Disproving UF5

[abunickabhi]

i have even made a new intro, and added some new editing stuff to the video alg discussion.

i am also planning to do some 5-style video walkthrough solves this month.

It will help people make judgement on when to use it, how much to use it and when to resort to UF 3-style.

 

[abunickabhi]

Crossed 109k Letter quads,

My progress with letter quads and 5-style each year,

end of 2016 - 5k letter quads all the easy MU gen algs and setup to that discovered and listed out, referred to Ishaan's algs and Oleg's algs and added them to my sheet.

end of 2017 - 23k letter quads, refuted viability of 5cycle corners, and focus now only on edges. Made a subset of 5-cycles only for Roux LSE.

end of 2018 - 48k letter quads, stopped genning DF buffer and focus solely on UF5. Discussion with Yongqiang peng, regarding the derivation of various categories of 5-cycles and mathematics of it (most of the comms are not of the form [A,B]).

end of 2019 - 58k letter quads, very less free time this year hence only 10k+ LQs added. Also I did not have a laptop, and had to save money to buy one so that I could start genning again. Drilling of algs, and comparing times with 2 3-style exec.

end of 2020 - 97k letter quads, thanks lockdown for creating lot of time. Coded some software for 5-style Yo notation, reconstructions, cube images, latex pdf export, keyboard cube trainer. Completed 17k UF5 algs which rotationless,speed optimal and have 0-2 regrips.

March 2021- Currently at 109k letter quads done!!! Yippee, near completion. All mirrors and inverses checked.

 

[PCCuber]

Your dedication to this inspires me a lot. I haven't even learnt full Oll yet lol.

 

[guelda]

Man this is motivation, you rocks! Definitely inspiring as PCCuber wrote.

 

[abunickabhi]

Will be most likely be finishing with ABXX subset in video format (video of alg and its execution), by this summer. I would love if people are interested in contributing to the alg execution.

If you are interested in showing the fingertrick for any future case, and make a small video executing it 5 times, it will be much helpful. Most of the submissions for videos for non-ABXX cases will be handled on the #5style-fingertricks channel on the 5-style discord. I hope to see a few contributions.

 

[abunickabhi]

The total hours upto the year 2020, that I have spent on making this method work is, 50 hours on fingertrick video making, 50 hours on documentation, 10 hours on kb trainer, 100 hours on software, 700 hours on making algs, 500 hours on stuff that I replaced later on, 1000 hours on letter quads making. I am not counting the time when I do background thinking of UF5, so overall about 100 days of human time has been spent till now.

I hope my estimate is approximately correct.

 

[abunickabhi]

After the release of Gil's software on fingertricks visualisation today, it will make my task of documenting fingertricks for UF5 set much easier.

Until now, if I had to show the fingertrick for any 5-style edge alg, I would have to make a video about it, or write it down (less expressive way, and sometimes confusing). Now the role of the fingertrick analyst, is a bit easier. If anyone of you guys are interested, the job is to just see if the playout from the fingertrick lab has regrips, and how to remove those regrips by some fingertricks.

I will be also changing the format of my each 5-style edge video on the 5AlgSolve channel, how instead of setup, and alg execution 5 times, I will include the screen grab from the speedcubestats window as well, at the end of each case video.

 

[abunickabhi]

Good solve LLK!
First person other than me who has used a 5-cycle in a solve.

The alg used was:

(UF-UL-UB-BD-UR ), athough he traced it from the UB buffer.

 

[abunickabhi]

Okay so I have finished shooting the entire ABXX series in video format. Editing it too just too much time (about 20 minutes per video). The videos will be released in sequence with frequency of one alg per day on the 5AlgSolve YT.

ACXX and ADXX video series will be much faster, fingers crossed.

Also, the video of all these algs are linked to the 5-style sheet, so you can just click on a case to view the fingertrick for the alg in video format.

Also, Letter Quads update, now I am currently at 113,286 today before going to sleep on 26th May 2021.
Cheers and Happy Summer!

 

[abunickabhi]

I crossed 20k verified algs on my sheet today. So, 20k edges which are 5-cycles which I can say is quite good, and a bit faster than executing 2 3-style UF algs.

Verifying 20k took about a year worth of time.

 

[Deleted member 36445]

After the release of Gil's software on fingertricks visualisation today, it will make my task of documenting fingertricks for UF5 set much easier.

Until now, if I had to show the fingertrick for any 5-style edge alg, I would have to make a video about it, or write it down (less expressive way, and sometimes confusing). Now the role of the fingertrick analyst, is a bit easier. If anyone of you guys are interested, the job is to just see if the playout from the fingertrick lab has regrips, and how to remove those regrips by some fingertricks.

View attachment 15653


I will be also changing the format of my each 5-style edge video on the 5AlgSolve channel, how instead of setup, and alg execution 5 times, I will include the screen grab from the speedcubestats window as well, at the end of each case video.

sausage fingerrrs

 

[Deleted member 36445]

a bit faster than executing 2 3-style UF algs.

"memorizing 20k algs to be a bit faster than 3 style" - abunickabhi 2021

 

[DuckubingCuber347]

Just a bit.

 

[abunickabhi]

sausage fingerrrs

Maybe in future updates, Gils hires a computer graphics guys, and the person designs realistic fingers maybe, that look less like sausage.

 

[abunickabhi]

Tried to compile some statistics today.

Total number of wide moves in my 5-style edge set = 4382/209073 = 2.095% (which is a lot less than the slice moves MES which constitute the major chunk of each alg.)
So, wide moves do exist, and they are quite fun to do, r and l are more common as compared to u and f. I try to avoid b,d wide moves as they are not fingertrickable.

Out of all the letter quads that I have annotated, about 37.8% represent Object images, 31.7% represent action images, 13.7% represent Person image, 6.5% represents an emotion or a human feeling, 5.6% represents language literal and 4.65% represents a place or a city. Cool stats indeed!

 

[abunickabhi]

I have taken a bit of a break from pushing this algset, as I have a full time research job since June 2021.
There are still 180ish people in the discord helping me out clarify lot of letter quads and alg fingertrickability which is cool.

I think once comp come back in India in 2022 potentially, I would be interested to pursue this project a bit more.
I am happy to have spent time in UF5 since 2017. I just cannot believe that 2014 me thought of this idea. The idea of cycling 5 pieces is just insane ngl, R' F' S2 E R E' S R' f R.

 

[DuckubingCuber347]

I've always wondered, are the algs you have at the end of your posts a secret message or is it just a random 5-style alg?

 

[abunickabhi]

I've always wondered, is the algs you have at the end of your posts a secret message or is it just a random 5-style alg?

It is not cryptic enough to hold a message. It is just a random case that I have on my mind while typing out something, R' D' M F M2 F' U2 M U2 D R.

 

[abunickabhi]

Update:

I am not focussing on improving the alg quality nowadays as it is a laborious task.
I just try to do example solves using 5-style and post it on example solve threads on SS.

 

[Cubing Forever]

I am not focussing on improving the alg quality nowadays as it is a laborious task.

I have an idea for that:
You could make an alg improvement channel on the 5-style discord and post a random case and it's alg everyday for people to improve on it. Just like how you do alg-of-the-day.