Glassolver project

[CubePhysics]

Hi. You remember when Gods Number was finally found? We'll we intend to do just that... But more. With this we will be using Glass. Google Glass. The only thing I'm having problems with is coding it in C++. I can do it just fine with Java, but, quite frankly, I don't want to wait a couple days for it to solve.

HOW IT WORKS

Utilizing Google Glass's camera, you will be able to take pictures of the cube's faces, and will automatically calculate God's Algorithm to solve it (like the Rubiks app sort of already does). The cool part, though, is that (for starts) it will work with 2x2, 3x3, 4x4, and 5x5. In the first update, it will also include all variants of the V-CUBE, and the "funky" Rubiks Cuves (void, tower, etc). And, periodically, we will add new cubes, like the Bermuda Triangles and Crazy variants of cubes.

THE TEAM

right now, it's just me, a guy with and old MacBook Pro, iPhone 5S, Nexus 5 and Glass. And a dream.

WHO WE WELCOME

-anyone good with algorithms
-C++, Control C, Java coders (esp people who know how to port Java to C++.
-UI developer
-Glass Developers
-Good with image to data working
-Hackers (to look at code from the Rubiks app)

REQUIREMENTS

-Must know, one way or another, how to solve the 2x2, 3x3, 4x4, and 5x5 cubes. EXCEPTIONS CAN BE MADE.
-Skype Acct.

Please post questions/comments/errors/emissions/anything!

 

[qqwref]

Wait, you're going to generate optimal solutions for the 4x4x4 and 5x5x5? You do know nobody's figured out how to do that, right? Got any great ideas that might make it possible?

The idea of having a Google Glass program to look at and solve a few common cubes is a good one, and I'm interested to see what you'll end up with. I just think you don't quite realize how difficult some things are

 

[Coolster01]

Great idea, but now we won't be able to wear Google glass in competition...

 

[CubePhysics]

Were we ever allowed to wear glass in comp?

 

[CubePhysics]

Wait, you're going to generate optimal solutions for the 4x4x4 and 5x5x5? You do know nobody's figured out how to do that, right? Got any great ideas that might make it possible?

The idea of having a Google Glass program to look at and solve a few common cubes is a good one, and I'm interested to see what you'll end up with. I just think you don't quite realize how difficult some things are

Yeah, it's gonna be hard. But the only reason we haven't been able to figure it out is that no one wants to put a 100 million supercomputer into the project. We're trying to work with Google for that. Why they would give it to us? Because of the Google Glass part. It would be brilliant marketing for them, and it would ease the tension devs have with Glass now, because they think it has a limited amount of possibilities.

 

[qqwref]

But the only reason we haven't been able to figure it out is that no one wants to put a 100 million supercomputer into the project.

Wrong. The 3x3x3 has 4.3*10^19 positions; the 4x4x4 has 7.4*10^45 positions. That means optimally solving a 4x4x4 position is about 10^26 times as hard as optimally solving a 3x3x3 position (again, unless you have some amazing idea). If a typical desktop computer can optimally solve a 3x3x3 position in one millisecond (and it can't) then a computer a trillion times as powerful would take over 5000 years to optimally solve a 4x4x4 position. With or without supercomputers, it's VERY far from being feasible.

Besides, even if you could somehow do it, the amount of computation required would be gigantic - a user could certainly not expect a solution any time soon, even with all the supercomputers in the world working on it. And the power used to carry out that computation would certainly not be worth your "brilliant marketing" of a few hundred app users. I think Google has much more efficient advertising ideas lined up.

 

[Jakube]

What you can do is calculating nearly optimal solutions and show them. Like using kociemba two-phase algorithm for the 3x3x3. There is a 4 (?) step solver for the 4x4x4. And you can do similar algorithms for even bigger puzzles.

 

[mark49152]

Optimal 4x4 and up may be infeasible but I still like the Glass idea. Non-optimal would be more interesting, to make it like a tutor. Imagine: "next we pair the edges, try like this...". And "no wait, you screwed up!"

 

[Lucas Garron]

Just don't try to do everything from scratch.

Optimal solutions are useless here; either use an existing Kociemba solver that can get you near optimal in milliseconds, or do something cool like a human-understandable solution.

http://cubecheater.efaller.com/
https://bitbucket.org/IngeneroiOS/cubetimer-scramblers/
http://cube20.org/

This is certainly interesting; just make sure that you have some idea of what you're doing.

 

[tim]

I'm having problems with is coding it in C++. I can do it just fine with Java, but, quite frankly, I don't want to wait a couple days for it to solve.

1.) C++ is very similar to Java. Porting it from one language to the other is trivial.
2.) Don't expect any significant speed boosts from using C++ instead of Java. Java's performance isn't as bad as you think it is.

 

[Kirjava]

Are you sure you didn't misunderstand what the cube20 project actually accomplished?

 

[CubePhysics]

Wrong. The 3x3x3 has 4.3*10^19 positions; the 4x4x4 has 7.4*10^45 positions. That means optimally solving a 4x4x4 position is about 10^26 times as hard as optimally solving a 3x3x3 position (again, unless you have some amazing idea). If a typical desktop computer can optimally solve a 3x3x3 position in one millisecond (and it can't) then a computer a trillion times as powerful would take over 5000 years to optimally solve a 4x4x4 position. With or without supercomputers, it's VERY far from being feasible.

Besides, even if you could somehow do it, the amount of computation required would be gigantic - a user could certainly not expect a solution any time soon, even with all the supercomputers in the world working on it. And the power used to carry out that computation would certainly not be worth your "brilliant marketing" of a few hundred app users. I think Google has much more efficient advertising ideas lined up.

I'm sorry, but I AM right. Maybe it would take ONE computer 5000 years, but lolno. No ****ing way I would have a computer work on a single solve for more than a day. And PS: supercomputers aren't really just one computer.

 

[CubePhysics]

Optimal 4x4 and up may be infeasible but I still like the Glass idea. Non-optimal would be more interesting, to make it like a tutor. Imagine: "next we pair the edges, try like this...". And "no wait, you screwed up!"

I'm liking what I'm seeing .

 

[Mike Hughey]

I'm sorry, but I AM right. Maybe it would take ONE computer 5000 years, but lolno. No ****ing way I would have a computer work on a single solve for more than a day. And PS: supercomputers aren't really just one computer.

Are you saying Google actually has 100 million supercomputers? I was always under the impression that Google's "supercomputer" was simply many millions of ordinary computers networked together rather efficiently.

And if we were just talking 100 million ordinary computers, that only removes 8 zeroes from the problem. Not enough to even make a minor dent in the 4x4x4 God's Algorithm problem. And 5x5x5 is so much harder than even that.

Truth is, if all you do is handle the 3x3x3 efficiently (not even optimally), it will make for a pretty good marketing scheme. Most people don't even care about 4x4x4 or 5x5x5. So I'm not sure why you're even bothering talking about them - they're surely not worth your time. Focus on 3x3x3; it would be very effective.

 

[CubezUBR]

im no computer nerd, but i think google glasses cant handle that much stuff