• 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 40,000+ people from around the world today!

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

Converting algorithms to letters?

j45th

Member
Joined
Jan 3, 2019
Messages
5
I've been thinking about this quite a bit while learning zbll, and I think it could be extremely useful in memorizing large algorithm sets with a memory palace. My original thought was this: Give each trigger a letter and then just memorize the first two triggers and try the rest. The obvious problems with this: too many triggers, doesn't account for non triggers mid alg, like U2 etc, and it doesn't account for the entire algorithm. I will continue to work on this but I would love to hear the ideas of others, my goal is to engineer a system where an algorithm can be reduced into a visual of one object(or a set of object working together, for example, "a dog eating meat").

Comment your ideas, no matter how complex and far fetched they sound!
 

abunickabhi

Member
Joined
Jan 9, 2014
Messages
6,682
Location
Yo
WCA
2013GHOD01
YouTube
Visit Channel
Already been working on it


David Singmaster notation is good to generate scrambles and stuff, but not good enough for memorization.

Edit:
I have also added ways to memorize wide moves and rotations.

For wide moves,
u - QA
u' - BP
u2 - RC
f - SV
f' - WT
f2 - XU
l' - NH
l - MG
l2 - OI
r - NJ
r' - MK
r2 - OL

x - li
x' - mi
y - gi
y' - kj
z - qp
z' -dp
(I chose these letter pairs as these were rarer in the algorithm string, and had strong imagery in my letter pair scheme)

For normal face turning moves,
U - A
U' - B
U2 - C
D - D
D' - E
D2 - F
L - G
L' - H
L2 - I (Letter i)
M - M
M' - N
M2 - O
R - J
R' - K
R2 - L
S - S
S' - T
S2 - U
E - P
E' - Q
E2 - R
F - V
F' - W
F2 - X
B - Y
B' - Z

There are also an insane amount of cancellations I have come up with, but it will be pretty advanced for you ig.
This system takes time to get accustomed to.

For fun reasons, I have decided to name this system as Yo notation!

Happy memorizing...
(P.S. I have not used this system to memorize ZBLL, as I use Roux method to solve a 3x3)

If you have any further doubts or need help or clarification, drop a mail to [email protected]
I am always open to debates yo.

Lol, nice read.

Also, I think I am not the first person to come up with this system.

But, I have invested about 1 year to perfectly engineer this system and make it work.
I will not publish the 10,000ish algs I have converted to images using the Yo notation anytime soon.

But, I will be posting videos of execution of this triggerless algs (length ~12 moves) on this channel:
https://www.youtube.com/channel/UCa7dTclUqnR9VwSeOCpRwAQ
 
Last edited:

CraZZ CFOP

Member
Joined
Jul 30, 2018
Messages
95
WCA
2018HUGE02
It would be a really good idea for TSLE memo.

RUR'=A
RU2R'=B
RU'R'=C
U=X
U2=Y
U'=Z
So the 4 corners alg for TSLE would be: XCYA
and RUR'U2RUR' for one of the edge in slot cases would become AYA
 
Last edited:

j45th

Member
Joined
Jan 3, 2019
Messages
5
This system takes time to get accustomed to.
Are you accustomed to this system? Also I do think it is a very interesting system, however very long. For example, if im learning zbll, there is only a very limited amount if turns ( L, R, R2, etc) that are actually used in algorithms. I'm currently working on something very similar to yours, but it includes both common triggers, and common transition moves. But thats where my system lacks to yours, mine doesn't account for certain moves, ei S, E, etc so it cannot be directly applied to blind comms. I think it could be easily modified however. I also think one problem with that system would be how long your visual memo is. In mine it includes both triggers and turns, to minimize visual length. While I do think yours is definitely a great option once it is concreted into your mind, my biggest problem would be trying to fit an entire 12 alg set into one room.
 

j45th

Member
Joined
Jan 3, 2019
Messages
5
In fact I make mistakes in turning (like R instead of R' in normal notation), but in this I do not make any mistake.

You should make a program that converts typical algorithms to your system. I'd also be interested in seeing how many algs you can learn in day(or a speed test of the sort).
 

Tao Yu

Member
Joined
Jun 29, 2011
Messages
1,172
Location
Ireland
WCA
2012YUTA01
YouTube
Visit Channel
For something as "small" as ZBLL I feel that you shouldn't worry about getting every part of an alg precisely encoded into an image. As long as you train your algs regularly, your brain should be able to fill in a lot of the gaps. If I were you, I'd just leave out all of the non trigger moves. Something that simply jogs your memory should be enough in my opinion.

When I learned ZBLL, I never found it hard to memorize an alg. I had a lot more trouble remembering which alg corresponded to which case. I think that some way of mapping all 493 cases to a unique image (which you could then connect to an image for the alg) would be more useful, although I've never been able to think of a good system for this.

For something like 5-style, which I know Abhijeet has been developing, a letter based mnemonic system makes more sense because there's an easy way to encode all cases to four letters, and there are far more algs, making it harder for your brain to fill in missing information.
 
Last edited:

abunickabhi

Member
Joined
Jan 9, 2014
Messages
6,682
Location
Yo
WCA
2013GHOD01
YouTube
Visit Channel
For something as "small" as ZBLL I feel that you shouldn't worry about getting every part of an alg precisely encoded into an image. As long as you train your algs regularly, your brain should be able to fill in a lot of the gaps. If I were you, I'd just leave out all of the non trigger moves. Something that simply jogs your memory should be enough in my opinion.

When I learned ZBLL, I never found it hard to memorize an alg. I had a lot more trouble remembering which alg corresponded to which case. I think that some way of mapping all 493 cases to a unique image (which you could then connect to an image for the alg) would be more useful, although I've never been able to think of a good system for this.

For something like 5-style, which I know Abhijeet has been developing, a letter based mnemonic system makes more sense because there's an easy way to encode all cases to four letters, and there are far more algs, making it harder for your brain to fill in missing information.


I agree on Tao's point.
For ZBLL, the only tough point is to find that mapping and recognition which can be easily practiced by alg trainer.

There are some reductions in the letter based mnemonic system, when the algset only consists of [R U D] like say in UFR corner algs, but it easier just to remember the commutator and how it works, rather than develop a system to memorize it.
If a system with same triggers are translated into letter based mnemonic system, the algs look somewhat like: https://docs.google.com/spreadsheets/d/1XiNdbdbj14t6p-38t_IfRyLEyjYZzR6D1M16dTZudpM/edit?usp=sharing
Remembering like this is not a pleasure-able experience and it is better to understand how the commutator works.

For ZBLL, some kind of TeamBlind jargon to remember triggers effectively is good enough to make algs stick when we start first learning them.
 

j45th

Member
Joined
Jan 3, 2019
Messages
5
For something as "small" as ZBLL I feel that you shouldn't worry about getting every part of an alg precisely encoded into an image. As long as you train your algs regularly, your brain should be able to fill in a lot of the gaps. If I were you, I'd just leave out all of the non trigger moves. Something that simply jogs your memory should be enough in my opinion.

Yeah, this is kind of what I was after.
I haven't tried just leaving out the non trigger moves, but that makes a lot of sense now that I think about it.
My "plan" once I find a good system is to be able to map out full 1lll, so I'm sure as long as I'm not learning algs insanely fast, this would work.
Thank you!
 

j45th

Member
Joined
Jan 3, 2019
Messages
5
There are some reductions in the letter based mnemonic system, when the algset only consists of [R U D] like say in UFR corner algs, but it easier just to remember the commutator and how it works, rather than develop a system to memorize it.
If a system with same triggers are translated into letter based mnemonic system, the algs look somewhat like: https://docs.google.com/spreadsheets/d/1XiNdbdbj14t6p-38t_IfRyLEyjYZzR6D1M16dTZudpM/edit?usp=sharing
Remembering like this is not a pleasure-able experience and it is better to understand how the commutator works.
Very interesting to look at the translation of these, and I definitely agree that it wouldn't be useful on 3 style, but something like this, plus mapping technique for zbll? Sounds useful.
 

kubnintadni

Member
Joined
Jan 4, 2018
Messages
71
I've been thinking about this quite a bit while learning zbll, and I think it could be extremely useful in memorizing large algorithm sets with a memory palace. My original thought was this: Give each trigger a letter and then just memorize the first two triggers and try the rest. The obvious problems with this: too many triggers, doesn't account for non triggers mid alg, like U2 etc, and it doesn't account for the entire algorithm. I will continue to work on this but I would love to hear the ideas of others, my goal is to engineer a system where an algorithm can be reduced into a visual of one object(or a set of object working together, for example, "a dog eating meat").

Comment your ideas, no matter how complex and far fetched they sound!

Already been working on it

David Singmaster notation is good to generate scrambles and stuff, but not good enough for memorization.

[...]

There are also an insane amount of cancellations I have come up with, but it will be pretty advanced for you ig.
This system takes time to get accustomed to.

For fun reasons, I have decided to name this system as Yo notation!

Happy memorizing...
(P.S. I have not used this system to memorize ZBLL, as I use Roux method to solve a 3x3)

[...]

I've also been working at a system like these for a while. It's almost finished, so I'll probably be cross-posting it here and at artofmemory within a week. In my first attempt (this is my third) I went with something similar to Yo notation (though I was not aware of Yo Notation at the time), but abandoned it after making some examples and seeing many instances of repeat images. After some fiddling I decided that I was never going to avoid repeat images, since cubing algorithms naturally end up with similar move sequences, and so I decided to go the opposite direction and see if I could encourage repeat images to show up <i>on purpose</i> whenever the move sequence meant the same thing. This may well have been a mistake, but I'm still excited to try it out (though that won't happen for a while after I finish it since it uses my number system... which I have barely started filling out. :/

I would be interested in how @abunickabhi deals with heavily repeating images given as Yo Notation will generate a fair number of them, and he is tackling large algsets. My approach was to embrace it and hope its not too much of a problem, but a more general way to deal with it would be awesome.

Or if anyone has come up with a system that somehow minimizes repeat images. The optimum would be applying a reversible hash function (Is that a contradiction in terms? I genuinely don't know.) to the alg and then memorizing the result, which should have each image in your system showing up with roughly the same frequency, and then translate back. Of course, that is wildly impractical, so if anyone has found a way to get some of the benefit of a hash function like that while still being able to be translated into cube moves instantly by humans, that would be a really cool approach, too.
 

abunickabhi

Member
Joined
Jan 9, 2014
Messages
6,682
Location
Yo
WCA
2013GHOD01
YouTube
Visit Channel
I've also been working at a system like these for a while. It's almost finished, so I'll probably be cross-posting it here and at artofmemory within a week. In my first attempt (this is my third) I went with something similar to Yo notation (though I was not aware of Yo Notation at the time), but abandoned it after making some examples and seeing many instances of repeat images. After some fiddling I decided that I was never going to avoid repeat images, since cubing algorithms naturally end up with similar move sequences, and so I decided to go the opposite direction and see if I could encourage repeat images to show up <i>on purpose</i> whenever the move sequence meant the same thing. This may well have been a mistake, but I'm still excited to try it out (though that won't happen for a while after I finish it since it uses my number system... which I have barely started filling out. :/

I would be interested in how @abunickabhi deals with heavily repeating images given as Yo Notation will generate a fair number of them, and he is tackling large algsets. My approach was to embrace it and hope its not too much of a problem, but a more general way to deal with it would be awesome.

Or if anyone has come up with a system that somehow minimizes repeat images. The optimum would be applying a reversible hash function (Is that a contradiction in terms? I genuinely don't know.) to the alg and then memorizing the result, which should have each image in your system showing up with roughly the same frequency, and then translate back. Of course, that is wildly impractical, so if anyone has found a way to get some of the benefit of a hash function like that while still being able to be translated into cube moves instantly by humans, that would be a really cool approach, too.

The reversible Hash function, that's a good one.
Actually, in Yo notation, images do not repeat at all, in fact, I have created here is a library of 150,000 images that I have made (26 English alphabets, so ~26x25x24x23 which is 200,000 more than my number, but I weed out the cases that are very rare.)

There is no need of getting all complicated about repeated moves causing repeated images. Like for the example of UFR corner algs which is about 380, I have smartly tried to make each one of the alg into Yo notation and then having a unique set of letters resulting in newer images.

I will be making a separate video on the channel on how to do this smart encoding.
If you want to collaborate with me, you can drop an email to [email protected]

ufralgyo.png
 

kubnintadni

Member
Joined
Jan 4, 2018
Messages
71
The reversible Hash function, that's a good one.
Actually, in Yo notation, images do not repeat at all, in fact, I have created here is a library of 150,000 images that I have made (26 English alphabets, so ~26x25x24x23 which is 200,000 more than my number, but I weed out the cases that are very rare.)

There is no need of getting all complicated about repeated moves causing repeated images. Like for the example of UFR corner algs which is about 380, I have smartly tried to make each one of the alg into Yo notation and then having a unique set of letters resulting in newer images.

I will be making a separate video on the channel on how to do this smart encoding.
If you want to collaborate with me, you can drop an email to [email protected]
Are you using a 4 digit letter system of some sort (I feel like I saw a post of yours talking about something like that?), or a 2 digit with 2 images per loci? If it's 4 digits, I could see how that might avoid repeats since 8 moves is longer than most repetitive sequences.
 
Top