• 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!

Guide: Human Thistlethwaite Algorithm for Fewest Moves - HTA for FMC

Cubenovice

Forever Slow
Joined
Feb 6, 2010
Messages
2,291
Location
Vlaams-Brabant (Belgium)
Those of you following the FMC results in the weekly comp or the example solve thread may have noticed an increase in my FMC results.

This is a result of my quest to explore the limits of Human Thistlethwaite for FMC.
I have now had a few sub 40 solves with a best result of 35 HTM.

Here's my guide on Human Thistlethwaite for FMC (work in progress).
perhaps it can be of some use for those who want to try FMC but are afraid of blockbuilding ;-)

I know HTA probably cannot compete with the traditional blockbuilding approach but the method is quite easy to learn and could get you better results than a pure CFOP solve (excluding skips...)


Edge orientation
During EO pay attention to the following:
- Try a few different variations in EO to check what gives best continuation for edges / corners
- Try to get some U/D edges on U/D face
- Preferably leaving both with the same number of remaining edges
- With the final F or B face move place remaining edges on the same slice
- This enables adjusting U/D faces followed by a simple L or R move to solve two U/D edges

Solve edges
- Try to solve two edges at a time (set up by final move of EO)
- Try not to use too many R U R’ style stuff for edges but if you do: pay attention to U/D corners: check if an ADF will orient the corner in the D face
- If you get a 1x1x3 bar: try to preserve by placing it opposite to the final edge to be solved on the U/D face. When you place it next to an unsolved edge you will unsolve a corner.
- At your final R or L check try both CW and CCW for the resulting corner orientation case


Orient corners
- If you have more than 5 corners left you may want to go back and tinker some more with the previous step.
- Regular OLL’s can be used but beware of the move count.
- If you run into Superman you better take back some moves and try something else...
- Sune, Pi and chameleon are OK for move count but make sure to use mirrors and inverses to increase your chances of cancelling moves. Especially with Sune’s I find that I can often take a version that will cancel stuff.

Also: Don’t dismiss Heises’ “two corner twist” alg R’ D L2 D’ R too fast.
- It actually saves moves vs. pi and chameleon! You typically want to try both options to check which gives best continuation into “separation of corners”

Separate corners:
- At this stage I used to bring U corners to U and D corners to D but lately I have started to “just” separate. Direction to sort the corners to depends (again) on possible cancellations.
- You might as well combine “U to U” with the fix bad edges step later on which typically involves an M2.
- I do not have enough documented solves yet to see if one or the other gives better overall results…

Pair up corners:
- Ortega all the way! Note: these algs influence edge permutation, also in the equator.
Don’t worry: E edges will remain in E (This is the reason why I do not solve the equator early on in the solve)
- Worst case scenario: Y-perm…
- Best case: no pairs: good old R2 F2 R2 make sure to take the mirror that cancels with your last move.

Fix bad edges:
- Did you check the mirrors of the last alg?
- Your last alg will have an effect on the number and positions of your bad edges.
- Your righty Ortega -alg may have cancelled a move with pairing the corners but if it leaves a nasty edge case you may want to go back and check the lefty version.
- Or even the versions on “D” if you have the Ortega case with two bars :)


4 bad edges: Heise mentions setting up the edges for a D’ M2 D swap. This can actually work out just fine if you separated U/D corners onto the “wrong” face. But unfortunately you cannot predict this.

Other options:
U2 M2 U2 M2 or U2 R2 U2 R2 U2 R2 stuff, especially if you can use this to solve some edges and:eek:r cancel some moves

2 bad edges: via edge 3-cycle
Note: If you do not have too many edges left after separating the corners you could also skip the fix bad edges step and go straight into solving the endgame.


End game
As Heise mentions: this can be done with nothing but double turns but for FMC I obviously prefer the shortcuts.

Some typical stuff for this stage:

Edge 3-cycles (you will often find this on the E slice)
3 edges
M U M’ U2 or B2 M B2 M’ depending on cancellations (3 possible start / finish moves!)
If you’re a diehard HTA user: R2 U2 F2 U2 R2 D2 B2 D2 or starting with D2 depending on cancellations

Swapping of edge pairs
4 edges -2 pairs on R and U slice
R2 U2 R2 U2 R2 U2 or starting with U2 depending on cancellations

4 edges – 2 pairs on M slice
M2 U2 M2 U2 or starting with U2 depending on cancellations

4 edges – 2 pairs offset on L and R slice, normal swap on D
F2 U’ D’ R2 U D F2 D2 or starting with D2 depending on cancellations (thx Atilla)
Diehard HTA version:
R2 U2 F2 U2 R2 B2 D2 B2 (this is a variation of the double turn edge 3-cycle)

6 edges - normal swap on U + 4-cycle on E (D solved)
F2 U D R2 U D B2 D2 or starting with D2 depending on cancellations (variation on Atilla’s)

6 edges - normal swap on D + 4-cycle on E (U solved)
F2 U’ D’ R2 U’ D’ B2 U or starting with D2 depending on cancellations (variation on Atilla’s)

Done? Hell no!
Now the fun really starts, it is time for some info on insertions.
Corner cycles typically make for good insertions, regardless of method.
Edges insertions typically are very tricky.
This is where the unique features of HTA kick in. HTA is great for edge insertions!

A listing of the possible insertions
- Ortega case with three bars? Those who solve this via an A-perm will realize this is just a corner 3-cycle. You can check for an insertion of this one at the end of the solve (take care to cycle in the right direction though)
- Fix bad edges: edge swapping makes for great insertions. Try insertions if the regular edge fix takes too many (set up) moves
- 4 bad edges: swap edges with U2 M2 U2 M2 or U2 R2 U2 R2 U2 R2
- Here you have several options:
- Case 1 Sticker just the two “red” edges and swap with any ”blue” edge (or vice versa)
- Case 2 Sticker four edges if you have an edge pair that can be solved with swapping.
- If you do not find a good insertion to solve all four than just treat as case 1

- 2 bad edges: use an edge 3-cycle
- Cycle Red – Blue – Red or Blue – Red – Blue “swap” a Red and a Blue edge
- You only have to sticker the two bad edges (but may sticker preferred 3rd edges)
- Endgame: this is all edge swaps and 3-cycles so again insertions are possible

Notes:
If you do a “Bad edge fix” insertion you typically should do this before end game insertions.
This because you will probably use “any” edge in your insertion. Using “any” edge will affect edge permutation in your endgame.

Want to explore more starts?
Have a go at the inverse scramble...
Or try some NISS: do EO on the inverse and then switch back to normal with (EO)' as premoves.

Want less moves for EO?
xyz moves are your friend: By simply re-orienting the cube you change the number of mis-oriented edges.
Drawbacks:
- recognition, especially when using different U/D colors as usual.
- less mis-oriented edges does not perse mean fewer moves so choose wisely


For info on the actual HTA method:

Ryan Heise's page on HTA - The Original
Speedsolving Human Tistlethwaite Help/ Discussion Thread
 
Last edited:

mariano.aquino

Premium Member
Joined
Dec 18, 2009
Messages
104
Location
Buenos Aires, Argentina, Argentina
WCA
2011AQUI01
YouTube
Visit Channel
hey! such a good idea, HTA is very interesting way of learning how some things "work"
i tried a couple of variations, however i havent got to any conclusions yet. what do you think about this:
start by doing guimond for corners, followed by sort-of-L6E on the whole cube to orient edges, and then continue with normal hta
i´ll give it a try and comment!
 

uberCuber

Member
Joined
Jun 24, 2010
Messages
3,921
Location
Tucson, Arizona, USA
WCA
2011THOM01
hey! such a good idea, HTA is very interesting way of learning how some things "work"
i tried a couple of variations, however i havent got to any conclusions yet. what do you think about this:
start by doing guimond for corners, followed by sort-of-L6E on the whole cube to orient edges, and then continue with normal hta
i´ll give it a try and comment!

No matter how you follow through with it, I can't see corners-first being a good way to approach FMC at all.
 

Brest

Moderator
Staff member
Joined
Sep 30, 2010
Messages
2,440
YouTube
Visit Channel
Here's my guide on Human Thistlethwaite for FMC (work in progress).
Nice work dude! I will have to set some time aside to go through this in detail. Looks great.

No matter how you follow through with it, I can't see corners-first being a good way to approach FMC at all.
You should have a look at some of Attila's FMC solutions. Try this and this and this for some examples. iirc he only uses corners 1st and usually has sub 30 FMC solves.
 

Cubenovice

Forever Slow
Joined
Feb 6, 2010
Messages
2,291
Location
Vlaams-Brabant (Belgium)
what do you think about this:
start by doing guimond for corners, followed by sort-of-L6E on the whole cube to orient edges, and then continue with normal hta

I'v been thinking about a variation of this for a very long time: EO followed by Guimond-ish corner orientation.
For this I need to make a set of HTA safe "Guimond algs". HTA safe meaning algs should keep U/D edges in U/D and E edges in E. Permutation within these layers may change.

I already have a set of over 60.000 HTA safe corner orientation algs computed by Jaap Scherphuis: multiple algs for all possible corners orientations.
I still need to extract some algs for some of the "nasty" corners orientations, for instance three corners left but not on the same face...
 

TMOY

Member
Joined
Jun 29, 2008
Messages
1,802
WCA
2008COUR01
No matter how you follow through with it, I can't see corners-first being a good way to approach FMC at all.
You just have no clue about what you're saying. I use cornrs first too and I've won several FMC comps with it.
The main problem is the predominance of HTM counting (which just makes no sense when talking about CF, and which is the main reason why I can't really take that event as a serious one).
 
Joined
Sep 25, 2009
Messages
91
Location
Odijk the netherlands
WCA
2009PAUL01
Nice explanation of HTA FMC, i however prefer the solve 2x2 block, and then just look for a way to expend that 2x2 to a f2l with atleast the LL edges oriented.

Can you check youre private messages?
I still can't find a nice way too solve my last 3 cycle of edges/corners.
 

Cubenovice

Forever Slow
Joined
Feb 6, 2010
Messages
2,291
Location
Vlaams-Brabant (Belgium)
Pretty much any method can be used for FMC, but it's a very bad idea to restrict yourself to only one imo.
Nice explanation of HTA FMC, i however prefer the solve 2x2 block, and then just look for a way to expend that 2x2 to a f2l with atleast the LL edges oriented.
Let us not forget that all this is just a fun exercise to find the limits of HTA.
For full-on FMC I also use the more standard blockbuilding approach and seizing all opportunities I get.

Can you check youre private messages?
I still can't find a nice way too solve my last 3 cycle of edges/corners.
Currently writing page 3 (not kidding) of my reply with a full walk through of your solve.

I'll probably be at the Dutch Cube day.
Perhaps you can type out a solve or three and then we can go over the insertions in real-life.
I'll bring my assum FMC stickers, you bring something to drink ;)

No matter how you follow through with it, I can't see corners-first being a good way to approach FMC at all.
Atilla, Francois, Irontwig and Guus to name a few have proven CF to be capable of very good FMC results.
I hope to do the same for HTA some day ;)

Fun method
Fo Sho
The cube looks better than lbl in a speedsolve.
Why?
 
Last edited:
Joined
Sep 25, 2009
Messages
91
Location
Odijk the netherlands
WCA
2009PAUL01
Currently writing page 3 (not kidding) of my reply with a full walk through of your solve.

I'll probably be at the Dutch Cube day.
Perhaps you can type out a solve or three and then we can go over the insertions in real-life.
I'll bring my assum FMC stickers, you bring something to drink ;)

sounds great =D, im really interested in that 3 page walk though of my solve :).
i will practise some more FMC and type out some of my results, shall i also type out all the other combinations i tried out in that solve?
looking forward to it.
Tristan
 

Cubenovice

Forever Slow
Joined
Feb 6, 2010
Messages
2,291
Location
Vlaams-Brabant (Belgium)
I see that Sebastiano Tronto has mentioned this thread in his most excellent FMC tutorial.
So I better put in some real life examples showing the techniques mentiones earlier in this thread.


Scramble 1 of the speedcubers.de weekly competition (somewhere back in 2012....)

scramble
R2 B2 L2 D B2 R2 D B2 L2 B2 D2 F' R U2 B' U F' R F' U' F' U'

inverse
U F U F R' F U' B U2 R' F D2 B2 L2 B2 D' R2 B2 D' L2 B2 R2


F' B' - Edge orientation
switch to inverse with premoves F B
. L' R D R' U2 D L' D L : - Corner orientation
U' L2 - corner cycle would pair up corners (solve with random commutator, insert cycle later)
D' F2 B2 - bring corners to correct layer, leaves edge cycle and edge 2+2 swap
F B - undo premoves at this stage cancels 2 moves

insertions:
. pair swap U2 L2 U2 L2 U2 L2 cancel 1 move
: edge cycle L2 + U' D F2 D' U cancel 3 moves
+ corner cycle L2 U' R2 U L2 U' R2 U cancel 3 moves

final solution for inverse scramble
U2 L2 U2 L2 U2 L R D R' U2 D L' D L U' R2 U L2 U' R2 D F2 D' L2 D' F' B'

for regular scramble:
B F D L2 D F2 D' R2 U L2 U' R2 U L' D' L D' U2 R D' R' L' U2 L2 U2 L2 U2 = 27 HTM

----------------------------------------------------------


speedcubers.de
scramble 2 of week 2012-12

B2 D2 R D2 B2 U2 F2 U2 L B2 L2 D R2 B' R' U' B2 R' B2 L

R L2 B' - edge orientation
L U2 R' - place U/D edges
R' D L2 D' R' - corner orientation
D L2 U2 D' R2 - separate corners note the extra moves
R' F R' B2 R F' R' B2 R2 - pair up corners with A-perm
note that there are no bad edges
R2 F2 R2 F2 R2 F2- solve 2 edges and line up two for swapping
F2 B2 U2 F2 B2 - solve edges
U F2 U' D R2 U' - done

but why would I bother lining up faces for an edge cycle?
last line is much better as F2 U' D R2 in which the F2 cancels out

this makes 32 HTM:
R L2 B' L U2 R2 D L2 D' R' D L2 U2 D' R F R' B2 R F' R' B2 F2 R2 F2 R2 B2 U2 B2 U' D . R2

now take out the A-perm and insert D B2 D' F2 D B2 D' F2 at . to cancel 1 additional move for final solution:
R L2 B' L U2 R2 D L2 D' R' D L2 U2 D' F2 R2 F2 R2 B2 U2 B2 U' D2 B2 D' F2 D B2 D' F2 R2 = 31 HTM


-------------------------------------------------


FMC round 382
R2 F' B' D B D B' D' L2 R' F R2 F' B L' F2 B U2 D2 R' L2 B2 D' R' U' F U F' D2 B2


F B' L' B' – orient edges
U' D2 R' U2 D2 R - orient all but 2 corners
R’ F L F’ R F L’ F’ – corner cycle instead of the typical HTA two corner swap
D L2 – separate corners
B2 D' B2 D2 . R2 D' R2 D2 – pair up corners
D' U2 R' L B2 F2 L' R D' – leaves 2+2 edge
At . insert edge swap D2 L2 D2 : L2 D2 L2 to cancel 2 moves

Now replace the initial corner cycle by insertion : to cancel two additional moves:
At : insert D R’ D’ L2 D R’ D’ L2 to cancel 4 moves

Final:
F B' L' B' U' D2 R' U2 D2 R D L2 B2 D' B2 L2 D' R' D' L2 D R D L2 R2 D' R2 D U2 R' L B2 F2 L' R D' = 36 HTM, 30 STM
 
Top