Algorithm Database Discussion

[Ranzha]

ELL images on the dev version look like this:

But if you'd like I could add a view option to make them appear like this:

You did it again XP.
Lol. No worries.

Sounds great! I think either would be okay. I sincerely think that the former one is better for those who use ELL first, if any.

 

[miniGOINGS]

Here we go.

http://www.ai.univ-paris8.fr/~bh/cube/

When dealing with LL solutions for EO, EP, CO, and CP here are the optimal movecounts;

Fridrich (OLL + PLL)
9.22 + 11.21 = 20.43

Backwards (PLL + OLL)
8.1 + 12.8 = 20.9

Corners (CLL + ELL)
9.18 + 11.27 = 20.45

Edges (ELL + CLL)
7.87 + 11.73 = 19.6

1LLL (LL)
12.58

 

[Ranzha]

Here we go.

http://www.ai.univ-paris8.fr/~bh/cube/

When dealing with LL solutions for EO, EP, CO, and CP here are the optimal movecounts;

Fridrich (OLL + PLL)
9.22 + 11.21 = 20.43

Backwards (PLL + OLL)
8.1 + 12.8 = 20.9

Corners (CLL + ELL)
9.18 + 11.27 = 20.45

Edges (ELL + CLL)
7.87 + 11.73 = 19.6

1LLL (LL)
12.58

It's interesting that Backwards, the inverse of the preferred Fridrich LL system, has a negative effect on the movecount, whereas the preferred CLL + ELL is worse than Fridrich and its inverse, ELL + CLL, has a sub-20 movecount.

I'mma try this!

 

[miniGOINGS]

Mhm. I'm not sure in which metric the movecounts were generated with, but I think I should find out.

 

[Kenneth]

... and its inverse, ELL + CLL, has a sub-20 movecount

Sounds fine but is much worse than it looks, it is low because ELL first is way easy, almost intuitive and have short algs and I think 14 cases but CLL last is horrible, many have long algs and it is around 80 cases (1:6 of ZBLL), absolutly nobody uses it because of that.

I was learning that once, or... I was using a method that did VHF2L, EP + 1 corner and then the last three corners. then I was planning to expand to 2 steps, learned a few cases but found it was so many of them... I choosed to do CLL/ELL instead and later switched to COLL (because I hated to see oriented edges flip during CLL).

 

[miniGOINGS]

Sounds fine but is much worse than it looks, it is low because ELL first is way easy, almost intuitive and have short algs and I think 14 cases but CLL last is horrible, many have long algs and it is around 80 cases (1:6 of ZBLL), absolutly nobody uses it because of that.

I was learning that once, or... I was using a method that did VHF2L, EP + 1 corner and then the last three corners. then I was planning to expand to 2 steps, learned a few cases but found it was so many of them... I choosed to do CLL/ELL instead and later switched to COLL (because I hated to see oriented edges flip during CLL).

Actually, it doesn't look like it.

"permutation and orientation of edges (13 configurations, average : 7.87),
then permutation and orientation of corners (42 configurations, average : 11.73)."

Doesn't sound too bad to me.

 

[Cride5]

OK, an update has been made to the alg database. The following updates were made:

1. BUGFIX: Swap CLS + ELS
2. BUGFIX: Sorted out COLL/CLL/ELL plan view
3. FEATURE: "add" link on view pages, for ease of access
4. FEATURE: Allow removal of individual algs (currently admin only)
5. FEATURE: Improved parsing of user input to allow: "`", "R'2" and "Rw"
6. BUGFIX: Fixed verification problem for algs ending in x2 rotations
7. FEATURE: Improved duplicate detection, for duplicate algs posted with different rotations
8. FEATURE: Bulk algorithm upload
9. FEATURE: Link to alg.garron.us for algorithm animations
10. FEATURE: Initial rotations in separate column, final rotations hidden
11. FEATURE: All AUFs automatically replaced with y rotations for LL algs

I've also re-processed the ZBLL and COLL algs to change AUFs for rotations, so that the move-count stats are better.

For more details on updates and planned features, click here.

If you want any features that aren't in the ToDo-list please let me know below...

 

[Kenneth]

Sounds fine but is much worse than it looks, it is low because ELL first is way easy, almost intuitive and have short algs and I think 14 cases but CLL last is horrible, many have long algs and it is around 80 cases (1:6 of ZBLL), absolutly nobody uses it because of that.

I was learning that once, or... I was using a method that did VHF2L, EP + 1 corner and then the last three corners. then I was planning to expand to 2 steps, learned a few cases but found it was so many of them... I choosed to do CLL/ELL instead and later switched to COLL (because I hated to see oriented edges flip during CLL).

Actually, it doesn't look like it.

"permutation and orientation of edges (13 configurations, average : 7.87),
then permutation and orientation of corners (42 configurations, average : 11.73)."

Doesn't sound too bad to me.

Can't be correct, 27 orientations times 12 permutations divied by 4 makes a total of 81

 

[miniGOINGS]

Can't be correct, 27 orientations times 12 permutations divied by 4 makes a total of 81

Yea, it does sound strange. I'll look into it.

 

[Ranzha]

OK, an update has been made to the alg database. The following updates were made:

1. BUGFIX: Swap CLS + ELS
2. BUGFIX: Sorted out COLL/CLL/ELL plan view
3. FEATURE: "add" link on view pages, for ease of access
4. FEATURE: Allow removal of individual algs (currently admin only)
5. FEATURE: Improved parsing of user input to allow: "`", "R'2" and "Rw"
6. BUGFIX: Fixed verification problem for algs ending in x2 rotations
7. FEATURE: Improved duplicate detection, for duplicate algs posted with different rotations
8. FEATURE: Bulk algorithm upload
9. FEATURE: Link to alg.garron.us for algorithm animations
10. FEATURE: Initial rotations in separate column, final rotations hidden
11. FEATURE: All AUFs automatically replaced with y rotations for LL algs

I've also re-processed the ZBLL and COLL algs to change AUFs for rotations, so that the move-count stats are better.

For more details on updates and planned features, click here.

If you want any features that aren't in the ToDo-list please let me know below...

Thanks SO MUCH for this. This is way much and a HUGE improvement.

 

[Ranzha]

BUG?:

Tried uploading algs for the mirror of ELL #3.
This is ELL #3:

This is what I want:

Nothin' doin'. Just keeps adding to ELL #3

 

[StachuK1992]

If someone were to fix ELS and CLS, so one could actually add stuff, that'd be super awesome, and I'd add a lot.

Sorry for the bump, but this has been bugging me.

Edit:
Also, a CPLS section would be nice.
If someone were to add it to the database, I'd happily put up some algs.

Is there any way a regular user like me could add the CPLS section?
Or does one have to have higher permissions? :/

 

[Cride5]

I'm pretty sure uploading is possible - just log into the wiki and there should be an 'add' link in the 'Mode' section.

It appears as if ELS and CLS are still swapped though. I thought I'd fixed this so not sure why it's still happening. Will check it out when I've got time.

 

[Escher]

I don't have much to add personally to this project but I recently found this Polish Algorithm Database. It seems to have a ridiculous amount of algs, I don't know if anybody fancies doing the work of inputting them all (any of them??) into the ss wiki.

 

[oprah62]

I don't have much to add personally to this project but I recently found this Polish Algorithm Database. It seems to have a ridiculous amount of algs, I don't know if anybody fancies doing the work of inputting them all (any of them??) into the ss wiki.

Wow that database is insanely good. I love how they orangize it all. Thanks for sharing.
edit:500th post!

 

[StachuK1992]

I'm pretty sure uploading is possible - just log into the wiki and there should be an 'add' link in the 'Mode' section.

I can't seem to add CPLS. :/
That, or I'm just dumb.

 

[Cride5]

I'm pretty sure uploading is possible - just log into the wiki and there should be an 'add' link in the 'Mode' section.

I can't seem to add CPLS. :/
That, or I'm just dumb.

Yup, CPLS hasn't been added as a 'group' yet. Each time a new group is added a developer needs to define it within the system. The reason for this is that the validation system needs a semantic definition of the group to allow it to recognise algs which belong to that group. Without this none of the automated verification/organisation would work.

@Escher: That polish database is seriously cool! This is exactly what I tried to achieve with the wiki database, but the big problem is I'm very short of spare time these days.

If there are any php/mysql developers interested in getting involved in the project give me a shout. Looking at the huge number of algs already contributed to the wiki, I think this is a really worthwhile project.


EDIT: If folks fancy populating the database there is a 'bulk' upload function. Just log into the wiki, select the alg set, and then click on 'bulk'. For example:
http://www.speedsolving.com/wiki/index.php/Special:AlgDB?mode=bulk&view=default&puzzle=3&group=CLS
(requires wiki login)

 

[Anthony]

omfg..
I'm learning EG2..
now..

 

[oprah62]

omfg..
I'm learning EG2..
now..

Have you found any good algs? I've tried to CE them but most of them are just bad.

 

[danthecuber]

Bump

If you all aren't aware...I tried uploading CLS and ELS algs, and there are still swapped. What's worse, in the ELS section of the database, which is holding CLS algs, only has 23 cases there, but there's supposed to be 104 cases. I uploaded alg for every CLS case, every alg was successful, but it looks like they went to the wrong cases, or something like that. No new cases were created, when they were supposed to. The same thing is happening with the CLS section, only 5 cases are there and there is supposed to be 21 cases. WE NEED YOU CRIDE5!!!