





PostPosted: Wed Jan 05, 2000 9:24 am    Post subject: Brian's Brain etc
 
Brian's Brain seems to be quite popular. While toying around with other rules I found "Jason's Brain" 3/23/4 (S/B/States). This has many nice gliders etc and occasionally will create the pascal triangle/sierpinski triangle pattern that seems to be much more frequent in Brian's Brain.

Trying to find a 5 state Brain type rule is much more difficult. So far 35/234/5 is close, but the "diagonal lines" tend not to send out any gliders, so the system dies down to a stable condition fairly quickly.
Back to top

Jason Reply with quote
Guest





PostPosted: Wed Jan 05, 2000 9:32 am    Post subject: Question about 1D totalistic rules
 
With the 1D totalistic CA rules, the "C" parameter specifies the count of states. In previous CAs I have used 2 as a minimum for states (ie alive or dead). 3 states gives alive,dead and the "ghost" state. Higher than 3 gives multiple "ghost" states.

I noticed in the 1D totalistic rules you (Mirek) have, some have a value C0. How does this relate to the final patterns? For example I have been able to reproduce the "champagne" rule, but only with 2 states (dead/alive), and obviously this does not have any shaded cells within the "champagne glasses". So what does C0 (and C1) really mean?

Excellent CA site Mirek.
Back to top
Post new topic   Reply to topic    Mirek's software Forum Index -> MCell Feature discussion
Goto page Previous  1, 2, 3, ... 9, 10, 11  Next
Page 2 of 11

Choose Display Order
Display posts from previous:   
User Permissions
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum

 
Jump to:  


Skin Created by: Sigma12
Powered by phpBB © 2001, 2002 phpBB Group