Memory Methods

[LaffyTaffyKidd]

ty hait, but do u think that hexadecimals could work for corner/edge permutations?

Nevermind.. thats a really retarded question! (it won't work obviously)

 

[clincher]

How would that system work for corner orientation? I like that hexadecimals method but my BLD method doesn't have an EO step someone reply please!!

 

[Mike Hughey]

How would that system work for corner orientation? I like that hexadecimals method but my BLD method doesn't have an EO step someone reply please!!

Even though it's probably slower than pure visual, I still use it for corner orientation.

I use 4 digits. The way I do it, the first (most significant) digit is for bottom corners that need to go counterclockwise, the second digit is for top corners that need to go counterclockwise, the third digit is for bottom corners that need to go clockwise, and the fourth digit is for top corners that need to go clockwise. The first (most significant) bit in any given digit is for the corner in the BL position of that particular layer, the second bit is BR, the third bit is FR, and the fourth is FL. You can arrange them any way you want, but this works well for me.

So if you had UFL UFR DFR DBR DBL needing to go counterclockwise and UBL DFL needing to go clockwise, you'd have E318. I build it by doing all the pieces from one layer first, then the other. It may seem strange, but it works well for me. So I'd look at the top, see the 3 pieces that need rotating, and say to myself "0308". Then I'd look at the bottom and fill in the rest, "E318". Then I memorize those 4 digits and I'm good to go. Nice thing about it, if you can "see" hexadecimal well, is you can count the bits in the first 2 digits and the bits in the last 2 digits, and if they don't differ by a multiple of 3, you know you messed up. I had to work with hexadecimal A LOT a number of years ago, and so it's pretty instantaneous for me to convert from bits to hexadecimal and back, which is why it works so well for me.

 

[LarsN]

CO, CP and EO are all visual for me, but I've been using a vocal method for EP. All edges are given both a consonant and a vowel. 12 consonants (BDFKLMNPRSTX) and 11 vowels (the starting edge only needs a consonant). Unfortunatly there is only 6 vowels in english (9 in danish) but for the other edges I use a vowel combo (IO, AI). Then using 3-cycle you come up with something like this: BUDALIT(first cycle), KIOR(second cycle). Just repeat these words out loud.That way I find the vocal sounds much easier to remember and you know that a cycle of even egdes will always end with a vowel.
Sometimes you even get real word. Very easy to remember.

 

[Jack]

Okay, I have a question about memorizing. I tried timing my edges using 3 cycle compared to edges using M2, and M2 was about twice as fast. I am getting faster at memorizing, but my solving usually takes around twice as long as memorization using 3 cycle. So I want to switch to M2, but I don't know any way to memorize M2 other than with colours (where you take the 2 colours of an edge and turn it into an image). Right now I memorize visual for EO, CO, and CP then remember EP with letter pairs (how most people remember big cubes). I am really slow at memorizing with colours, so does anyone know a method using letter pairs that could work for remembering orientation and permutation at the same time? Thanks!

 

[Mike Hughey]

Yes - I just realized this as I was playing with Chris Hardwick's all-commutator method for 3x3x3. (I was wondering the same thing as you until I realized this last night.) You can use a method like is typically used for big cubes BLD. Assign letters to each cubie face on the cube. So for edges, you'd have 24 - A-X. Now, just memorize your first face (I always start with A), and then see where that cubie face needs to go. And so on. When you're done, you'll have only memorized about half of the letters. I just started playing with this, so I'm not sure, but it looks like you do have to pay attention to the end of a cycle, wrapping back to the first one, in case you have a piece flip to account for at the end.

The same can be done for corners. Again 24 letters for the 8 corner pieces.

Really pretty neat - I don't know why I never saw that before - it seems so obvious now.

 

[tim]

The same can be done for corners. Again 24 letters for the 8 corner pieces.

If you use a fixed buffer (like in M2), you only need 21 letters for 7 corners and 22 letters for 11 edges.

 

[Kenneth]

I'm trying to learn a method where:

----------
Obsolete, look here for the changes I have made for this part.

RUF and RDF are R-corners, RU, RF and RD are R-edges (the colour of the R-side)
FUL and FDL are F-corners, FU, FL and FD are F-edges
LUB and LDB are L-corners, LU, LB and LD are L-edges
BUR and BDR are B-corners, BU, BR and BD are B-edges
------------

If the peice shall go to the U-layer, it is a T-piece (top), if it is going to the D-layer it's a D-peice and for edges there are also M-pieces for the middle layer ones.

Here I'm using "T" instead of "U" and that because T is a consonant, U is not and that is important as you wil see below (the rest of the faces has got conconant letters so those are OK)

OK, that's for permutation but I also memorise orientation and for that i use three wovels, A, E and O.

----------
Obsolete, look here for the changes I have made for this part.
A is for oriented peices, E is for clockvise corners and O is for counter clockvise corners and unoriented edges.
----------

The wovels I put in the midle of the "pice charstring" I use. A peice can for example be "BET", "RAM" or "ROD".

Then, to memorise the small proto words I get from that I use normal words (mostly Sweidish ones =), Some are as you can see above already words like that but some are not. "LAM" makes for example "LAME" and so on.

So, you remember normal words, but you can directly from the word "decompile" the positioning and orientation needed for the peice.

 

[philkt731]

How would that system work for corner orientation? I like that hexadecimals method but my BLD method doesn't have an EO step someone reply please!!

Even though it's probably slower than pure visual, I still use it for corner orientation.

I use 4 digits. The way I do it, the first (most significant) digit is for bottom corners that need to go counterclockwise, the second digit is for top corners that need to go counterclockwise, the third digit is for bottom corners that need to go clockwise, and the fourth digit is for top corners that need to go clockwise. The first (most significant) bit in any given digit is for the corner in the BL position of that particular layer, the second bit is BR, the third bit is FR, and the fourth is FL. You can arrange them any way you want, but this works well for me.

So if you had UFL UFR DFR DBR DBL needing to go counterclockwise and UBL DFL needing to go clockwise, you'd have E318. I build it by doing all the pieces from one layer first, then the other. It may seem strange, but it works well for me. So I'd look at the top, see the 3 pieces that need rotating, and say to myself "0308". Then I'd look at the bottom and fill in the rest, "E318". Then I memorize those 4 digits and I'm good to go. Nice thing about it, if you can "see" hexadecimal well, is you can count the bits in the first 2 digits and the bits in the last 2 digits, and if they don't differ by a multiple of 3, you know you messed up. I had to work with hexadecimal A LOT a number of years ago, and so it's pretty instantaneous for me to convert from bits to hexadecimal and back, which is why it works so well for me.

First, why do u switch between layers for this number? Also, how (and how fast) do you translate this when youo are executing? Do u usually solve two at a time, or what?

 

[Mike Hughey]

First, why do u switch between layers for this number? Also, how (and how fast) do you translate this when youo are executing? Do u usually solve two at a time, or what?

I guess all I'm trying to accomplish with the order of the digits is to keep the counterclockwise pieces grouped together, and the clockwise pieces grouped together. That way, I can immediately see if it's an impossible memorization or not (has to be a difference of a multiple of 3 between the number of bits in the two halves of the number). There's no real reason why you have to memorize in this order, though - if you decide to try this, I recommend you use whatever order works best for you.

I'm really fast with hexadecimal - that comes from having done some machine language coding back in the 80's writing some video games for the TRS-80 Color Computer. So translating back and forth was always very fast for me. But really, eventually it just becomes a pattern that you "see", so I don't think my hexadecimal experience means all that much any more. I probably take about 20 seconds on average to do corner orientation in the execution phase, but keep in mind that I'm not that fast at BLD - I typically run 3:00 to 3:30 for an overall solve, of which 2 minutes is execution, so 20 seconds for CO is really not bad, relatively, for me.

As for how I do it, I prefer to do them in 3's, if that's possible. If I have 3 CCW and 3 CW, I'll almost always do the CCW first, then the CW. Also, if I have 4 CCW and 1 CW, I'll do the easiest 3 CCW first, then do the other 2. But if I have 4 CCW and 4 CW, if it's reasonable I'll try to do them 4 at a time by doing one layer at a time. (But if they're grouped 3 on a layer, I'll still do 3 at a time, then the last 2.) I guess it's a little chaotic, but it works for me.

 

[philkt731]

ok i have another question, hen theres 1 corner on top that needs to go CCW, and two adjacent on the bottom that need to go CCw, does anyone ever AUF and then rotate the cube to do all three at the same time, or what else would you do?
Or if you needed to do 1 5 7 CCW, what would you do?

Also , for EO, you could do something like Mike's hexadecimal and have a three digit number. The first one for the top layer with order of significance, B, L, R, F. The second digit for the E edges, with order BL BR FR FL. The thrid digit being the bottom layer, with order, B, L, R ,F. So F39 would be the same as (1 2 3 4 5 8 9 11) (using Macky's number systme)

Lastly, I've been testing a Roman rooms method for CP, replacing the rooms with rooms/ objects in the upstairs of my house (ie. my sis's room, picture of my mom, stairs, office, my room, light switch, parents room, bathroom), but how can I do a good method for EP, since Roman rooms usually only has 8, and anyway my upstairs only has those 8 things, and I probably shoulldn't use the same setting as CP. What do you guys use, who don't use numbers, or transate the numbers to something else easier to memorize?

 

[AvGalen]

If 1 5 7 = UFL DFL DBR I would just do F'

For Roman Rooms EP (12 edges, 8 locations) you can either use pairs (2 edges in 1 location) or use the floor and ceiling of the room. Each room has 8 corners and 6 middle spots on the walls(4)/floor(1)/ceiling(1)

 

[magicsquares]

I'm sorry but can anyone help me on how Roman Rooms are actually used for blindfold cubing? I don't know how you use those locations... For example, how do you memorize that this edge goes there.. An explanation like that... Sorry for the dumb question.

 

[Mike Hughey]

I use Roman Rooms for big cubes only; I just use visual memorization for 3x3x3 permutations. For big cubes, I use letters A-X (I guess you'd only need A-L for 3x3x3) and put them in pairs, like Chris Hardwick does (see http://www.speedcubing.com/chris/memorizing.html), and then combine several of those images for a single location in my room, interacting with the object inside that room. For instance, I use my living room (I'm a keyboardist, as you will be able to tell), so I have spots in the room corresponding to my synthesizer stack, organ, and piano as three of the locations. Then I can imagine the various images somehow interacting with the items. For instance, this morning I did one where I had Atom Ant and Barney Rubble playing a duet on the synths (A and B, a 2-cycle), and I also had Fred Flintstone sitting on the bench of the piano but facing away from it playing a QiX videogame machine with a ROdent named Linus (F Q X R O L - I got 6 in a single location). By combining several items at each location, I can usually fit a whole 4x4x4 (except corners) in a single Roman Room. I have a couple of other locations in the adjacent room in our house (our dining room) as extras in case I run out. I guess I'm actually using a combination of Roman Rooms and journeys for this (which I suspect anyone does who does this), because I have a standard "journey" through my room that I always use. (I guess I should mention that I've only been doing this for a few days now, but I've probably used it on 20 solve attempts at this point.)

I would think the Roman Rooms method would be great for MultiBLD; I'll probably give that a try sometime after the Virginia Open. (I gave away too many of my cubes about a month ago, so I didn't have enough 3x3x3's to try MultiBLD until very recently.) For MultiBLD, I'd want to be able to store the whole thing, so I'd probably put my CO hex at one location, EO hex at another location, and then CP would probably take 2 locations or so and EP 3, on average. So one room should be plenty for a single cube. I'll have to make more rooms to do this, though - so far I've been going with just 2 (since that's all I need for a 5x5x5).

Oh, and by the way, I do EO exactly like philkt731 suggested. And I would do the same as Arnaud for 1 5 7 CCW, except that I would actually do B' instead of F' because I do my orientations with U' R' U R U' R' U R instead of doing them on the top or bottom. R and U edges are fastest for orienting corners for me.

 

[magicsquares]

Ahh.. So Roman Rooms is just like a journey but it all happens in a single room or something like that. Did I get it right?

 

[AvGalen]

http://www.mindtools.com/pages/article/newTIM_06.htm

There is no need to restrict this information to rooms: you could use a landscape or a town you know well, and populate it with memory images.

 

[Mike Hughey]

Another quote from the same article:

The Roman Room technique is most effective for storing lists of unlinked information, while the journey method is better for storing lists of ordered items.

Technically, the Roman Room technique is not really appropriate for things like memorizing a cube. But by performing a journey in the room, it becomes the journey method, and is therefore quite useful. As it turns out, using a Roman Room for a journey is a pretty good technique.

I find the Roman Room approach useful because it compartmentalizes the journey rather nicely. You can move on to a different room for the next type of pieces, for instance. (Or perhaps the next cube, for MultiBLD.)

 

[isaganiesteron]

what i dont seem to get and i have not found any information about is memorizing using pochmanns methods.. more on the edge permutation, because when the edges change orientation while permuting them, how can you memorize this?

 

[masterofthebass]

I've said this a bunch elsewhere, but I doubt you could find it from the search. I started BLD with Pochmann's old method and couldn't think of a memo system. I finally just practiced with color pairs enough that I was able to do it. I just memorize white-red, blue-yellow, etc. It's not a great memo-system, but it works for 1 cube pretty reliably. Another option is to assign each of the 12 edges 2 images or letters, one for each orientation. You would need to have 24 letters or images to make a story with. As long as you have 24 things to memorize, you can theoretically use anything. Any of these methods can work for bigger cubes while using r2.

 

[Marcell]

I learned it like this: Assign words to the cubies using the first two letters of the colors on each cubie, like White-Red = WR = war(Or: world record ), RedWhite = RW = row... or whatever you like. Use these words to make a story in your mind. This is actually very close to the images variant.