• 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!

[Help Thread] Human Thistlethwaite Help and Discussion

mr. giggums

Member
Joined
Feb 14, 2010
Messages
394
Location
Batavia, Northern Illinois
WCA
2011BERN02
Do robots solve the cube with HTA? If they do solve the cube with this "method", does that mean we are learning how to...
think like a robot? 0__0
(Just kidding).

anyway, what is speed HTA?

No robots (the good one's at least) solve with only two steps instead of four. The steps are reducing to <U,D,F2,B2,R2,L2> and then solving the rest.
See here for more details.
 
H

Hershey

Guest
Oh fail! I meant the Thistlethwaite algorithm, not Human Thistlethwaite Algorithm.

Just a wild guess: Solving a cube with HTA as fast as you can?

How is speed HTA different from regular HTA? Are there less steps? If there are, what are the steps used in speed HTA?
 
Last edited by a moderator:

nigtv

Member
Joined
Jul 13, 2009
Messages
82
Basically following the Heise HTA approach but using 2xOLL for orienting corners.
I think 60 HTM is "fair" for this approach
Just using intuitive algs (I didn't know how to build commutators at this point), if I focused on using the least amount of turns in each step, but focusing only step by step, I was almost always using 54htm.

HTA looked interesting so I tried it for a while. Never got fast or super low turn counts, but it was fun from an abstract point of view. Good primer on basic group theory with cubes.
 

Cubenovice

Forever Slow
Joined
Feb 6, 2010
Messages
2,291
Location
Vlaams-Brabant (Belgium)
How is speed HTA different from regular HTA? Are there less steps? If there are, what are the steps used in speed HTA?

Steps will be the same but using large alg sets for the various steps.
Orienting corners: normally done per layer
With a 200+ alg set you can orient all corners at once (algs must keep EO and E slice edges must stay in E)
Alg length 6 - 10 HTM (or 11, don't have the list on this PC; will edit later)

Fix bad edges:
Alg set vs fully intuitive

End game:
Alg set vs semi intuitive (how intuitive are edge commutators?)


Just using intuitive algs (I didn't know how to build commutators at this point), if I focused on using the least amount of turns in each step, but focusing only step by step, I was almost always using 54htm.

HTA looked interesting so I tried it for a while. Never got fast or super low turn counts, but it was fun from an abstract point of view. Good primer on basic group theory with cubes.

I never solve HTA for speed and will admit that my low movecount solutions posted in this thread were not done linearly.
I find HTA a nice method for "puzzling", trying different options for the various steps to try to come to a short solution.
 

Cubenovice

Forever Slow
Joined
Feb 6, 2010
Messages
2,291
Location
Vlaams-Brabant (Belgium)
fmc.mustcube.net 353:

Scramble: B' L D2 F' B R D R L' B2 L D2 B2 R' U' L2 R' D2 L' D' U B' D U' B2 R2 L U2 R U2

F2 D' F' (EO)
R F2 D' R L' D L' (edges and 4 corners)
F U2 F2 U' F2 U' F2 U2 F (well know OLL to finish corners)
U F2 U' L2 U' . (separate U/D corners)
F2 U F2 U2 L2 U L2 (pair corners with ortega)
L2 U2 D2 R2 (4 edges)
R2 L2 U2 R2 L2 D (4 edges) leaves U perm

At . insert F2 L’ R U2 R’ L (no cancellations but much better than U perm)
(I left in all the cancellations for clarification)

Final with cancellations:
F2 D' F' R F2 D' R L' D L' F U2 F2 U' F2 U' F2 U2 F U F2 U' L2 U' F2 L’ R U2 R’ L F2 U F2 U2 L2 U’ D2 L2 U2 R2 L2 D = 42 HTM

Decided to ignore the bad edges...
 

Cubenovice

Forever Slow
Joined
Feb 6, 2010
Messages
2,291
Location
Vlaams-Brabant (Belgium)
Weekly comp 29

(inverse scramble) L B U2 D L2 U' R2 F' R U' F2 U R F B' R2 F' L

D' F2 L' B' (EO, note the F2 insertion) 4
R D' B2 L D2 B2 (solve U/D edges and complete a U/D face) 10
B' D2 B D B' D B (this is why it pays off to know your Sune's) 16
B2 L2 B2 D (pair up and align corners) 19
R2 D L2 R2 U' R2 (fix bad edges) 25
R' L B2 R L' D2 (edge cycle) 30
D2 F2 U' D' R2 U D F2 (swap 2 edge pairs)

Final solution for inverse scramble:
D' F2 L' B' R D' B2 L D2 B D2 B D B' D B' L2 B2 D R2 D L2 R2 U' R L B2 R L' F2 U' D' R2 U D F2 = 36 HTM

OK, I admit:
my initial solution was 40 HTM because of a sub optimal edge pair swap...

Atilla Horvath was kind enough to open my eyes :tu
 

Cubenovice

Forever Slow
Joined
Feb 6, 2010
Messages
2,291
Location
Vlaams-Brabant (Belgium)
THX for posting, it is nice to see that HTA is getting more attention.

Some tips for your next video:
- clearly explain the consecutive steps in HTA
EO
solve UD cross
orient corners (using Heises two-twist and or regular OLL)
separate U/D corners (intuitive)
pair up corners (typically with Ortega)
fix bad edges (explain what bad edges are and how to solve)
solve edges (possible just using 180° turns but not requirted)

- you should consider solving in a more common cube orientation, at least use white or yellow on top
- even Heise mentions that it is OK to use shortcuts in the final step
- use rehearsed solves so you can focus on teaching instead of solving (prepare your story)
- cover various cases of EO and bad edge fix

That H perm made me cringe though...
an M2 set up into an alg that starts with M2?
 

Cubenovice

Forever Slow
Joined
Feb 6, 2010
Messages
2,291
Location
Vlaams-Brabant (Belgium)
More example solves, enjoy:

Example solve thread #1170
scramble F U2 B L2 D2 L2 F R2 F D2 F2 D' R2 B2 L' D U B U2 R' F'

D U' L' B (EO) 4
R' D' L' (U/D edges and one face) 7
L' D2 L D L' D L' (Sune competes 2 faces) 13
U' L2 (separate U/D corners) 15
F D' F R2 F' D F' (pair up corners) 22
L2 R2 D L2 R2 (fix bad edges, leave L7E) 27
R2 B2 R2 B2 R2 B2 (3 edges) 31
L2 D2 R' L' B2 R L D2 (swap two edge pairs) 39


Example solve thread #1183
scramble: F2 R2 D F2 L2 D' R2 U2 R2 U' R F2 U' B' D2 B' R D2 F D' R'

D B' R2 F (EO) 4
L' B2 R2 U' R F2 (U/D edges + orient 4 corners) 10
R D2 R2 D' R2 D' R2 D2 R (orient 4 corners) 19
R F' R B2 R' F R B2 R2 (pair up corners) 27
R2 B2 R2 B2 R2 B2 (1 HTM edge swap to fix bad edges) 28
D' R2 B2 R2 B2 R2 B2 (edge swap to fix and solve bad edges) 35
U' L' U2 D2 R2 U2 D2 L' D U2 (edge swap + align faces) 45

Alg Garron:
D B' R2 F L' B2 R2 U' R F2 R D2 R2 D' R2 D' R2 D2 R R F' R B2 R' F R B2 R2 R2 B2 R2 B2 R2 B2 D' R2 B2 R2 B2 R2 B2 U' L' U2 D2 R2 U2 D2 L' D U2



Example solve thread #1185
Scramble L2 F2 L' B2 D2 F' L' R' F2 D' F2 R' U2 R' L B' L2 U2 R D B2 U F' U2 B2

R U D L' F' (EO) 4
B2 R2 D2 R (edges) 8
R' U' R U' R' U2 R (corners) 13
R2 U' R' U L2 U' R (corners) 19
R U' R F2 R' U R' (pair up corners) 25
U' D B2 U D' R2 (edge cycle) 31
D' B2 (set up) 33
D' R2 L2 U2 R2 L2 D' (edge swap) 40
B2 U (align stuff) 42

not fixing bad edges is not always a bad idea :)

Alg Garron: R U D L' F'B2 R2 D2U' R U' R' U2 R'U' R' U L2 U' R2U' R F2 R' U R'U' D B2 U D' R2D' B2D' R2 L2 U2 R2 L2 D' B2 U


Weekly forum competition 2011-31
Scramble U2 B' U' B' D B' U L B D' B F' L B R2 B' U2 R'

U’ F D’ L’ B (EO) 5
L’ U’ L’ D R2 D2 R’ D2 L2 (edges and face) 14
L U L’ U L U2 L . (Sune for last three corners) 20
F2 U B2 (separate U/D corners) 23
U’ B2 U B2 U2 R2 U R2 (pair up corners) 31
F2 B2 D (fix bad edges, leave pseudo ending) 34
R’ L U2 R L’ (edge cycle magically solves part of pseudoness) 39
F2 (undo pseudoness, leaves three edges) 40
At . insert U’ D R2 D’ U F2 to cancel 3 moves U F2 F2 U = U2 to make 43

final solution:
U’ F D’ L’ B L’ U’ L’ D R2 D2 R’ D2 L’ U L’ U L U2 L U’ D R2 D’ U2 B2 U’ B2 U B2 U2 R2 U R2 F2 B2 D R’ L U2 R L’ F2 = 43 HTM
 
Last edited:

miotatsu

Member
Joined
Oct 21, 2010
Messages
27
Location
Minnesota
YouTube
Visit Channel
some info on AHTA
First step: EO
like EOLine except without the line. good tutorial here http://cube.crider.co.uk/?p=eoline

Second step: equator
here is how I do this step: I find/place a pair of edges that need to be fixed onto the L/R slice (I use F and B for the EO so I am free to turn L and R in the next step). Next I find a U/D color and place it above the pair, and place the opposite color below. now when you place the two U/D edges it also partially solves your equator. usually doing this you either end up with a nearly solved E slice or solved.

Third step: corner orientation
for a very long time now I have been wondering about the algs for this. I was convinced up until yesterday that I would at some point need to generate them myself. However, I stumbled upon this by chance:
http://www.speedsolving.com/forum/s...Long-Comeau-Belt-Method)-Guide-and-Algorithms
The parity+EPOCLL step of this method is exactly what is needed for AHTA.
BGOI8.png


Fourth step: corner permutation
There are only 5 cases, I generated my own algs for these but it would be very easy to generate others if you do not care for these:
if D is solved do Y or J perms on U, else you will have one of these 3:

Double J: (place the correct corners on L)
F2 U' R2 U D R2 D' F2

Double Y: (no setup needed)
F U' R' B' L2 B R U F'

YJ: (place the correct pair on top left)
F' L U L F D2 F' L' U' L' F

Fifth step: edges
I do this intuitively, you can use things along the lines of M2 U2 M2 U2
(note that it should be easy to generate algs for all cases though, there are not that many)

Final step: edges
again, should be easy to generate algs for it but I use intuitive edge commutators and an occasional H-perm

I hope this information helps anyone else out there interested in AHTA :)
 

Cubenovice

Forever Slow
Joined
Feb 6, 2010
Messages
2,291
Location
Vlaams-Brabant (Belgium)
some info on AHTA
First step: EO
Second step: equator
Third step: corner orientation
Fourth step: corner permutation
There are only 5 cases, I generated my own algs for these but it would be very easy to generate others if you do not care for these
Fifth step: edges
Final step: edges

THX for sharing; HTA needs more love!

Some feedback/questions on your steps:
Second step: equator
I always look at this step as "solving U/D corners"
Do you think one of the "mindsets" has advantages over the other?

Third step: corner orientation
LCBM has been briefly discussed further up the thread.
My personal feelings though are that using such a large alg set goes a little against the spirit of Human Thistlethwaite.
Do you actually use this alg set or do you use regular OLL and or Heise's two twist?
I use both.
In some cases like Pi OLL the Heise alg with set up actually saves moves!

Fourth step: corner permutation
Props for finding your own algs but please have a look at these common Ortega algs, they may save you some moves :)

Double J: (place the correct corners on L)
F2 U' R2 U D R2 D' F2
Ortega: (place the correct corners on F) R2 U R2 U2 F2 U F2 or the left handed version

Double Y: (no setup needed)
F U' R' B' L2 B R U F'
Ortega: (no setup needed) R2 F2 R2 or any variation of it

YJ: (place the correct pair on top left)
F' L U L F D2 F' L' U' L' F
Ortega (place correct corners on top F): R U' R F2 R' U R' or the left handed version

Why the alternative versions?
I prefer to pick the versions that cancel moves with the previous step :)


Fifth step: edges
" you can use things along the lines of M2 U2 M2 U2"
I suppose you mean "fix bad edges"
Typically you would use single move turns on U / D at this stage to bring the edges into the right orbit and exchange with M2.
Unless you perhaps use set up moves including F and B to enable edge pair swaps to fix edges? I sometimes do something similar with R2 U2 R2 U2 R2 U2.
You will find some in my example solves.
I will keep M2 U2 M2 U2 in mind as it provides an extra opportunity to cancel moves.

Final step: edges
yep, this is all about edge cycles and edge swaps including H-perm

When solving for Fewest Moves I like to insert the edge cycles to save extra moves.
Especially the equator edges are very likely to cancel moves as they are (obviously) in the same orbit for most of the solve. As a result you will have multiple situations where they can be cycled.

But to end the solve in true Thistlethwaite it is sometimes nice to end in all double turns.
Unfortunately this adds two moves to the edge 3-cycles...
 

miotatsu

Member
Joined
Oct 21, 2010
Messages
27
Location
Minnesota
YouTube
Visit Channel
Some feedback/questions on your steps:
Second step: equator
I always look at this step as "solving U/D corners"
Do you think one of the "mindsets" has advantages over the other?

Third step: corner orientation
Do you actually use this alg set or do you use regular OLL and or Heise's two twist?
I use both.

Fourth step: corner permutation
Props for finding your own algs but please have a look at these common Ortega algs, they may save you some moves :)

Fifth step: edges
" you can use things along the lines of M2 U2 M2 U2"
I suppose you mean "fix bad edges"

second step:
I think of it as making the crosses but at the same time I focus on setting up a solved or near solved equator. I don't think there is any advantage of one viewpoint over the other.

third step:
as I have just found this set recently I only have learned a couple of the algs. I do plan on using the full set. normally up until this point and until I know a fair amount of them I solve using the standard OLL algs. My old system was like this:
OCLL, flip cube, OCLL, use Heise's two corner swapper if there was parity.
it never occurred to me to fix the parity beforehand heh

Fourth step:
dang I had no idea about those ortega algs. I will definitely start using those!

Fifth step:
that was a typo on my end, I meant single U turns, not double. that is just an example of fixing 4 bad edges on U

glad for the feedback and very excited about those ortega algs, thanks :D

edit: those ortega algs do not preserve E, still nice though.

double edit
Fourth step:
Double J:
R2 U R2 U2 F2 U F2
Double Y:
F U' R' B' L2 B R U F'
or the modified ortega version which is maybe more finger trick friendly:
R2 F2 R2 F2 U' D' R2 U D F2
YJ:
F' L U L F D2 F' L' U' L' F (ortega alg does nor preserve equator, adding a fix would be overkill)
 
Last edited:

Cubenovice

Forever Slow
Joined
Feb 6, 2010
Messages
2,291
Location
Vlaams-Brabant (Belgium)
True, the Ortega algs do not preserve the E slice but who cares?
The E slice is never more than a fast 6 mover away from solved and can be solved together with U/D edges.
 

miotatsu

Member
Joined
Oct 21, 2010
Messages
27
Location
Minnesota
YouTube
Visit Channel
I agree that for normal HTA the ortega algs are better. for AHTA however the equator is done right after EO and must be preserved for the remainder of the solve.
 

Cubenovice

Forever Slow
Joined
Feb 6, 2010
Messages
2,291
Location
Vlaams-Brabant (Belgium)
I feel that solving the equator reduces the opportunities for combining step 2 and 3.

Check my weekly comp 33 solution below, it has 12 moves for EO and orienting all corners ;-)

scramble U2 L2 B2 D F2 L' D2 F' R' L F2 D2 R' B' L F B' U2 (= inverse of comp scramble)

D2 B2 L2 B (EO) 4
R2 U2 R U D' L D L ( U/D edges, note the U to orient all corners) 12
B2 . D' R2 D' B2 (separate corners into U / D layers) 17
R2 U' R2 U2 B2 U' B2 (pair up corners, leaves 4 bad edges) 24
at . insert B2 L2 R2 F2 L2 R2 to fix bad edges, solve some and cacel 3 moves 27
D (ADF leaves 6 edges) 28
D2 R L B2 R L U2 L2 (last 6 edges via a modification of Atilla's pair swap alg) 35

With cancellations:
D2 B2 L2 B R2 U2 R U D' L D L' R2 F2 L2 R2 D' R2 D' B2 R2 U' R2 U2 B2 U' B2 D' R L B2 R L U2 L2 = 35 HTM
 

Cubenovice

Forever Slow
Joined
Feb 6, 2010
Messages
2,291
Location
Vlaams-Brabant (Belgium)
Guimond style corner orientation algs for HTA:
- Corners and edges permute
- keep EO
- keep edges in correct orbits

First orient 3 corners on D face, then orient the remaining 1 to 5 corners in a single alg

Set 1: misoriented sticker on FDL:
- L D' R2 D L standard 2-twist from Ryan Heise's tutorial
- R' D R2 D' R2 D R
- R' F2 U' F2 U R makes a nice <FRU> after y'
- B' R2 D R2 D' B
- R B2 R D' L' D' L
- F U2 F U F' U F makes a nice <RU> after y'
- R D L' D B2 L' D R
- F' U' F' U' D' F U2 F makes a nice <RUD> after y'

AUF chosen so (apart from the last 2 cases obviously) there is always a solved corner on UFL
AUF for last two: Both misoriented corners point to F
Reason: recognition

Set 2: misoriented sticker on FDR: will add later

These are just the top algs from the list I got from Jaap; I received multiple algs per case.
Will look through the list for more fingertricky algs later.

Eventually I will make up a list with diagrams.
Has anyone got a link to nice diagrams of the Guimond U layer orientations?
 
Last edited:

Cubenovice

Forever Slow
Joined
Feb 6, 2010
Messages
2,291
Location
Vlaams-Brabant (Belgium)
current avg12: 36.80 (C = 5.01)

has anyone ever sub20'd with this god damn method?

Why does it always have to be about speed?
Right... Speedsolving.com :)

Thom,
correct me if Im wrong but doesn't your experience in Roux help with quick recognition of edges the endgame?
But I understand that having the edges all over the place makes for nasty recognition.
 

Kirjava

Colourful
Joined
Mar 26, 2006
Messages
6,121
WCA
2006BARL01
YouTube
Visit Channel
yeah, the ending is my fastest part, human kociemba helps too

I think my biggest problem is with EO efficiency and corners in general

you'll see why I'm going for speed soon enough :)
 
Top