Cube Explorer 5.00 released - with slice moves

[nlCuber22]

Works for me. Not sure what your problem is.

Can you load the Cube Explorer page?
http://kociemba.org/cube.htm

 

[Chapuunka]

It works for me.

 

[PeterNewton]

Try http://chodecircus.com/area51/.

 

[Mattb10M]

I got it now. It was something with my network settings.

Thanks for the help though

 

[cubacca1972]

Outstanding version of Cube Explorer. Thanks.

If you were to release a 4x4x4 solver that performs this well, I would probably weep with joy.

 

[riffz]

Outstanding version of Cube Explorer. Thanks.

If you were to release a 4x4x4 solver that performs this well, I would probably weep with joy.

So would I, but it's not gonna happen any time soon.

 

[rubixfreak]

I have a problem with the programm:
my Norton internet security 2010 says that the usage of the program is highly risky and that it has a suspect behaviour.
Therefore, Norton always automatically removes the .exe when I want to run the prog.

Does anyone else has the same problem?

 

[Herbert Kociemba]

I have a problem with the programm:
my Norton internet security 2010 says that the usage of the program is highly risky and that it has a suspect behaviour.
Therefore, Norton always automatically removes the .exe when I want to run the prog.

Does anyone else has the same problem?

Interesting. Indeed Cube Explorer has a build in Web-Interface, but this is disabled by default. But maybe Norton finds the code where you have the option to open a port on your machine.

 

[rachmaninovian]

Outstanding version of Cube Explorer. Thanks.

If you were to release a 4x4x4 solver that performs this well, I would probably weep with joy.

well, clement's solver works pretty well =P

 

[Christopher Mowla]

Outstanding version of Cube Explorer. Thanks.

If you were to release a 4x4x4 solver that performs this well, I would probably weep with joy.

well, clement's solver works pretty well =P

What is the minimum number of half turn moves Clement's solver outputs for this parity case?

 

[cubacca1972]

well, clement's solver works pretty well =P

I have tried and failed several times at getting it to run on Ubuntu. I just can't seem to get the syntax right. Sigh.

 

[rachmaninovian]

Outstanding version of Cube Explorer. Thanks.

If you were to release a 4x4x4 solver that performs this well, I would probably weep with joy.

well, clement's solver works pretty well =P

What is the minimum number of half turn moves Clement's solver outputs for this parity case?

I think I tried to run the solver for that case but my computer ran out of memory. I can still try I guess....but from a speedsolving point of view, R U R' U' + lucas' BLD parity + U R U' R' is sufficiently fast.

 

[Christopher Mowla]

I think I tried to run the solver for that case but my computer ran out of memory. I can still try I guess...

Well, as they always say, "You can't do the same thing over again and obtain a different result". So, if you do try again (which I would appreciate) maybe use a different move restriction than the one you used before (or if you used a move restriction before, do not use one at all). Even if the solution is not short at all, any solution it yields will be interesting to me.

If it still doesn't work out, thanks for trying...I just wanted to test what the current 4X4X4 solvers can do.

 

[rachmaninovian]

I think I tried to run the solver for that case but my computer ran out of memory. I can still try I guess...

Well, as they always say, "You can't do the same thing over again and obtain a different result". So, if you do try again (which I would appreciate) maybe use a different move restriction than the one you used before (or if you used a move restriction before, do not use one at all). Even if the solution is not short at all, any solution it yields will be interesting to me.

If it still doesn't work out, thanks for trying...I just wanted to test what the current 4X4X4 solvers can do.

immediately after pm-ing you the solver spat out some solutions.
F2 U r2 F2 l F2 l' U2 r U2 F2 r F2 r U' F2(16)
F2 U r' F2 r' F2 U2 r' U2 l F2 l' F2 r2 U' F2(16)
F2 r2 U r' F2 r F2 r F2 U' r U F2 U' r F2(16)
F2 r2 U l' U2 r U2 l F2 U' r U F2 U' r F2(16)
F2 r' U F2 U' r' U F2 r' F2 r' F2 r U' r2 F2(16)
F2 r' U F2 U' r' U F2 l' U2 r' U2 l U' r2 F2(16)

so its 23 qtm minimum so far...

will continue editing this post if i find new stuff

 

[ezh]

I'm having the same problems. What did you do to fix it?

 

[cuBerBruce]

I've noticed that Cube Explorer does not generate all optimal solutions to the following scramble:

U2 R L' B2 R' L

Cube Explorer generates these solutions:
R L' B2 R' L U2 (6f*)
R B2 R' L U2 L' (6f*)
B2 R' L U2 R L' (6f*)

But omits this solution:
L' B2 R' L U2 R (6f*)

My belief is that this fourth solution is "pruned out" due to certain logic in the algorithm that also "prunes out" solutions that are equivalent due to switching the order of commuting moves. I note that Reid's optimal solver (that the Cube Explorer optimal solver is based on, to my understanding), also omits this fourth solution.

I note that all four of these solutions are essentially cyclic shifts of one another (allowing for switching the order of commuting moves). But only one gets omitted when you click the green arrow button multiple times to get "all" solutions.

For speedcubing purposes, I would think it might be nice to be able to have an option not to skip these not so trivially redundant solutions.

I also note that JAcube will produce all four of these solutions (using options "fao").

 

[Herbert Kociemba]

I've noticed that Cube Explorer does not generate all optimal solutions to the following scramble:

U2 R L' B2 R' L

Cube Explorer generates these solutions:
R L' B2 R' L U2 (6f*)
R B2 R' L U2 L' (6f*)
B2 R' L U2 R L' (6f*)

But omits this solution:
L' B2 R' L U2 R (6f*)

My belief is that this fourth solution is "pruned out" due to certain logic in the algorithm that also "prunes out" solutions that are equivalent due to switching the order of commuting moves. I note that Reid's optimal solver (that the Cube Explorer optimal solver is based on, to my understanding), also omits this fourth solution.

I note that all four of these solutions are essentially cyclic shifts of one another (allowing for switching the order of commuting moves). But only one gets omitted when you click the green arrow button multiple times to get "all" solutions.

For speedcubing purposes, I would think it might be nice to be able to have an option not to skip these not so trivially redundant solutions.

I also note that JAcube will produce all four of these solutions (using options "fao").

The reason is that the position has a symmetry (mirror symmetry at the RL-plane) and my program prunes the search tree. In this case Cube Explorer uses R for the first move, but never L' because of the symmetry.

If you observe the second maneuver R B2 R' L U2 L' maneuver in the RL-mirror you get L' B2 L R' U2 R and this is just your missing L' B2 R' L U2 R.

Of course it would be no problem to omit the symmetry check. For symmetric positions this just would increase the time to find the first optimal solution.
Btw., I never looked at the source coude of Mike Reids optimal solver, but presumably he just does this symmetry check too.

 

[cuBerBruce]

The reason is that the position has a symmetry (mirror symmetry at the RL-plane) and my program prunes the search tree. In this case Cube Explorer uses R for the first move, but never L' because of the symmetry.

If you observe the second maneuver R B2 R' L U2 L' maneuver in the RL-mirror you get L' B2 L R' U2 R and this is just your missing L' B2 R' L U2 R.

Of course it would be no problem to omit the symmetry check. For symmetric positions this just would increase the time to find the first optimal solution.
Btw., I never looked at the source coude of Mike Reids optimal solver, but presumably he just does this symmetry check too.

OK, that makes sense. I checked Reid's solver, and it has a "compile-time option" (trivial edit of source code required) to enable/disable the symmetry check. (With a little bit of editing, one can easily make it a runtime option.) With it disabled, it generates all four maneuvers for the case I mentioned. And yes, I assumed that "fixing" this (in Cube Explorer) would have a performance impact, so that's why I suggested it be a user option.

Also, it's cumbersome generating optimal solutions one at a time, and copying/pasting the results to an editor. Is there an alternative to this? OK, I see there is a trick of leaving a cubie gray, and it will create a window showing the solutions (including ones normally omitted by the symmetry check!), instead of adding the cube to the main window. It's just not intuitive that this feature exists from the user interface. I had to do a Google search to find it. (It seems I've probably seen this feature before, but forgot how to enable it.)

 

[qqwref]

Maybe there should be an option somewhere, "optimal search (allows undefined pieces)". When it is turned off, your cube must be completely filled in, and everything goes to the main pattern window; when it is turned on, everything goes in the new window, even if all the pieces are filled in. And if it's turned off and you try to solve a cube that isn't filled in, the error box should suggest turning it on.

 

[cuBerBruce]

Maybe there should be an option somewhere, "optimal search (allows undefined pieces)". When it is turned off, your cube must be completely filled in, and everything goes to the main pattern window; when it is turned on, everything goes in the new window, even if all the pieces are filled in. And if it's turned off and you try to solve a cube that isn't filled in, the error box should suggest turning it on.

qqwref's idea might be somewhat of an improvement, but I think it still must be more clear to the user "optimal search (allows undefined pieces)" is the feature that would allow you to generate a list of all optimal solutions for a completely specified cube.

If I already have a bunch of cubes in the main window (perhaps loaded from a file that was generated by a GAP program), I think I should be able to select one (or possibly multiple ones), and then select a menu item or click a button to generate a list of optimal solutions (and perhaps continue on to suboptimal ones as the existing window for incompletely specified cubes does). I don't think it should be necessary to have to transfer each cube back to the facelet editor before generating the list of optimal solutions.