hacker.org

A Rush-Hour like sliding block games could be interesting, if it hasn't been suggested before.

Not far, if you look at the high score table you'll see people slow down in the early 20ish.

I think that bitfield representation and operations is still the fastest option. The idea is more that, among your thousands of "equations", both during the elimination and substitution steps of gaussian, only a few will contain the parameter/variable you are trying to do something with.

Gauss is a good start, of course, but what you have to remember is to apply it to the context of this puzzle. We have huge matrices of 0's and 1's, but only a tiny fraction of it is actually a '1', which is what you are interested in. Most of the time you spend iterating through the matrix does not ...

By trying different puzzles :wink: . You could start by writing a brute-force solver. It won't get you very far, but will give you the general idea. Then, add stuff to that program, try to eliminate solutions trees early on. You should solve some problems by hand too to get used to the puzzle and to...

It's not cheating...some puzzles simply have a definite "solution". Runaway Robot's being fairly obvious, but Crossflip and Oneofus can essentially be represented as mathematical problems (which have been researched before - hence why so many people have done all the levels). Even if you k...

Only the puzzles with 2-3 colors. I only noticed this at 158 because neither my two algo can solve it so I wanted to verify some data by hand.

Edit : Found the problem. Moving on again.

you're presumably not actually using the flash/java app anyway

But yeah, that makes sense.

Did you run out of shapes or something? There aren't as many colors/shapes on screen then in the page's source code.

Simple, but nice puzzle, btw. Trying to break the level 158 wall (what the heck is with that one anyway?).

You could always try to make an algorithm that doesn't need much memory? Mine doesn't even reach a meg.