The Batch Solver - Generate large algorithm sets automatically!

[IsThatA4x4]

so i wanted to make a sort of a PBL step for the 3x3, where CO is done and the E slice needs to be preserved. this is what i put as the scramble:

<U, D,
R U R' F' R U R' U' R' F R2 U' R', --J perm
R' U R' F' R U R' U' R' F R2 U' R' U R2, --J perm on D
F R U' R' U' R U R' F' R U R' U' R' F R F', --Y perm
L2 R2 F R U' R' U' R U R' F' R U R' U' R' F R F' R2 L2 --Y perm on D
>

there should be only 8 cases, but the program somehow got 172. i didnt check them all out sice i didnt want to run it for this long. what could the problem be? (M and S slice edges are equivalent and dont need to be oriented. CO is done and every corner is in the correct layer. E slice is done. at the end of the alg all pieces except the M and S slice should be solved)

EDIT: oh and the algs are <R, U, D, F> gen, prune is 4 and search is 8, pre- and post- adjust are both <U, D>

What exactly did you put in the orientations and equivalences box?

 

[mk8]

What exactly did you put in the orientations and equivalences box?

1:{UF UB UR UL DF DB DR DL}

 

[mk8]

welp i again fixed it myself with using z2, but i would still like to know what the problem was and how to fix it

 

[OreKehStrah]

1:{UF UB UR UL DF DB DR DL}

That is the problem. Remove the 1: because that tells the solver those pieces are equivalent and the orientation doesn’t matter.
Edit: misread because I am dumb. Disregard this lol.

 

[mk8]

guess whos here with more annoying questions

this is partly kind of a puzzle theory question, but its related to the batch solver:
whats a subset of moves/sequences (for the scramble) that 1. doesnt break EO 2. preserves the E slice and 3. preserves all the corners? im making an algset that permutes edges of the M and S slices after EO (corners and E slice have already been done), but i cant get the program to give me all the cases. i only use M2 and S2 so i dont break EO, but that only gives line, cross or checkerboard on top/bottom, and i have found (by playing with the cube) cases when there are 3 D edges on U. is this even possible? (kociemba cube solvers have given me pretty long <R U F B L D> solutions to these when i manually entered them in, so im not sure if i can manage this with only <M U D>, maybe <M U D R> or <M U D S>?)

an alternative i could do earlier in the solve is to solve DR and DL together with CP, and then only do regular L6E instead of L8E. but i cant figure out a move subset that preserves the belt and CO, but breaks CP and EO on M and S. any help would be very cool

 

[ruffleduck]

guess whos here with more annoying questions

this is partly kind of a puzzle theory question, but its related to the batch solver:
whats a subset of moves/sequences (for the scramble) that 1. doesnt break EO 2. preserves the E slice and 3. preserves all the corners? im making an algset that permutes edges of the M and S slices after EO (corners and E slice have already been done), but i cant get the program to give me all the cases. i only use M2 and S2 so i dont break EO, but that only gives line, cross or checkerboard on top/bottom, and i have found (by playing with the cube) cases when there are 3 D edges on U. is this even possible? (kociemba cube solvers have given me pretty long <R U F B L D> solutions to these when i manually entered them in, so im not sure if i can manage this with only <M U D>, maybe <M U D R> or <M U D S>?)

an alternative i could do earlier in the solve is to solve DR and DL together with CP, and then only do regular L6E instead of L8E. but i cant figure out a move subset that preserves the belt and CO, but breaks CP and EO on M and S. any help would be very cool

<M2, S2, M' U2 M, M U2 M', S' U2 S, S U2 S'> ?

 

[IsThatA4x4]

guess whos here with more annoying questions

this is partly kind of a puzzle theory question, but its related to the batch solver:
whats a subset of moves/sequences (for the scramble) that 1. doesnt break EO 2. preserves the E slice and 3. preserves all the corners? im making an algset that permutes edges of the M and S slices after EO (corners and E slice have already been done), but i cant get the program to give me all the cases. i only use M2 and S2 so i dont break EO, but that only gives line, cross or checkerboard on top/bottom, and i have found (by playing with the cube) cases when there are 3 D edges on U. is this even possible? (kociemba cube solvers have given me pretty long <R U F B L D> solutions to these when i manually entered them in, so im not sure if i can manage this with only <M U D>, maybe <M U D R> or <M U D S>?)

an alternative i could do earlier in the solve is to solve DR and DL together with CP, and then only do regular L6E instead of L8E. but i cant figure out a move subset that preserves the belt and CO, but breaks CP and EO on M and S. any help would be very cool

<M2, U, D> probably works no?

if not then <M2, U, D, M' U2 M> should definitely work

 

[mk8]

also, an option to filter out all cases under a certain amount of moves (i dont need the solution for M2 in my sheet cuz its kinda obvious) and to hide failed cases would be pretty nice

EDIT: and something that probably doesn't even require that much effort: if a case fails, instead of a "solution" field there would be a "scramble" field with the (shortest) scramble used to achieve that state, so then you could manually check the failed cases in cube explorer instead of running the entire search again

 

[trangium]

so i wanted to make a sort of a PBL step for the 3x3, where CO is done and the E slice needs to be preserved. this is what i put as the scramble:

<U, D,
R U R' F' R U R' U' R' F R2 U' R', --J perm
R' U R' F' R U R' U' R' F R2 U' R' U R2, --J perm on D
F R U' R' U' R U R' F' R U R' U' R' F R F', --Y perm
L2 R2 F R U' R' U' R U R' F' R U R' U' R' F R F' R2 L2 --Y perm on D
>

there should be only 8 cases, but the program somehow got 172. i didnt check them all out sice i didnt want to run it for this long. what could the problem be? (M and S slice edges are equivalent and dont need to be oriented. CO is done and every corner is in the correct layer. E slice is done. at the end of the alg all pieces except the M and S slice should be solved)

EDIT: oh and the algs are <R, U, D, F> gen, prune is 4 and search is 8, pre- and post- adjust are both <U, D>

I think the problem is your J perm on D algorithm.

You wrote: R' U R' F' R U R' U' R' F R2 U' R' U R2
It should be: R' U R' F' R U R' U' R' F R2 U' R' U' R2

Hopefully this fixes your problem.

 

[Kyoobies]

Okay I've done what I could but I'm officially stuck and need help. I'm creating an alg set for my master pyraminx stuff, but this is post reduction so I only need a Jing's pyraminx for the puzzle definition (Completely ignoring the top tip center though, and instead solving L4E and face centers). And I did what I could, but when generating the algs it says that there's 131 cases it found. This was before I realized that I actually was messing up and said that the top tip also needs to be solved. Once I fixed that and said that it can be in any position, it actually jumped up to over 500 possible cases somehow.

But accounting for AUF, there should only be about double the amount of algs as L4E- which is 35. Because the rotation of the top centers takes the place of the top tip center, and that only leaves cases where the the bottom center is solved, or cases where the bottom center needs to swap with the front center. I'm assuming 71 cases total since it's double L4E, plus a pure centers flip for 71.

So I'm not sure what I'm doing wrong and I don't know how ignoring some pieces made it more cases instead of less, but I'm guessing I'm doing my definition wrong with the centers or something. If anyone can help that would be awesome, thank you!

 

[IsThatA4x4]

Okay I've done what I could but I'm officially stuck and need help. I'm creating an alg set for my master pyraminx stuff, but this is post reduction so I only need a Jing's pyraminx for the puzzle definition (Completely ignoring the top tip center though, and instead solving L4E and face centers). And I did what I could, but when generating the algs it says that there's 131 cases it found. This was before I realized that I actually was messing up and said that the top tip also needs to be solved. Once I fixed that and said that it can be in any position, it actually jumped up to over 500 possible cases somehow.

But accounting for AUF, there should only be about double the amount of algs as L4E- which is 35. Because the rotation of the top centers takes the place of the top tip center, and that only leaves cases where the the bottom center is solved, or cases where the bottom center needs to swap with the front center. I'm assuming 71 cases total since it's double L4E, plus a pure centers flip for 71.

So I'm not sure what I'm doing wrong and I don't know how ignoring some pieces made it more cases instead of less, but I'm guessing I'm doing my definition wrong with the centers or something. If anyone can help that would be awesome, thank you! View attachment 20057

Putting 1: without curly braces after specifies that you want pieces to be permuted but not oriented, if you want to completely ignore them 1:{} will do that
I don't know what your alg set is meant to do as you didn't specify it, so I'm only guessing at the issue here

 

[mk8]

ok so this time, im completely lost and dont know what to do

im genning a (sort of) f2l algset, but without the cross, so the pair's edge could be in any of the cross pieces' positions

according to my calculations, (not accounting for advanced cases where either of the pieces is in the wrong slot) this should be 48 more cases than normal (its ehh, but theyre all really short so thats not the problem)

anyway, im not sure what subgroup/scramble should i do. theres just so many inserts & stuff in f2l, so idk which i should include and which are unnecessary. can someone help?

edit: just to clarify, i want to gen standard f2l algs which only affect the single slot

edit2: also if the case number is too big im ok with a short step at the start that reduces the number of cases (such as putting all the E edges into U or putting U/D edges into all cross slots, ...). any ideas would help

edit3: and also, it doesn't have to be the standard f2l approach, im just trying to do f2l minus cross in the fewest moves possible (with a tolerable algcount ofc)

 

[Kyoobies]

Putting 1: without curly braces after specifies that you want pieces to be permuted but not oriented, if you want to completely ignore them 1:{} will do that
I don't know what your alg set is meant to do as you didn't specify it, so I'm only guessing at the issue here

I did try this initially, along with several minor changes (tip centers and trivial tips can't change permutatio is why). It was actually less cases when I made a mistake and forgot to tell it to ignore the trivial tip that's connected to the tip center. Fixing that and telling it to ignore top tips actually adds more cases despite solving less pieces; and if I remove that and say tips must he solved it goes back to 131.

And what I'm try to do is generate all of l4e for pyraminx, but with the addition of the face centers that a master pyraminx (or jings pyraminx) has. I'm also completely ignoring the U tip center (although on a regular jings pyraminx it's technically impossible to solve edges and centers without solving tip centers).

 

[Kyoobies]

I think I figured it out. Jings Pyraminx has an extremely abnormal application of orientation for the face centers. But, it has been done before and I can translate the definition file by hand (thank you twizzle explorer). I don't have time to do it now but I should have all I need so well see how it goes!

 

[mk8]

im tryna gen CMSLL rn (basically CxLL that disregards the M and S slices) and i thought it would be simple, but nope, dumbass me somehow messed up again
D layer corners and E slice are solved (and need to be preserved), i need to solve U layer corners, everything else (U/D edges) is irrelevant

Equivalences: 1:{UF UR UB UL DF DR DB DL}
Scramble: <U, R U R' U R U2 R, R U R' F' R U R' U' R' F R2 U' R' U'>     (Us, sunes and jperms)

solving R U F gen with prune 5 and search 5 and pre-/post-AUF is U only
instead of the 42 cases i should have gotten i got 20k+, no idea why
help very moch appreciated

edit: ok now its giving the correct cases, all i did was reorder the edges in equivalences. why is this? (i have put the in order M then S , rather than U then D)

also im trying a sort of a eg-1-ish thing, where DFR and DFL are swapped and i need to solve them along with U corners. the scramble i put is

R2 F U F' R2 F U' F U F2 <U, R U R' U R U2 R, R U R' F' R U R' U' R' F R2 U' R' U'>

everything else same as before
the case number is ridiculously high again
what in the world am i doing wrong?
(R2 F U F' R2 F U' F U F2 swaps DFR and DFL)

 

[trangium]

im tryna gen CMSLL rn (basically CxLL that disregards the M and S slices) and i thought it would be simple, but nope, dumbass me somehow messed up again
D layer corners and E slice are solved (and need to be preserved), i need to solve U layer corners, everything else (U/D edges) is irrelevant

Equivalences: 1:{UF UR UB UL DF DR DB DL}
Scramble: <U, R U R' U R U2 R, R U R' F' R U R' U' R' F R2 U' R' U'>     (Us, sunes and jperms)

solving R U F gen with prune 5 and search 5 and pre-/post-AUF is U only
instead of the 42 cases i should have gotten i got 20k+, no idea why
help very moch appreciated

edit: ok now its giving the correct cases, all i did was reorder the edges in equivalences. why is this? (i have put the in order M then S , rather than U then D)

also im trying a sort of a eg-1-ish thing, where DFR and DFL are swapped and i need to solve them along with U corners. the scramble i put is

R2 F U F' R2 F U' F U F2 <U, R U R' U R U2 R, R U R' F' R U R' U' R' F R2 U' R' U'>

everything else same as before
the case number is ridiculously high again
what in the world am i doing wrong?
(R2 F U F' R2 F U' F U F2 swaps DFR and DFL)

The issue is probably your Sune algorithm.

You wrote: <U, R U R' U R U2 R, R U R' F' R U R' U' R' F R2 U' R' U'>
Should be: <U, R U R' U R U2 R', R U R' F' R U R' U' R' F R2 U' R' U'>

Hopefully this fixes it!

 

[Stasios]

Hi! I am quite new to this and so I don't know exactly how to go about it, and the tutorial hasn't helped much. How do I label the edges on the E layer as non-permutated and non-oriented? Thank you very much, help would be appreciated!

 

[trangium]

Hi! I am quite new to this and so I don't know exactly how to go about it, and the tutorial hasn't helped much. How do I label the edges on the E layer as non-permutated and non-oriented? Thank you very much, help would be appreciated!

To indicate that the orientation of pieces don't matter, place "1: " at the start of the line.
To indicate that the permutation of pieces don't matter, surround them with curly braces.
The pieces in the E layer are FR, BR, BL, FL.

Putting this together, in unique orientations and equivalences, put 1: {FR BR BL FL}

 

[trangium]

I created official documentation of all features of the batch solver. (I actually made it a while ago, but I figured I should post it here.) If you're curious about a more advanced or recent feature not covered in the tutorial, this is where to look. I've added quite a few features since the original post, and the documentation explains them all. (I usually report my development updates in my discord server, but the documentation should always be up-to-date.)

 

[Burrito]

Hi, I am trying to gen TSLE algs (for zz, inserts last f2l edge and orients corners) and BatchSOlveer keeeps crashing.

Prune: 17

Subgroup: R, U, D, r, S, M

Scramble: < R U R' U R U2 R', U> R U' R' U2

ANy heel;p[ is appreciated.