海角大神

Stanford-trained mathematician Tomas Rokicki has broken his own record, proving that any (solvable) Rubik鈥檚 Cube position can be cracked in or fewer.

For anyone who鈥檚 spent hours twirling the sides of the colored puzzle, this news might just be frustrating. But for mathematicians, this is a big success for elegant computer solutions.

Rubik鈥檚 Cubes, in case you forgot about the 鈥�70s sensation, challenge players to spin sides of 3x3x3 cube in order to make each surface a single color.

Last year, students at Northeastern University proved they could solve even the most twisted configuration in as few as 26 moves. In March, one-upped them by whittling down that number to 25. used a computer to churn through possible paths toward victory and filtered out any repeats. This is how most mathematicians solve puzzles nowadays when there are 43 quintillion possible positions 鈥� teach a computer to do it, and just wait for it to consider every possibility.

But this takes a really long time. Rokicki sped up the process by finding a clever way to divvy up the possible configurations into sets. It still took his PC to come up with a solution.

The leap from 25 to 23 moves came courtesy of , the supercomputers that brought us 鈥�Spider-Man 3鈥� and 鈥淪urf鈥檚 Up.鈥� Rokicki programmed the Sony processors to whiz through permutations in between rendering movie special effects.

We already know that there is no position that takes to solve. (My first reaction to this was, 鈥淲hat do you mean? Take a finished Rubik鈥檚 Cube and spin it 21 times.鈥� But Rokicki found that whatever configuration you land on after that 21st twirl can be solved in fewer than 21 rotations.)

This quirk suggests that, after more refinement, it鈥檚 possible that computers could push the cap down to 20 moves. Rockicki is already chugging away at 22. But it will take a lot of time and processing power to reach that number.

[Via ]

Also check out:
Computers master the game board
Wisdom of the crowd triumphs in alternate reality games