[2010 Ideas] New scramblers

qqwref (2010-01-20 08:57:43 +0000)
I think it is time to update our scramblers for the new year. There are two main changes: - The current Square-1 scrambler does not include images. I suggest the following one, which does: http://mzrg.com/miniSites/scramblers/sq1_wca.html. These images are completely generated by Javascript, so it will run fine and only requires one external file (wz_jsgraphics.js). The images are not graphically perfect but they are more than good enough to allow the scrambler to tell that they have correctly applied the scramble, something which is very important in Square-1 and which was not possible until now. The scrambler is Jaap's, that is, the same as the current one. I slightly modified the code so that some scrambles have (0,0) after them - this signifies that the last slice turn should be "canceled out" and thus ensures that all scrambles are 40 moves htm. I think this is an important change. - We should switch to a random-state, optimal (or near-optimal) Pyraminx scrambler. Here is an optimal one: http://mzrg.com/miniSites/scramblers/sc ... nx2009.htm. It would be nice to improve the images, using javascript drawing routines as with the Square-1 scrambler, but this hasn't been implemented yet. In addition, it would be nice to modify the 2x2 and Pyraminx scramblers, which are optimal, to instead always generate 11-move scrambles. This is a simple change to the scramble generation function, and it ensures that very easy scrambles cannot be easily detected. If it seems like this might be a bit too much work for the scramblers, an alternate idea is to generate scrambles of 8 through 11 moves, rather than 0 through 11 (this is also a very simple change in the code). That way, most scrambles would be unchanged, but the very easy ones would be shown with longer solutions, so that competitors and judges would not know about very easy scrambles beforehand. I would also like to take this time to rehash a separate proposal I made earlier: viewtopic.php?f=4&t=546. The idea is to discard scrambles that are too close to solved; ignore the part about Clock, since I realized later that that was silly. But I think it would still be useful to discard 2x2, 3x3, Pyraminx, Square-1 scrambles that are too close to solved, to eliminate cases where many people at a competition get very nice single times which cannot be beaten without a similarly ridiculous scramble. I don't expect this to be used, but it would be nice to open it up for debate.
MadsMohr (2010-01-20 10:57:52 +0000)
Very nice work! The current [url=http://www.worldcubeassociation.org/regulations/scrambles/scramble_pyraminx2009.htm:mizdfsf4]pyraminx scrambler[/url:mizdfsf4] has improved images. Perhaps you could use that?
deadalnix (2010-01-20 13:54:35 +0000)
I already have a javascript 2x2x2 scrambler generator, with optimal solution. It has a god algorhytm precalculation, so can avoid too easy scramble easily. the code run fine on firefox and chrome, but is really slow under IE due to his poor javascript engine. I can release it if the WCA is interested in.
qqwref (2010-01-20 23:33:35 +0000)
Here's what should be the new Pyraminx scrambler: http://mzrg.com/miniSites/scramblers/sc ... nx2010.htm. I made two changes from the one in the first post: - Changed the images to the ones we use now, they are much prettier. - Modified the scrambler so it always generates sequences of at least 7 moves (not counting tips). It is still completely random-state but this will make it less obvious when a very easy scramble comes along.
Ron (2010-02-14 09:18:33 +0000)
Hi Michael, Thanks for your contributions. Your programs will be part of WCA Regulations 2010. Please explain why your Square-1 scramble often adds (0,0). This will only undo the last / of the previous move. Thanks, Ron
Pedro_S (2010-02-14 13:06:00 +0000)
I say we should use the / again. It eliminates confusion on wheter to do the last / or not, and doesn't present any problems, in my opinion.
jbcm627 (2010-02-14 21:19:49 +0000)
I'd also like to address the following usability issues: - In most browsers default settings, some printed scrambles don't appear on a single page. - In most browsers default settings, no colors will show up when printing, since the images rely on the css background-color property. The following programs fix the above issues by using images instead of background colors, and by adjusting the height of the image output: 2x2: http://koii.cubingusa.com/lib/scrambles ... e_222.html 4x4-7x7: http://koii.cubingusa.com/lib/scrambles ... _cube.html pyra: http://koii.cubingusa.com/lib/scrambles ... /pyra.html sq1: http://koii.cubingusa.com/lib/scrambles ... e/sq1.html (the [url=http://koii.cubingusa.com/lib/scrambles/standalone/wz_jsgraphics.js:1sxoncan]wz_jsgraphics[/url:1sxoncan] file was what was modified) The following images are needed: http://koii.cubingusa.com/lib/scrambles ... /scrbg.zip And the pyraminx and sq-1 scramblers both use Michael's new code, with slightly modified image generation. I would also like to request that any future scramblers be written as javascript classes. I would like to work on converting the current scramblers soon, but it is rather tedious/boring work. The point of this would be to have one program capable of generating any number of scrambles for any cube, as discussed here: viewtopic.php?f=4&t=846&p=4519&sid=07953d7fcdb85a5ca6c4a378da364db1#p4510
qqwref (2010-02-15 10:13:49 +0000)
[quote="Ron":j6vuzsvd]Please explain why your Square-1 scramble often adds (0,0). This will only undo the last / of the previous move.[/quote:j6vuzsvd] Exactly. The (0,0) undoes the last / and thus provides a scramble of exactly 40 half turns as defined in the regulations; if we never allowed scrambles to finish with a (0,0) there would be no way to make some scrambles 40 turns. So this is required because of the decision to abolish the /. The only way to ensure that all scrambles are 40 moves again, without using (0,0) moves, is to put the / back in.
Ron (2010-02-24 22:38:28 +0000)
Hi Michael, Is there a final version of the Square-1 scrambler that we should use? Thanks, Ron
qqwref (2010-02-25 00:51:43 +0000)
Either mine or jbcm's are working final versions. However, I would like to suggest adding the / move back in, if possible. I can quickly edit the scrambler to include this if you want.
Pedro_S (2010-02-25 11:25:56 +0000)
[quote="qqwref":29hp9crz]Either mine or jbcm's are working final versions. However, I would like to suggest adding the / move back in, if possible. I can quickly edit the scrambler to include this if you want.[/quote:29hp9crz] Agreed. The scramble string doesn't get much longer, and it eliminates confusion. EDIT The images are not working for me in the official link http://www.worldcubeassociation.org/reg ... m=5&len=40
Ron (2010-02-27 23:15:40 +0000)
Can someone please send me the correct files?
jbcm627 (2010-02-28 05:45:31 +0000)
[quote="Ron":12rf37nf]Can someone please send me the correct files?[/quote:12rf37nf] Here is a zip file containing necessary scramble files: http://koii.cubingusa.com/lib/scrambles/standalone.zip If the sq-1 needs the / added back in, qq will need to give you a new sq-1 scramble file, but I would like to keep the changes I made in the wz_graphics.js file to output images instead of background colors.
Ron (2010-04-11 20:35:19 +0000)
Thanks for posting the correct programs! Regarding the additional (0,0) move. The most important thing is that we all use the same scrambling strategy. For this we have a program. I like the new version very much. Great contribution to our community! For me the current discussion on the length of the scramble is only important because the regulations are a bit of a contradiction. More important though is that we should move to a random position scramble program, as soon as possible. Until then we have to live with the current software, and I dislike the idea of changing the notation only because of an inaccuracy in the software. Bottom line is that we have a standard solution for scrambling and that the quality of the scrambling is good enough unless proven otherwise or unless improved by a true random position scramble program. Have fun, Rn
DanCohen (2010-04-11 21:50:48 +0000)
[quote="Ron":3oq8odvc]Thanks for posting the correct programs! Regarding the additional (0,0) move. The most important thing is that we all use the same scrambling strategy. For this we have a program. I like the new version very much. Great contribution to our community! For me the current discussion on the length of the scramble is only important because the regulations are a bit of a contradiction. More important though is that we should move to a random position scramble program, as soon as possible. Until then we have to live with the current software, and I dislike the idea of changing the notation only because of an inaccuracy in the software. Bottom line is that we have a standard solution for scrambling and that the quality of the scrambling is good enough unless proven otherwise or unless improved by a true random position scramble program. Have fun, Rn[/quote:3oq8odvc] so do we do a slice move at the end of a scramble or not? Having a (0,0) at the start isn't necessarily the issue, its more about the end of the scramble in my opinion.
Ron (2010-04-12 05:28:14 +0000)
[quote:5ompimao]so do we do a slice move at the end of a scramble or not?[/quote:5ompimao] We use the current program 'as is' until we have a better solution. The quality of the scrambles is an issue until we have a random position scramble program.
WydD (2010-05-01 19:10:55 +0000)
Hi everyone, The new scrambler of the SQ1 is great but very slow even on recent navigators due to the massive charge of the library that generates graphics. I made a really slight modification to the code in order to use svg/vml technologies that are far quicker. I used the following library : http://raphaeljs.com/ which supports Firefox 3.0+, Safari 3.0+, Opera 9.5+ and Internet Explorer 6.0+ and other svg-friendly navigators. The rendering is even also prettier than before :) You can find the code here if you like, http://loic.doesntexist.com/cube/scramb ... 0_wydd.htm
deadalnix (2010-05-02 02:03:31 +0000)
It doesn't work on IE6 but very nicely on chrome and firfox 3.5 for as far I have tested. As long as IE6 is clearly outdated (and tools like http://code.google.com/intl/fr/chrome/chromeframe/ can allow it to use a modern renderer) I think this isn't an issue.
WydD (2010-05-02 12:50:08 +0000)
Well it's weird that the code does not work on IE6, it should use vml. Unfortunately, I can't test it correctly here (i'm on a mac with a teaked ie6). I'll look into it with a correct ie6 at work :)