Congratulations, rokicki !!!

Phil Pilgrim (PhiPi)

Tomas Rokicki, a familiar name in the Parallax forums, got a mention on Slashdot today for proving that the Rubik's Cube can be solved from any position in 25 moves or less. The article and links make fascinating reading.

Kudos, Tomas!

-Phil

Post Edited (Phil Pilgrim (PhiPi)) : 3/27/2008 6:05:46 AM GMT

rokicki

Thanks, Phil!

I'll have to explore this "sandbox" thing.

Anyway, this is one of several things that has been keeping me away from the prop recently.

-tom

skylight

For those rubiks enthusiasts that cant get their cube back to normal An Excellent Rubiks Cube Solver

mpark

That is too awesome. I was reading that slashdot article and wondering why the name Rokicki sounded so familiar, and then I thought, naw, it couldn't be... could it?

But you needed a workstation with 8GB of memory and around 1500 hours of time on a Q6600 CPU running at 1.6GHz? Whatsamatta, a Prop not good enough for you?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Michael Park

PS, BTW, and FYI:
To search the forum, use search.parallax.com (do not use the Search button).
Check out the Propeller Wiki: propeller.wikispaces.com/

Beau Schwabe

rokicki,

Great job! ...I can get about 2/3rds of·a cube solved in relatively few moves, and with a little mechanical cheating· I can solve it the rest of the way.

I'd offer the 2.5Gig / 3.2GHz machine I·own, but you wouldn't like me standing over your shoulder while you use it.· ...I just wouldn't leave you alone asking questions.





▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Baggers

Great job rokicki [noparse];)[/noparse]
I can see why you've been soooo busy lately [noparse]:)[/noparse] smarty pants lol

Post Edited By Moderator (Chris Savage (Parallax)) : 4/1/2008 5:22:13 PM GMT

stevenmess2004

So rokicki, when do we see the version that runs on the prop? Lets see, if we partition all the matrices and use an SD card or 4 and maybe a Hydra RAM card. But processing time could be a problem. Did I say good job? I have never been able to solve those things (probably because the stickers got pulled off and put in the wrong places) so this is very interesting to know. While solving this did you end up figuring the quickest to the solution from any position or just that it was possible?

rokicki

Right now my program uses 5GB of RAM to run, so when we get that much memory on the Prop (and it has to be
fast, so SD card access just won't do it), then we'll talk. Prop 3 maybe?

The program is currently being run in a mode that just shows it is possible. There is another mode which lets it
find specific solutions to the full set of positions, but that takes longer. Sometimes much longer.

(By the way, I've been working on this for *years* on and off; it's not just an overnight thing. Also, I made
an error in the paper; I said the Q6600 runs at 1.6GHz when in reality it runs at 2.4GHz. Oops.)

Phil Pilgrim (PhiPi)

It appears that Tom Rokicki has done it again: this time proving that the Rubik's Cube can be solved from any position in 23 moves or less. Go Tom!

-Phil

LoopyByteloose

The first time I saw a Rubix cube, I solved it in about 10 minutes. The people that introduce me to it had been working for days without progress.

Mainly I did so by first lining up the corners to their appropriate faces and then getting the top and bottom colors complete. After that it seemed easy to resolve by visual analysis with trial and error.

But having the discipline to demonstrate a minimal number of moves is beyond me.

I have no idea of what I do [noparse][[/noparse]other than reduce the variables to 2 axis rather than dealing constantly with manipulating 3 axis, but I still can solve it in 10 - 20 minutes by this approach.

People just hate it when you do that.

I suppose those stages can be presented in a mathematical paper. The motion of the corners is somewhat independent of the center crosses. With 3 color faces on each corner, it seemed best to solve those first.

So there are two main phases - the corners first and the crosses last. The second phase is worked faster by having any two opposite faces first resolved and they figuring out the last four according to direct observation.

Maybe this approach can run on a Propeller. Good luck.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PLEASE CONSIDER the following:

Do you want a quickly operational black box solution or the knowledge included therein?······

···················· Tropically,····· G. Herzog [noparse][[/noparse]·黃鶴 ]·in Taiwan

Post Edited (Kramer) : 6/12/2008 4:37:44 PM GMT

LoopyByteloose

BTW, I have a dodecahedron icosohedron [noparse][[/noparse]20 triangular sides with 5 colors] Rubix-like puzzle worked out that can be created and played in computer graphics, but cannot be physically built. {I was really going to try to take the physical version to market, but alas it is absolutely beyond the limitations of machine mechanics.}

If anyone wants to collaborate on that, I am more than willing to share the wealth and glory with them. I'll never get it done alone. I'd even accept just a mention.

Oddly enough, it also serves as a wonderful 3-D model of the relationships of Yin-Yang and the the 5 elements in Chinese medicine, even encompasses the circulation on the 12 meridians of the body.

So far, it is just yet another bit of useless Asian knowledge that I've uncovered.

Kudos to you, Tom Rokicki.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PLEASE CONSIDER the following:

Do you want a quickly operational black box solution or the knowledge included therein?······

···················· Tropically,····· G. Herzog [noparse][[/noparse]·黃鶴 ]·in Taiwan

Post Edited (Kramer) : 6/16/2008 10:11:49 AM GMT

Dan Taylor

I am a big fan of the Ribix's Cube! And am able too solve it with no help.
But 23 moves is amazing! Is there anyplace he shows you how do solve it in 23 moves?!?!?!

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Dan Taylor

rokicki

Remember, this is a proof that there are no positions that require more than 23 moves. While it is true that
I did indeed compute a "solution" for every position, there is no way I could store that many solutions,
so the program essentially finds the solution, marks it as existing (one bit in memory) and continues along,
not remembering the solution itself.

For actual speedsolving techniques, there are many, and they tend to be very interesting. One that I really
like builds a two-by-two corner and fixes the other three "faces" using a number of moves; that's a very
neat technique compared to the usual cross, first two layers, last layer algorithms. But I'm no speed cuber.

Kramer, if you were able to solve the cube within 10 minutes of your first exposure you are a genius.
It took me many months my first time, and lots of notebook paper. Even now I typically take about two
minutes a solve, which means I do not "qualify" as a speed solver (in general, I refuse to memorize lists
of algorithms; I prefer to rediscover the algorithms as I need them).

Anyway, thanks for all the attention, guys! Yeah, I've got fires on many fronts at the moment.