Life In 19x19 http://prod.lifein19x19.com/ |
|
question about shapes and formations http://prod.lifein19x19.com/viewtopic.php?f=15&t=5246 |
Page 1 of 1 |
Author: | dohduhdah [ Sat Dec 31, 2011 9:56 pm ] |
Post subject: | question about shapes and formations |
Hi! First of all, happy new year! Would you say the three identical black shapes are equally spaced in each of the four quadrants? http://bayimg.com/eALjpaadK greetings and thanks in advance for any suggestions, Niek |
Author: | Dusk Eagle [ Sun Jan 01, 2012 1:39 am ] |
Post subject: | Re: question about shapes and formations |
I don't really get the question. Is it a go question, or more of a riddle? |
Author: | Loons [ Sun Jan 01, 2012 1:47 am ] |
Post subject: | Re: question about shapes and formations |
I would say no. The groups' connections differ: 'A' are connected by a large knight's move, 'b' a knight's move, 'c' a diagonal move. Also the thickness/overconcentration of the stones involved. Happy new year. |
Author: | EdLee [ Sun Jan 01, 2012 1:52 am ] |
Post subject: | |
He means this: |
Author: | ez4u [ Sun Jan 01, 2012 2:19 am ] |
Post subject: | Re: |
EdLee wrote: He means this: Ah, now I understand (which I did not before). Since the grid is helpful in dispelling any purely optical illusions, I guess the question comes down to whether we choose to interpret "spaced" in the original statement through any go-related filters or are content to base our answer on the purely physical arrangement. Since there is no indication one way or another in the original post, the question is purely about the person answering rather than the diagram. Anyway, I pass. (Remember Bill's dictum, "Tenuki is always an option!") ![]() |
Author: | dohduhdah [ Sun Jan 01, 2012 10:58 am ] |
Post subject: | Re: question about shapes and formations |
Dusk Eagle wrote: I don't really get the question. Is it a go question, or more of a riddle? Like EdLee indicates in the picture with marked stones, my question is more or less whether there is a way to abstract from the particular shapes of groups of connected stones when considering how they are spatially related on the go board. |
Author: | hyperpape [ Sun Jan 01, 2012 12:02 pm ] |
Post subject: | Re: question about shapes and formations |
dohduhdah wrote: Dusk Eagle wrote: I don't really get the question. Is it a go question, or more of a riddle? Like EdLee indicates in the picture with marked stones, my question is more or less whether there is a way to abstract from the particular shapes of groups of connected stones when considering how they are spatially related on the go board. If you let us know why you're asking that question and what your constraints are, we might be able to give advice on whether there's a way to do it that fits your interest. |
Author: | dohduhdah [ Sun Jan 01, 2012 12:47 pm ] |
Post subject: | Re: question about shapes and formations |
hyperpape wrote: dohduhdah wrote: Dusk Eagle wrote: I don't really get the question. Is it a go question, or more of a riddle? Like EdLee indicates in the picture with marked stones, my question is more or less whether there is a way to abstract from the particular shapes of groups of connected stones when considering how they are spatially related on the go board. If you let us know why you're asking that question and what your constraints are, we might be able to give advice on whether there's a way to do it that fits your interest. I'm thinking about a way to store patterns in go in an optimal fashion. I'd like to program a bot that can learn from experience by accumulating knowledge (in the form of patterns it learns to recognize) based on playing many games. You can store the shape of a group of connected stones in a way that is independent of rotations, translations and reflections. Likewise it seems it must be possible to store patterns in a formation of groups on the board. The idea of patterns might be a bit vague. But consider the concept of atari. In a way this concept can be understood as all possible situations (*) where there is a group on the board that has only a single liberty left. As a program continues to learn, at some point it should come up with such concepts by itself and learn to distinguish them. For some concepts like 'atari' it's relatively easy and concrete, but for other concepts, like 'influence' or 'tewari' it's more fuzzy and abstract. (*) Where a situation is understood as a configuration of stones on the board as well as a history of past states that have occurred in the particular game where this configuration occurs, which allows one to determine if a particular move is allowed or not (taking the superko rule into consideration). |
Author: | dohduhdah [ Sat Jan 07, 2012 8:55 pm ] |
Post subject: | Re: question about shapes and formations |
dohduhdah wrote: I'm thinking about a way to store patterns in go in an optimal fashion. I'd like to program a bot that can learn from experience by accumulating knowledge (in the form of patterns it learns to recognize) based on playing many games. You can store the shape of a group of connected stones in a way that is independent of rotations, translations and reflections. Likewise it seems it must be possible to store patterns in a formation of groups on the board. The idea of patterns might be a bit vague. But consider the concept of atari. In a way this concept can be understood as all possible situations (*) where there is a group on the board that has only a single liberty left. As a program continues to learn, at some point it should come up with such concepts by itself and learn to distinguish them. For some concepts like 'atari' it's relatively easy and concrete, but for other concepts, like 'influence' or 'tewari' it's more fuzzy and abstract. (*) Where a situation is understood as a configuration of stones on the board as well as a history of past states that have occurred in the particular game where this configuration occurs, which allows one to determine if a particular move is allowed or not (taking the superko rule into consideration). I guess the easiest way to associate a location with any particular shape is to simply take the center of its bounding box as its associated location. Which means there are 1369 potential locations for shapes on a 19x19 go board. http://bayimg.com/kALhbAAdm |
Page 1 of 1 | All times are UTC - 8 hours [ DST ] |
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ |