Multiplayer 15 puzzle sim

Stefan, soup is just a guy who solves slidy puzzles. Ben is the programmer of this and has been since the beginning.

And unless I'm mistaken the scrambling algorithm is just the canonical random permutation one - for each piece, in order, swap it with a random piece not earlier than it - with a parity fix.
 
qqwref said:
And unless I'm mistaken the scrambling algorithm is just the canonical random permutation one - for each piece, in order, swap it with a random piece not earlier than it - with a parity fix.
Click to expand...

So, an interesting fact about the 15 puzzle is that this will *not* give the same distribution as "do lots of random moves", for what I think is the most natural definition of "random move" (pick a random piece adjacent to the hole and move it into the hole).

MRSS will leave the hole in a center square twice as often as a corner (in a stable distribution, the probability of being on a square is proportional to the number of neighbors).

Random permutations are fine, but this is something to be aware of.
 
You're right, it certainly won't give the same distribution as lots of random moves. Lots of random moves is an extremely inefficient and poor scrambling method for sliding puzzles, especially for larger puzzles - if you don't do enough moves, pieces tend to stay roughly where they started, and it's not clear how many moves are enough. Consider a 20x20 puzzle and imagine how many moves it would take to leave the 1 tile on the bottom right as often as on the top left! So yeah, as far as I know, none of the slidysims designed for speedsolving have ever used random move scrambling.
 
qqwref said:
You're right, it certainly won't give the same distribution as lots of random moves. Lots of random moves is an extremely inefficient and poor scrambling method for sliding puzzles, especially for larger puzzles - if you don't do enough moves, pieces tend to stay roughly where they started, and it's not clear how many moves are enough. Consider a 20x20 puzzle and imagine how many moves it would take to leave the 1 tile on the bottom right as often as on the top left! So yeah, as far as I know, none of the slidysims designed for speedsolving have ever used random move scrambling.
Click to expand...

Some WCA puzzles would also be very inefficient to scramble using lots of random moves, but we still use the resulting distribution to define a random state.
Since we *are* turning physical puzzles, the idea of "do lots of random moves" makes sense. We're just lucky that most puzzles have something that matches with "assemble the puzzle randomly". (Square-1 is the funniest case, but we use the most natural definition that works for both.)

Anyhow, it doesn't matter much. Just pointing out that "random state" is not well-defined here without further qualifications.
 
Lucas Garron said:
Some WCA puzzles would also be very inefficient to scramble using lots of random moves, but we still use the resulting distribution to define a random state.
Click to expand...
Honestly, it's nowhere near the same (except in Square-1, which is a bandaged puzzle). No cube or minx of any size needs more than a handful of moves to solve a given piece; a sliding puzzle of size n needs O(n) moves.

Lucas Garron said:
Just pointing out that "random state" is not well-defined here without further qualifications.
Click to expand...
Mathematically, it's never well-defined ("random" just means that it's not deterministic). Anyway, the consensus for sliding puzzles is that we want to be solving scrambles from a uniform distribution. Handscrambling random moves is just not a good model here; anyone who has played with a physical puzzle knows that for any reasonably small number of moves the scrambles are noticeably easy (compare megaminx, where a scrambling algorithm can only ever generate a tiny fraction of the position space, and yet still make positions that look scrambled and feel very random).
 
v19.0 (link is same as in OP)
Pasting scrambles from skype messages is easier
Fixed bug where pasting a scramble as the first solve after opening the sim would have a stupid tps
5x5 grids colour scheme
Slider to change distance between tiles
Option to enable/disable the black border around each tile
Changed save file format to allow saving during relays and marathons
Pasting a scramble for a different puzzle than the current one will ask if you want to switch to that puzzle
Width/height only relay pb files fixed
Fixed bug where you couldn't connect to a server if the puzzle was 10x10-20x20 (forgot to remove old size limit)
Leaderboards!!!!!!!!!!!
 
v19.2
Fixed bug where relays/marathons on multiplayer would only show the first puzzle as scrambled and the rest as solved
Usernames in the chat box are shown in bold text
Leaderboard window is non-modal (can switch between it and the main window without having to close the leaderboard window)
 
v19.3
Leaderboard button doesn't show in fullscreen from alt+enter
New lines in chat are replaced with backslashes so you can still copy/paste scrambles
You can run the same leaderboard query multiple times in a row now
Pasted scrambles are read slightly better
 
Last edited:
For the folks who have not updated to v19.6 (wwavelet, Paul, leon, gxx, shangshanruoshui, mylxc60, etc), please update.

The leaderboards are displaying times that are 1/1000th what they're supposed to be because of this, and the people on that small list there may be aware of that~
 
soup said:
For the folks who have not updated to v19.6 (wwavelet, Paul, leon, gxx, shangshanruoshui, mylxc60, etc), please update.

The leaderboards are displaying times that are 1/1000th what they're supposed to be because of this, and the people on that small list there may be aware of that~
Click to expand...
I'm still on v16.0, but I never go online. Does that matter?
 
v19.7
Pressing enter has the same effect as pressing get leaderboard
Username leaderboard filter
Ctrl+C to copy stuff from the leaderboard
Movecounts are correct in relays and marathons (whoops)
 
You must log in or register to reply here.

Similar threads

ben1996123
QCube Multiplayer
Replies
15
Views
4,852
Veerexx
Veerexx
liljthedude
Ryan Heise Multiplayer Cube
Replies
12
Views
2,665
Sa967St
Sa967St
L
Multiplayer Cube Simulator
Replies
6
Views
1,741
Richard
R
Jteddy
Speedsliding 2026
Replies
8
Views
128
pi=3.14
pi=3.14
Puzzle Maniac
4-Pad Results Thread
Replies
3
Views
287
acrystal
acrystal
Back
Top