It is currently Thu May 28, 2020 2:24 pm

 All times are UTC - 8 hours [ DST ]

 Page 1 of 1 [ 10 posts ]
 Print view Previous topic | Next topic
Author Message
 Post subject: Komaster #1 Posted: Wed Jun 26, 2019 10:03 pm
 Tengen

Posts: 4935
Liked others: 0
Was liked: 678
Bill, maybe now I sometimes understand how to calculate a count of a komaster ko. But what does it mean?! Why does the count of the local ko position include points elsewhere or of the temperature? How do we use such a count for analysing the whole board? What is the move value and what does it mean? Next, dreaming of monsters.

Top

 Post subject: Re: Komaster #2 Posted: Thu Jun 27, 2019 3:20 am
 Honinbo

Posts: 9732
Liked others: 3102
Was liked: 3208
RobertJasiek wrote:
Bill, maybe now I sometimes understand how to calculate a count of a komaster ko. But what does it mean?! Why does the count of the local ko position include points elsewhere or of the temperature? How do we use such a count for analysing the whole board? What is the move value and what does it mean? Next, dreaming of monsters.

The basic idea of evaluating positions by play is that, if a player makes a play that gains a certain number of points and the opponent replies with a play that gains the same amount, the value is the same as that of the original position. The first player plays in the position to be evaluated, and the opponent may reply in the environment of other plays. The play in the environment is assumed to be the largest gote play available and how much it gains is called the temperature of the environment.

Now suppose that we wish to evaluate the play, {7|5}. Let the temperature be t. Then the result of play at temperature t when Black plays first is 7 - t and the result when White plays first is 5 + t. When 7 - t = 5 + t, t = 1, and the result, no matter who plays first, is 6. So when the temperature is 1, Black can guarantee a result of at least 6 by playing locally to a local score of 7 or by replying in the environment when White plays locally, and White can guarantee a result of at most 6 by playing locally to a local score of 5 or by replying in the environment when Black plays locally. That means that the value of the game under these conditions is 6. The mean value of the position is also 6.

OC, these conditions may not exist. Suppose, for instance, that the temperature is 2. Then if Black plays locally White can reply to a value of 5, and if White plays locally Black can reply to a value of 7. Neither player will wish to play locally. However, suppose that we wish to estimate the value of the whole board. We may take 6 as the mean value (i.e., count) of the local position. The mean value is 6 at all temperatures 1 or higher.

Now suppose that the temperature is ½. Then if Black plays locally White can reply to a value of 6½, and if White plays locally Black can reply to a value of 5½. Each player will prefer to play locally. Black to play can guarantee a result of at least the mean value of the position, and White to play can guarantee a result of at most the mean value of the position, .

The temperature is an abstraction by which we may calculate mean values of positions and how much a local play gains, on average.

OK. Now let's consider the ko, K = {6||K|0}. As I have said, to evaluate it we have to consider the results of exiting the ko. First, let White be komaster of this ko. Berlekamp's rule is that when the komaster plays in the ko (normally by taking it) she must continue locally on the next turn if the koloser makes a play in the environment. (I have amended Berlekamp's rule to conform with my redefinition of thermography in 1998.)

If Black plays first at temperature t the result will be 6 - t and if White plays first the result will be 0 + 2t = 2t. (White takes the ko, Black replies in the environment with a gain of t, then White wins the ko and Black replies in the environment again. When 6 - t = 2t then t = 2 and the value of the game is 4, as is the mean value of the ko. Similar to the case of the simple gote, when t > 2 neither player will wish to play the ko, and we may estimate the value of the ko position as 4. And when t < 2 each player will prefer to play in the ko; the result when Black plays first is 6 - t, which is greater than 4, and the result when White plays first is 2t, which is less than 4.

Next, let Black be the komaster of this ko. Just because she is komaster does not mean that she is required to take the ko back. At temperatures above 2 Black will indeed prefer not to take the ko back, but to let White win it. At temperature 2 Black is indifferent between taking the ko back or not. Where being komaster pays off is when the temperature is less than 2. Then when White takes the ko Black, instead of replying in the environment, plays a threat large enough that White must answer (called a primary threat in my earlier theory) and then takes the ko back. White must play in the environment and then Black wins the ko. The result is 6 - t, the same as when Black plays first. That means that the value of the game is 6 - t. Black can guarantee a result of at least 6 - t and White can guarantee a result of at most 6 - t. When Black is komaster the value of the game depends upon the temperature. The mean value of the ko is not relevant when t < 2, and we may regard the value of the ko as the value of the game. OC, the temperature is an explicit parameter.

_________________
At some point, doesn't thinking have to go on?

Everything with love. Stay safe.

Last edited by Bill Spight on Thu Jun 27, 2019 4:20 am, edited 1 time in total.
Top

 Post subject: Re: Komaster #3 Posted: Thu Jun 27, 2019 4:10 am
 Tengen

Posts: 4935
Liked others: 0
Was liked: 678
Your reply is helpful. Now, I know at least what to think about. I guess I need to reread it several times before repeating my questions more precisely;)

Top

 Post subject: Re: Komaster #4 Posted: Thu Jun 27, 2019 4:12 am
 Honinbo

Posts: 9732
Liked others: 3102
Was liked: 3208
RobertJasiek wrote:
Your reply is helpful. Now, I know at least what to think about. I guess I need to reread it several times before repeating my questions more precisely;)

I have edited my note for correctness and, I hope!, clarity.

_________________
At some point, doesn't thinking have to go on?

Everything with love. Stay safe.

Top

 Post subject: Re: Komaster #5 Posted: Thu Jun 27, 2019 5:19 am
 Tengen

Posts: 4935
Liked others: 0
Was liked: 678
"The mean value of the ko is not relevant when t < 2, and we may regard the value of the ko as the value of the game."

Which do you call a count - the mean value of the ko or the value (What value?) of the ko?

Top

 Post subject: Re: Komaster #6 Posted: Thu Jun 27, 2019 6:26 am
 Honinbo

Posts: 9732
Liked others: 3102
Was liked: 3208
RobertJasiek wrote:
"The mean value of the ko is not relevant when t < 2, and we may regard the value of the ko as the value of the game."

Which do you call a count - the mean value of the ko or the value (What value?) of the ko?

It is always correct to call the value of the game the count. For instance, in this example at the dame stage with Black komaster the count is 6. That is what the local score will be, no matter who plays first. Why say any different? But with White komaster, without knowing who plays first the local score could be 6 or 0; in that case using the mean value as the count is OK. And, OC, at or above temperature 2 the mean value is appropriate.

_________________
At some point, doesn't thinking have to go on?

Everything with love. Stay safe.

Top

 Post subject: Re: Komaster #7 Posted: Thu Jun 27, 2019 7:28 am
 Tengen

Posts: 4935
Liked others: 0
Was liked: 678
Say we have the count C := 6 - t = 2t of the ko, where t = 2. This means that the environment has the largest local endgames with the counts C0 = 2 and C1 = 2. However, if Black starts, C contains C0; if White starts, C contains C0 and C1. So if we make a positional judgement and judge C0 = 2 and C1 = 2 for the two largest local endgames in the environment, we CANNOT judge C for the ko as a summand of the whole board position. I dislike this.

Top

 Post subject: Re: Komaster #8 Posted: Thu Jun 27, 2019 8:58 am
 Honinbo

Posts: 9732
Liked others: 3102
Was liked: 3208
RobertJasiek wrote:
Say we have the count C := 6 - t = 2t of the ko, where t = 2. This means that the environment has the largest local endgames with the counts C0 = 2 and C1 = 2.

The counts are estimates of territory, not gains.

Quote:
However, if Black starts, C contains C0; if White starts, C contains C0 and C1. So if we make a positional judgement and judge C0 = 2 and C1 = 2 for the two largest local endgames in the environment, we CANNOT judge C for the ko as a summand of the whole board position. I dislike this.

I am not sure what we are talking about. At temperature 2 let's suppose, as an example, that we have the ko, K = {6||K|0}, S = {10|0||0|0}, G0 = {0|-4}, G1 = {0|-4}, and G2 = {0|-2}. The overall count is 4 - 2 - 2 - 1 = -1. If Black plays first she wins the ko for a final result of 0, or she takes G0, after which White takes the ko and then Black wins the ko fight, again for a final result of 0. If White plays first he takes the ko and loses the ko fight for a final result of -2. In each case the result is 1 pt. more or less than the overall count of -1, which is what we expect at temperature 2.

At temperature 1 let's halve the values for the G's. G0 = G1 = {0|-2}, G2 = {0|-1}. The overall count is 5 - 1 - 1 - ½ = 2½. If Black plays first the final result is 3. If White plays first the final result is 2. In each case the result is ½ pt. more or less than the overall count of 2½, which is what we expect at temperature 1. Giving the ko a count of 4 would give us an overall count of 1½. White to play would lose ½ pt. Tilt!

_________________
At some point, doesn't thinking have to go on?

Everything with love. Stay safe.

Top

 Post subject: Re: Komaster #9 Posted: Thu Jun 27, 2019 10:13 am
 Tengen

Posts: 4935
Liked others: 0
Was liked: 678
You mean the theory is better than I thought? I see...

 This post by RobertJasiek was liked by: Bill Spight
Top

 Post subject: Re: Komaster #10 Posted: Thu Jun 27, 2019 10:23 am
 Honinbo

Posts: 9732
Liked others: 3102
Was liked: 3208
RobertJasiek wrote:
You mean the theory is better than I thought? I see...

Yes, Professor Berlekamp was a pretty smart guy.

_________________
At some point, doesn't thinking have to go on?

Everything with love. Stay safe.

Top

 Display posts from previous: All posts1 day7 days2 weeks1 month3 months6 months1 year Sort by AuthorPost timeSubject AscendingDescending
 Page 1 of 1 [ 10 posts ]

 All times are UTC - 8 hours [ DST ]

#### Who is online

Users browsing this forum: No registered users and 1 guest

 You cannot post new topics in this forumYou cannot reply to topics in this forumYou cannot edit your posts in this forumYou cannot delete your posts in this forumYou cannot post attachments in this forum

Search for:
 Jump to:  Select a forum ------------------ Life In 19x19.com General Topics    Introductions and Guidelines    Off Topic    Announcements    General Go Chat    Beginners    Amateurs    Professionals       Lee Sedol vs Gu Li    Go Rules    Forum/Site Suggestions and Bugs    Creative writing    Tournaments       Ride share to tournaments Improve Your Game    Game Analysis    Study Group    Teachers/Club Leaders       Teacher advertisements    Study Journals L19²GO (Malkovich)    1-on-1 Malkovich games    Big Brother Malkovich games    Rengo Games    Other versions of turn-based games Go Gear    Go Books    Go Book Reviews    Computer Go    Gobans and other equipment    Trading Post    New Products/Upgrades/Sales Go Club Forums    Go Club Discussions       Honinbo Go League    American Go Association Forum       Go Congress 2011 volunteers       AGA volunteers ( non-congress)    Australian Go Association    European Go Federation Forum    Singapore Weiqi Association    KGS    ASR League    IGS    OGS    Tygem    WBaduk    Turn Based Servers    Insei League Events    Kaya.gs       King of the Hill