my direct solving method?

[Kirjava]

You didn't say it was a new method, but claimed you created it - naming it something 'new' doesn't help the impression >_>.

As for copying the site, I wouldn't worry about it. It's not hard to churn out something like that, and I don't claim copyright - but a credit would have been nice. Alternatively, you could have just made it a forum post or a simpler webpage, maybe even on a single page. You obviously understand what you did, so I won't go on about it. You can imagine my surprise when seeing the webpage initially, though.

Anyway, it's good to see more people are looking into direct solving for big cubes. Keep it up!

 

[Kenneth]

r U2 r' U2 r' U2 r' U2 r
l' U2 l U2 l U2 l U2 l'

U2 r' U2 l U2 l' U2 F2 l' F2 r

Shorter than my original parity algs but these does not preseve middle layer centres (U and D). The first two saves 4 turns each, the last only one r2, but still =)

Have fun

BTW, to fix opposite 1x2 blocks of centres do r2 f2 r2 f2 ... or faster r2 y r2 y l2 y l2

 

[Kenneth]

r U2 r' U2 r' U2 r' U2 r
l' U2 l U2 l U2 l U2 l'

lol, forget those and try this instead:

r2 U2 r U2 r2
l2 U2 l' U2 l2



Ortega H OLL =)

 

[rachmaninovian]

bump: the CNY holidays meant that I could revive my site.

http://rachmaninovian.webs.com

I did my best to credit all those involved as possible; if I had left any details out please do kindly tell me.

What is new - more algs! and some new pics.

Comments are appreciated. Once again I apologize for whatever wrong I have done in the past.

 

[somerandomkidmike]

Thanks for posting this. I am going to try some of the center algorithms. I also plan on trying the 2 step 4x4 centers posted by TMOY. Right now I am working on fewest moves for the 4x4, and this is going to help a lot.

 

[rachmaninovian]

I do not believe that the system that I use for the middle edges are good for FMC; they are speed-optimized and not move optimized! xD if you want optimal algs you will have to generate your own algs...you will see that the middle edges are like a 2x2 that can be scrambled and solved with <R, L, U2, D2, F2, B2>. so erase of the edges of the 3x3, and setup the corners of the cube. note that not all algs generated will work; I had to go through hundreds of algs to find something that works, and is fingertrick friendly.

 

[somerandomkidmike]

I do not believe that the system that I use for the middle edges are good for FMC; they are speed-optimized and not move optimized! xD if you want optimal algs you will have to generate your own algs...you will see that the middle edges are like a 2x2 that can be scrambled and solved with <R, L, U2, D2, F2, B2>. so erase of the edges of the 3x3, and setup the corners of the cube. note that not all algs generated will work; I had to go through hundreds of algs to find something that works, and is fingertrick friendly.

Apperently some have helped, because I got an 89 turn solve. You're right that some need to be optimized more, but others have helped a bunch. But imagine if one day I could do 100 turns while speedsolving! I'd almost be guaranteed to have under 1 minute if I had a good cube. I will try to find better algorithms too. I'll help you optimize this method, because I think it could end up being better than reduction someday (not that it isn't already).

 

[rachmaninovian]

I do not believe that the system that I use for the middle edges are good for FMC; they are speed-optimized and not move optimized! xD if you want optimal algs you will have to generate your own algs...you will see that the middle edges are like a 2x2 that can be scrambled and solved with <R, L, U2, D2, F2, B2>. so erase of the edges of the 3x3, and setup the corners of the cube. note that not all algs generated will work; I had to go through hundreds of algs to find something that works, and is fingertrick friendly.

Apperently some have helped, because I got an 89 turn solve. You're right that some need to be optimized more, but others have helped a bunch. But imagine if one day I could do 100 turns while speedsolving! I'd almost be guaranteed to have under 1 minute if I had a good cube. I will try to find better algorithms too. I'll help you optimize this method, because I think it could end up being better than reduction someday (not that it isn't already).

I'm sure CLL and blockbuilding the first layer helped. I'm averaging about 130 turns without much thinking xD and again, I believe the speed of this method very much relies on how fast you can do each stage, not the move count..=D Any ideas on optimizing the method?

 

[somerandomkidmike]

I do not believe that the system that I use for the middle edges are good for FMC; they are speed-optimized and not move optimized! xD if you want optimal algs you will have to generate your own algs...you will see that the middle edges are like a 2x2 that can be scrambled and solved with <R, L, U2, D2, F2, B2>. so erase of the edges of the 3x3, and setup the corners of the cube. note that not all algs generated will work; I had to go through hundreds of algs to find something that works, and is fingertrick friendly.

Apperently some have helped, because I got an 89 turn solve. You're right that some need to be optimized more, but others have helped a bunch. But imagine if one day I could do 100 turns while speedsolving! I'd almost be guaranteed to have under 1 minute if I had a good cube. I will try to find better algorithms too. I'll help you optimize this method, because I think it could end up being better than reduction someday (not that it isn't already).

I'm sure CLL and blockbuilding the first layer helped. I'm averaging about 130 turns without much thinking xD and again, I believe the speed of this method very much relies on how fast you can do each stage, not the move count..=D Any ideas on optimizing the method?

I've got two things to say. First of all, if you lower your moves, then improve your look-ahead, you won't have to spend very much time on the steps anyway. When I am done the first layer and the opposite corners, it is usually done around 30 seconds. Considering that I just started using this method, I'd like to think that this is pretty good. I think with any method the speed of it relies on how fast you can do each stage. I think that statement is obvious. Second of all... have you tried block-building then CLL? Just kidding. I think to optimize it, you need to slow down the first few steps a little, so you can have some edges places already when solving the first layer. This can help your look-ahead for the next steps so you can use less turns. Also do center control when possible. Learn to recognize cases that are fast for center control. My best solve was a non-lucky solve where I just did all the steps in very few turns. I was not turning fast all but I did blockbuilding for the first layer, and center control when finishing the edges. I'll show EXACTLY how I solve it some other time, but right now I have to go sleep.

 

[rachmaninovian]

well...i tried doing your style and lookahead is hard...i'm abt 26-27s with first layer+corners and I don't know CLL =D the bad point abt this way is I cannot see the rest of the edges good enough..in my system onces the corners are solved I can lookahead very very well for the edges..i'd experiment more and get back to you.

EDIT: yes, I am currently experimenting with solving some of the opposite layer edges will inserting the first layer edges..in fact I am currently using it in normal speedsolving already. this gives me less pieces to solve thus less lookahead is needed..and I can execute as fast as i can =D
How do you exactly pair up the last 2 dedges of the R and L faces, and where do you pair them? right now as you know I leave one dedge to be solved with 15-16 algorithms which are in fact forms/conjulates of niklas, so I can lookahead to the midges easily..anyway take a look at step 4b on my page; I have updated it with some of TMOY's algs for oriented midges because in certain cases, oriented midges are a pain in the neck..so learning parts of other systems should prevent these bad cases(yet good cases for our friend here).

 

[somerandomkidmike]

I'd have to agree with you that it is the hardest part to see the other edges. It really takes a lot of practice (I am getting better at it). Once you practice for long enough, it does get easier. When I solve the last edges on both sides, the way that I do it varies depending on what is the fastest. Usually I will solve 2 edges at a time (a redge and a ledge). I think this way, you can have a better look-ahead. You can also do center-control a lot easier.

 

[rachmaninovian]

as you know, i solve the last...redge, I use conjulates/variations of niklas, so the only stuff I mess up is the r and l slice edges..and it is really easy to see the centres...
generally I try to lookahead to the next step so I'm compromising on my turning speed for that, but it is working well I guess. This i liken to my musical experiences, where i literally lookahead of where I am playing and thus I am able to sightread music well. In the same way, I try to minimize pauses and if I do pause I make my pauses worthwhile. This method very much relies on how much you can plan your solve on the spot to be efficient...
Also, I practise my execution speed for the algs of the last 2 midges...I think it is important that you can cut 1-2s there and give yourself a bit more time to think and process the fastest way to solve the centres.

On another note: I do think that your variation does do a lot of difference to the move count; but I recommend that you solve whatever piece you see at the moment. It has been proved to be faster than trying to find the shortest solution. I believed I have read from somewhere that Per does that...and Thom Barlow as well for his building of a 1x3x4 block on the left. I do not believe in the efficiency of that (move count wise) but I believe in the speed potential of that. Maybe you should try that too and see whether you get faster =D

Here is a breakdown of a random solve:
1st 2 centres - 5.33s
corners - 11.47s
finish redges and ledges - 26.13s
finish cage - 10.44s
centres (a bit smooth) - 16.63s

I did each step without inspection (except first 2 centres)..so total was 1:10.00 which is around my average after a warm up.

 

[somerandomkidmike]

Just so you know, I do solve the first thing I see at the moment. I have tried many ways of solving the first layer and corners, and this is just the way that has given me the best times. I know that recognition is in most cases more important than the move count, but I think I could get just as good recognition (with a lot of work) using my variation. I just did a couple of solves using your first layer (when I was on the bus) and most of the times were over 2 minutes. I will continue experimenting, and tell you the results later.

 

[rachmaninovian]

yea, I guess that different people will have different preferences..somehow I think it has to do with psychological stuff as well..I've mentally grouped my system into 5 parts (as above) and your system has 6 i think (first 2 centres, FL with empty dedge, LL corners, finish redges and ledges, finish cage, centres). not much diff..but wells..I took an average of 5 yesterday and I got 1:06.09, (1:02.83), 1:04.88, (1:18.89), 1:02.89 ---> 1:04.62. everything was nonlucky except the 2nd solve..I only needed to do 2-3 algs to solve the centres xD and the 4th solve i actually used the wrong alg to finish the midges...

On the other hand, I have been experimenting with pure cage which is FL, F3l (since you can ignore centres there are certain tricks you can do), LL corners+edges, and then final 5 centres. I'm able to get 1:40ish times sometimes (I don't know abt the average) and my pb with it is 1:32...I'm trying to see whether I can learn anything from pure cage that I can adapt to my system...

I guess that maybe one day our methods may become super hybridized xD

 

[Tweker]

I love that method, thanks for sedins a MP when you put your site back online. I just started looking at it, I can't wait to be able to solve my professor with that method. I've seen on youtube that at first you didnt have that much alg. Whats the minimum number of alg needed to solve using your method? (I'm not realy good at learning alot of alg fast so with that much alg it will be a long time till I am able to solve it on my own) If there is a way to finish it with less alg at first, let me know, if not well I guess I'll have to work on all these algs :O

 

[rachmaninovian]

yes...if you go thru my pages really slowly, you'd see that most of the "algs" are intuitive...like my algs from step 3a to 4a. if you go through them slowly on a cube I think it'll be very easy for you to remember as they flow very well and you might even come up with your own shortcuts it is very much intuitive, believe me...for step 4b...I learnt one for each case so that is 16 algs to learn...note that I astericked those I use most often; they are the most fingertrick friendly ones.

now to centres for 5x5...learn the simple ones first..like single cubie switches and stuff..then slowly advance...up to now I'm still discovering and learning more centre algs...of course to be most efficient knowing everything is best but it'll take awhile..and later on you will find that my centre algs are mostly intuitive..not all but you will be able to see that some are variations of the other..hope this LONG post helps
if you are really anxious and want a simpler way to finish the middle edges and learn the simplest centre algs, you only need something like 7-8 algs to learn.. the system I am currently using for middle edges is definitely faster then that primitive method but I guess it is better for you to start of something simpler..

 

[Neroflux]

im still stuck with keyhole experimenting.... what's keyhole?

 

[rachmaninovian]

the empty dedge that you are allowed to not care with at first, to ease your solving of the 3rd to 7th dedge.

niao me lor, ask me before le, still liddat lor, i spank you i tell you, naughty stepgrandson, i'd get blah to whack you, and monkey sweat to scold you.

 

[Tweker]

How do you do for the step4 with a 5x5x5? do you flip the edge( using the 4x4 alg) when they are not flipped corectly or there is an other way?


p.s. you were right, very intuitive, in fact, the only time I need to look at algs is part 4a and 4b, other then that I can go by myself and find it out thanks alot for the tutorial, I love it

 

[rachmaninovian]

you mean the case whereby I do M' U M' U M' U2 M U M U M U2?

I do it as (r l') U (r l') U (r l') U2 (r' l) U (r' l) U (r' l) U2 =) but I'm trying to find another way as this way is REALLY slow..

and also, 4a is really intuitive...I found almost every single case myself...on the day I decided to switch methods for the middle edges =D I cheated on the last one; I ran ron's cube solver (3x3 one) to find a solution for that.