Volunteers needed for cross solving study

[Dene]

This is awesome, Lars! Way to go! The OPness of colour neutrality is finally coming to show!! I may as well have my little input here, in saying that I can almost always find a case where on edge is already oriented and another 2 are extremely easy, for a cross. This isn't certain, but I generally come to expect a good cross on at least one face. COLOUR NEUTRALITY FOR LIFE!!!

 

[Stefan]

But because you insist, Stefan, here it is and it wasn't difficult to find :
B2 D L2 R2 U R2 D' L2 R' F' D' L R2 B2 F' D2 L' F2 D U2

Darn. I was of course hoping there isn't any. After all, there are just 102 cases for eight moves white cross, and it had to be one putting the white edges in F2L only. Figured there's a good chance this is impossible. Darn.

 

[AvGalen]

It looks like it will be finished within the next 2 days. What will be next?

I think all big conclusions can already be drawn from the current results. The rest of the crunching will just provide us with meaninglessly exact numbers and hopefully some fun scrambles.

 

[jazzthief81]

But because you insist, Stefan, here it is and it wasn't difficult to find :
B2 D L2 R2 U R2 D' L2 R' F' D' L R2 B2 F' D2 L' F2 D U2

Darn. I was of course hoping there isn't any. After all, there are just 102 cases for eight moves white cross, and it had to be one putting the white edges in F2L only. Figured there's a good chance this is impossible. Darn.

Well I never really looked into the 102 worst cases a lot. These are the patterns that can occur for cross on bottom:

4 edges in D slice: 10 cases
3 edges in D slice, 1 edge in E slice: 8 cases
1 edge in D slice, 3 edges in U slice: 20 cases
1 edge in E slice, 3 edges in U slice: 16 cases
1 edge in D slice, 1 edge in E slice, 2 edges in U slice: 8 cases
4 edges in U slice: 40 cases

 

[Bounb]

Should be done in one or two days we're at 60% now. Looking forward to the analysis.

 

[fw]

I think this whole thing is great. We should do this more often. I for myself have some CPU power, but almost never use it. And I think there are some other interesting cube-related problems which are unsolved just because it takes to long to calcuate the solution...

 

[cuBerBruce]

I think this whole thing is great. We should do this more often. I for myself have some CPU power, but almost never use it. And I think there are some other interesting cube-related problems which are unsolved just because it takes to long to calcuate the solution...

I've done several large analyses related to 3x3x3 and 4x4x4. A lot of these involved the creation and reading of large data files, which would not make them be very well suited for WAN execution. I assume this one only sends distribution results back to the host (a small amount of data), so it's works well for this type of distributed computing.

I was wondering if Lars used some type of existing software package or tool to develop this client & server.

(By the way, I noticed I hit the jackpot the other night - orientation #777.)

 

[jazzthief81]

I've done several large analyses related to 3x3x3 and 4x4x4. A lot of these involved the creation and reading of large data files, which would not make them be very well suited for WAN execution. I assume this one only sends distribution results back to the host (a small amount of data), so it's works well for this type of distributed computing.

I was wondering if Lars used some type of existing software package or tool to develop this client & server.

(By the way, I noticed I hit the jackpot the other night - orientation #777.)

Yeah, the amount of data sent back and forth is almost nothing. Retrieving a package is picking a number between 0 and 2047 and the result is the distribution of all the positions in that package (9 natural numbers).

I didn't use an existing library, just the Java API. It's all you need really. The server is a simple Java servlet that you can access through a URL.

 

[jazzthief81]

It looks like it will be finished within the next 2 days. What will be next?

Next will be the same analysis in QTM, and then same thing again for opposite color cross solving.

After that we can debate about what other things would be interesting to investigate.

I think all big conclusions can already be drawn from the current results. The rest of the crunching will just provide us with meaninglessly exact numbers and hopefully some fun scrambles.

Well, we could easily have made yet another good estimate, but now we will finally have the exact figures.

 

[AvGalen]

Well, we could easily have made yet another good estimate, but now we will finally have the exact figures.

If this was just about getting the exact figures I wouldn't have bothered. For me pi is simply 3.14, infinity is 999999999999999999999999 and 1/3 is 33%. With this logic I can be 100% sure that my logic is almost certainly wrong and/or correct.

For me this project is all about finding "weird" cases.

(p.s. I used to study applied physics)

 

[jazzthief81]

With this logic I can be 100% sure that my logic is almost certainly wrong and/or correct.

I'm not even going to ask you to elaborate on this...

 

[fw]

infinity is 999999999999999999999999
(p.s. I used to study applied physics)

Is that how all physicists handle math? If yes, that confirms some stereotypes I have heard of

Just kidding...

 

[Lucas Garron]

If this was just about getting the exact figures I wouldn't have bothered. For me pi is simply 3.14, infinity is 999999999999999999999999 and 1/3 is 33%.

Incredible! You can solve some twistypuzzles from any of more than an infinite number states?

 

[Johannes91]

For me pi is simply 3.14, infinity is 999999999999999999999999 and 1/3 is 33%.

Incredible! You can solve some twistypuzzles from any of more than an infinite number states?

Nah.

3x3x3 = 27
5x5x5 = 125
Megaminx is like a 3x3x3

They are all far from infinity!

 

[AvGalen]

For me pi is simply 3.14, infinity is 999999999999999999999999 and 1/3 is 33%.

Incredible! You can solve some twistypuzzles from any of more than an infinite number states?

Nah.

3x3x3 = 27
5x5x5 = 125
Megaminx is like a 3x3x3

They are all far from infinity!

I just realised that those numbers are actually pretty accurate :
1x1x1 = 1 (Start timer, touch cube, stop timer)
2x2x2 = 8 (That's about my average)
3x3x3 = 27 (Again, just about my average)
4x4x4 = 64 (I need to practise a bit more, but 1:04 should be possible)
5x5x5 = 125 (I am getting pretty close to 2:05)
megaminx is like a 3x3x3 only you do the same a couple of times (about 7 times slower for me, about 6 times for Erik so I guess that works)

I never realised this. Thanks for supplying such wonderfull math

 

[jazzthief81]

So it looks like it's going to be completed this evening and I would like to thank everyone once more for their contributions. This has been awesome. I had a quick look in the logs and the server received results from over 85 different machines.

I will a post some more detailed analysis about this study/experiment somewhere this weekend, but right now I'm in a bit of a hurry because I'm going to Rama's place (I heard something about a birthday party taking place ).

Kind regards,
Lars

 

[DanHarris]

The study is now complete You can view the data here - http://www.cubezone.be/colorneutrality.html

Some very interesting results! 40 positions at depth 8, and an overall weighted average of 4.8 turns, which I believe is a saving of around a move or so!

I look forward to your analysis Lars

Dan

 

[Kenneth]

Can someone maybe create generators for the 8 turn crosses?

I would love to see how many turns I need to solve those

 

[jazzthief81]

Can someone maybe create generators for the 8 turn crosses?

I would love to see how many turns I need to solve those

Yes Kenneth, I will do that as soon as I get back home on Sunday.

 

[cuBerBruce]

Wait a minute... Isn't the "total" value 18000000 too high?

The error appears to be orientation #8, depth 1. It appears it should be 1.393.424 instead of 19.393.424.