Is it proved that 2 points is the correct komi for 6*6ï¼

plato · Post by **plato** » Sun Apr 19, 2015 7:09 am

Is it proved that 2 points is the correct komi for 6*6？

emeraldemon · Post by **emeraldemon** » Tue Apr 21, 2015 10:33 am

Seems like 6x6 is not solved, at least as of 2009: http://erikvanderwerf.tengen.nl/pubdown ... GA2009.pdf

The largest solved board in that paper is 5x6.

Krama · Post by **Krama** » Tue Apr 21, 2015 2:26 pm

If we analyze the data we come up to a conclusion that it would take between 10^28 to 10^34 seconds to solve the 9x9 board. The hardware used was Intel Core2 3GHz but even with the state of the art supercomputers that would perform million times better we are still talking about more seconds than have passed since the beginning of the universe.

However if you had a modern processor it would take around 15 years to solve 6x6, so they have to develop some kind of parallel processing or even this task seems like a distant future.

gowan · Post by **gowan** » Tue Apr 21, 2015 2:31 pm

It is not clear to me that knowing that, say, 6x6 go is solved determines the proper komi. As far as I know, komi is supposed to make the game even, i.e. that both players have the same chance of winning. If the game is solved there is no chance as far as winning is concerned. So, what would be the function of komi?

emeraldemon · Post by **emeraldemon** » Tue Apr 21, 2015 3:09 pm

gowan wrote:It is not clear to me that knowing that, say, 6x6 go is solved determines the proper komi. As far as I know, komi is supposed to make the game even, i.e. that both players have the same chance of winning. If the game is solved there is no chance as far as winning is concerned. So, what would be the function of komi?

Solving the board tells you the outcome if both sides play optimally. For example in the 5x6 case the computer found a result of B+4. So you could say the correct komi is 4 points, since that would create a jigo with perfect play from both sides. What the komi "should" be may be a whole different question, but I assumed this is what he meant by correct komi in this case.

YeGO · Post by **YeGO** » Tue Apr 21, 2015 6:57 pm

There is a related question and discussion in this thread:
http://www.lifein19x19.com/forum/viewto ... 18&t=11608

There does not appear to be a rigorous, computer-verified solution for 6x6.

While some have claimed results for larger boards (see http://www.mathpuzzle.com/go.html which claims that B+4 is the answer for 6x6), these might be dubious given the lack of details in rule set, methodology, and mathematical proofs. Note that Erik van der Werk's solution for 5x6 refutes the result given on that page.

Life In 19x19