• Welcome to the Speedsolving.com, home of the web's largest puzzle community!
    You are currently viewing our forum as a guest which gives you limited access to join discussions and access our other features.

    Registration is fast, simple and absolutely free so please, join our community of 35,000+ people from around the world today!

    If you are already a member, simply login to hide this message and begin participating in the community!

My algorithm maker software

Joined
Oct 8, 2013
Messages
41
Likes
5
Location
Iran - Mashhad
WCA
2013AHMA01
Thread starter #1
After months of trying, finally I could write a program that got all the algorithms may lead to a particular state.
For example, to solve this case algorithms are calculated as follows.


algorithms are here
these are about 10000 algorithms!!!

All the possible algorithms are these and there is no faster algorithm to solve this case. and i didn't wrote longer algorithms.
you can test these algorithms as follow
f01 = f r01 = r ....
f11 = f' r11 = r' ....
f21 = f2 r21 = r2 ....

this is a new way to catch algorithms for your new methods. say your new methods here and I'll calculate algorithms that you need after approval of the Members.
 

Attachments

Joined
Oct 8, 2013
Messages
41
Likes
5
Location
Iran - Mashhad
WCA
2013AHMA01
Thread starter #7
my program is designed by python because I thought that's a fast language.
it's make all the possible algorithms to solve a special case. we have some database site just like Boca.bee.pl that every body can go there and add a new algorithm. These sites are no longer needed because we can find all the possible algorithms.
I can add a method in my program with some conditional statement. for example, PLL is an algorithm that Top face and first and second row does not change and ... then my program will provide all of PLL algorithms (in a special range)
the program is based on make all the algorithms and extract useful.
as you know we have 54 notations so we have 54^^10=... ten move algorithm and it is not possible to make all of them. All of my art is here! I got too complex ways to reduce computation time.
cube explorer just give you some good ways and there may be some better ways too. some times it's too hard to receive your algorithms one by one with cube explorer.
I'll show you my method in 2x2 first layer soon.

I'm sorry because i can't speed English very well!!!:)
 
Joined
Oct 8, 2013
Messages
41
Likes
5
Location
Iran - Mashhad
WCA
2013AHMA01
Thread starter #8
What do they mean?

Also, is it just 2x2, or can it go higher? We haven't got one like this for 4x4 and up; so this could be interesting if it does.
1 - they are all the possible algorithms to solve that special case
f01 = f r01 = r ....
f11 = f' r11 = r' ....
f21 = f2 r21 = r2 ....
2 - of course it can be use in bigger cubes too. The only problem is that for bigger cubes we have more processing time
processing time depends on your method too. for example I can't make PLLs as normal and i usually make them with Adding two F2L (it's hard so explain)
if you have a new method or you need some algorithms say me. I'll think about how can I make them!
3 - my program is for 3x3 and 2x2 now and i didn't wrote that for 4x4 or more but it does not take much time so write that for 4x4 because the program is wrote systematic.

if you can't understand my text say me :) because i cant speak very well
 
Joined
Oct 8, 2013
Messages
41
Likes
5
Location
Iran - Mashhad
WCA
2013AHMA01
Thread starter #10
why not put R R' or R2 instead so people actually understand it?
it is not too hard and i already do that for my currently algorithms. but that file is for when i explained my method in an Iranian forum. (Several weeks ago)

it's time to see the power of my program :)
After weeks of trying I made a method for 2x2 first layer.
you can see my method here
give your comments


if you have any problem with it, please say me

what do you think?
is that a useful method?
number of cases is good?
the method is comprehensible? or you need a tutorial!
 
Last edited:
Joined
Dec 18, 2007
Messages
7,830
Likes
33
Location
a <script> tag near you
WCA
2006GOTT01
YouTube
qqwref2
#11
my program is designed by python because I thought that's a fast language.
Python is generally fast to write programs in, but when it is actually executing, it is relatively slow. Of course for 2x2x2 solutions the speed is not important.

it's make all the possible algorithms to solve a special case. we have some database site just like Boca.bee.pl that every body can go there and add a new algorithm. These sites are no longer needed because we can find all the possible algorithms.
We could already find all possible algorithms. Sites like that are useful because they show the best (fastest) algorithms for each case. Can your program also find the best algorithms, or does it only generate huge lists of possible ones?

as you know we have 54 notations so we have 54^^10=... ten move algorithm and it is not possible to make all of them.
Uh what? Even on the 3x3x3 there aren't anywhere near 54 possible moves.

what do you think?
is that a useful method?
number of cases is good?
As far as I can tell your method is something like: (a) set up the four pieces of the first layer into one of the positions you give, (b) solve the first layer using one algorithm, (c) CLL. I am not sure the first layer pieces will always fit into one of the MM/MN/NN configurations. Also, it doesn't make sense to learn 300 algorithms for the first layer when you could just solve it with intuition - or, if you really want to learn algs, just memorize 80 and then use EG, so you only have to make a face instead of a layer. Your PDFs give way too many algs - nobody needs 1000 different algs. People will want one or a few options, at most. It's nice that you finished your program but you have to make the output useful to people or nobody will want to use it.
 
Joined
Oct 8, 2013
Messages
41
Likes
5
Location
Iran - Mashhad
WCA
2013AHMA01
Thread starter #12
Of course for 2x2x2 solutions the speed is not important.
of course it's too important! for example 14 move algorithms take 14000 century time with my computer.

We could already find all possible algorithms.
could you please show me a soft ware or site that do that for me?

Uh what? Even on the 3x3x3 there aren't anywhere near 54 possible moves.
i said 54 Notation
R R' R2 r r' r2 M E S x y z ....
Imagine we want to make all the the 10 move algorithms. so we have 54^^10 difference algorithms. a simple math!!

I am not sure the first layer pieces will always fit into one of the MM/MN/NN configurations.
i can't speak English very well but I'll make a tutorial soon and I'll show you that first layer pieces will always fit into one of my algorithms (some time you should make them with one move)

it doesn't make sense to learn 300 algorithms for the first layer when you could just solve it with intuition
well! you don't need to memorize all of them. you should just learn them! all of them are 1-7 move (one of them is 1 and several 7) so it's not very hard! some people spend too much time in order to solve the first layer in the best way. so anyway it's faster than waste your time.

if you really want to learn algs, just memorize 80 and then use EG, so you only have to make a face instead of a layer.
1 - oh oh oh! we have 120 EG algorithms and they are always longer than my algorithms. if you want to lean 80 of my alg and use EG so you need to learn..
91 (4.5 move) + 80(8.46 move) + 40 (CLL) = 409 + 676 = 1085
but on my method
324 (4.5) + 40 (CLL) = 1458
(4.5 and 8.86 is average of moves)
so it isn't much easier :)
2 - i think the EG method isn't a fast method. because you do first layer false and you should spend an additional time to fix it again!

I am very pleased to hear from you.
If you have any other idea please feel free to say me :)
 
Last edited:
Joined
Dec 18, 2007
Messages
7,830
Likes
33
Location
a <script> tag near you
WCA
2006GOTT01
YouTube
qqwref2
#14
of course it's too important! for example 14 move algorithms take 14000 century time with my computer.
How could it possibly be that much ?_?

could you please show me a soft ware or site that do that for me?
JACube, Cube Explorer, or the solvers here.


i said 54 Notation
R R' R2 r r' r2 M E S x y z ....
Imagine we want to make all the the 10 move algorithms. so we have 54^^10 difference algorithms. a simple math!!
Many of those are equivalent, so there is no need to try them all. You are searching for move sequences, so why try the same thing twice? x y z are just rotations; r f u d l b are equivalent to L B D U R F plus a rotation. So really there are only 27 possible moves, and only 18 if you don't allow slice moves. And there are actually far fewer than 54^10 (NOT 54^^10) sequences to try anyway, since you know you will never use the same type of move twice in a row (for instance R l2 or F F').

i can't speak English very well but I'll make a tutorial soon and I'll show you that first layer pieces will always fit into one of my algorithms (some time you should make them with one move)
OK, what case would you use for this? F R U F2 U F U2 R2 U' F2

1 - oh oh oh! we have 120 EG algorithms and they are always longer than my algorithms. if you want to lean 80 of my alg and use EG so you need to learn..
91 (4.5 move) + 80(8.46 move) + 40 (CLL) = 409 + 676 = 1085
but on my method
324 (4.5) + 40 (CLL) = 1458
(4.5 and 8.86 is average of moves)
so it isn't much easier :)
Your method only applies if you already know CLL, so we can ignore those algs. Thus there are only 80 new algs to learn if you want to do EG. That is 80*8.46=676 moves. (Why would someone learn 91 of your algs if they are learning EG?) But your method is 324*4.5=1458 moves. That looks like a big difference to me. Also, many EG algs are composed of simple sequences such as RUR' and FR'F'R, but your algs are more random, so each move is harder to remember. I would also add that your method has completely new recognition that must also be learned, whereas with EG the hardest part of the recognition is the same as CLL.

2 - i think the EG method isn't a fast method. because you do first layer false and you should spend an additional time to fix it again!
You should tell that to the current WR holder in 2x2x2.
 
Joined
Aug 1, 2011
Messages
1,716
Likes
12
Location
Hungary
WCA
2005HANT01
YouTube
Renslay
#15
JACube, Cube Explorer, or the solvers here.
I believed that Ron's Mini Cube Solver is an optimal solver, but this assuption seems false.

For this scramble:
R F2 R' F U2 F R U2 F' U2 z' (10,13)
if I say "max 11 moves", it gives
F U F2 U2 F U' F2 U R2 F' R2 (11,16)
and if I say "max 10 moves", then it gives the optimal solution (because for "max 9 moves" it gives nothing).

Does anyone know why it works in this way? Of course I can find optimal solutions, but why with multiple runs...

EDIT: okay, I think I know the answer. It searches for solutions lexicographically ordered, that's why it found the longer algorithm first.
 
Last edited:
Joined
Oct 8, 2013
Messages
41
Likes
5
Location
Iran - Mashhad
WCA
2013AHMA01
Thread starter #17
it's time to see the power of my program :)
After weeks of trying I made a method for 2x2 first layer.
you can see my method here
give your comments
well! here's some example solve.(green of front and yellow on top)

Scramble: D B' D' L2 U' R2 B' D2 L2 B'
Build a case: x' y2 (DBL is fix so we want to make the white face)(it's a MM case)
solution: r' u' r u f'

Scramble: F U2 R2 U2 R D' L' D' F L
Build a case: x R (DBL is fix so we want to make the white face)(it's a MM case)
solution: f' u' f2 u2 r2

Scramble: F' D2 L' B D2 R D2 R B2 R
Build a case: y x' U (DBL is fix so we want to make the yellow face)(it's a MM case)
solution: d f2 r f' d2 r

if you think some scrambles can't be solve say me.

Your method only applies if you already know CLL, so we can ignore those algs. Thus there are only 80 new algs to learn if you want to do EG. That is 80*8.46=676 moves. (Why would someone learn 91 of your algs if they are learning EG?) But your method is 324*4.5=1458 moves. That looks like a big difference to me. Also, many EG algs are composed of simple sequences such as RUR' and FR'F'R, but your algs are more random, so each move is harder to remember. I would also add that your method has completely new recognition that must also be learned, whereas with EG the hardest part of the recognition is the same as CLL.
in my method i make first layer and them we just need a CLL.
i can change my algorithms to make just one face (not one layer) so they will be 324/3=108 cases.
so in order to that we have two way to solve the cube

1 - make one face with one of 108 algs and then solve the cube with one EG so
108 (4.5) + 120 (8.46) = 486 + 1015 = 1501

2 - make one layer with one of 324 algs and then solve the cube with one CLL so
324 (4.5) + 40 (8.46) = 1458 + 338 = 1796

be careful because my method is on first layer so you should calculate some thing for first layer or face.

You should tell that to the current WR holder in 2x2x2.
of course EG is a really fast method!!! I just mean my method may be better because:
in EG we just make one face with irregular sides and then we use some longer algorithms to fix it again.
but in my method we don't need to make that irregular. so we don't use longer algorithms.

Meanwhile, as I told you, I can change my algs to use it for EG ( same 108 cases ) if you don't like to use my method with CLL, I can provide that in EG method for you.
 
Last edited by a moderator:
Joined
May 7, 2006
Messages
7,287
Likes
40
WCA
2003POCH01
YouTube
StefanPochmann
#18
well! here's some example solve.(green of front and yellow on top)

Scramble: D B' D' L2 U' R2 B' D2 L2 B'
Build a case: x' y2 (DBL is fix so we want to make the white face)(it's a MM case)
solution: r' u' r u f'

Scramble: F U2 R2 U2 R D' L' D' F L
Build a case: x R (DBL is fix so we want to make the white face)(it's a MM case)
solution: f' u' f2 u2 r2

Scramble: F' D2 L' B D2 R D2 R B2 R
Build a case: y x' U (DBL is fix so we want to make the yellow face)(it's a MM case)
solution: d f2 r f' d2 r

if you think some scrambles can't be solve say me.
None of those examples work. I suggest to use http://alg.garron.us/ to check before you keep wasting everybody's time. Also, what's up with the lower case letters?
 
Joined
Aug 1, 2011
Messages
1,716
Likes
12
Location
Hungary
WCA
2005HANT01
YouTube
Renslay
#19
None of those examples work. I suggest to use http://alg.garron.us/ to check before you keep wasting everybody's time. Also, what's up with the lower case letters?
They work for me. He just solve the first layer, as he mentioned (white / yellow).
Second example
Although I don't understand much more... Why is the R in the the "Build a case" phase?

of course EG is a really fast method!!! I just mean my method may be better because:
in EG we just make one face with irregular sides and then we use some longer algorithms to fix it again.
but in my method we don't need to make that irregular. so we don't use longer algorithms.


Meanwhile, as I told you, I can change my algs to use it for EG ( same 108 cases ) if you don't like to use my method with CLL, I can provide that in EG method for you.
That doesn't make any sense...
 
Last edited:
Joined
Oct 8, 2013
Messages
41
Likes
5
Location
Iran - Mashhad
WCA
2013AHMA01
Thread starter #20
None of those examples work. I suggest to use http://alg.garron.us/ to check before you keep wasting everybody's time. Also, what's up with the lower case letters?
ehhhh! I'm sorry if i wasted your time :/ but I checked all of them. all of them are true in order to this Notations


### Notice that this these are just solution of first layer!! because my method was on first layer!

for Example for the first one we do this
D B' D' L2 U' R2 B' D2 L2 B' x' y2 R' U' R U F' (The white was correct)
SECOND
F U2 R2 U2 R D' L' D' F L x R F' U' F2 U2 R2 (The White was correct)
THIRD
F' D2 L' B D2 R D2 R B2 R y x' U D F2 R F' D2 R (The yellow was correct)

please check it again!!

That doesn't make any sense...
how do you do the first face? anyway it's better than nothing!
 
Last edited by a moderator:
Top