Sierpinski Network

[qqwref]

For the curious, my current method is to use only corner presses to make all the groups of three non-corners in the smallest triangles either all black or all white. Then the second step is to solve everything without corner presses, and apply parity if necessary. I usually start from the top and work my way down.

You could try to predict the parity by counting the number of white corners plus the number of corner presses your plan would involve to solve the first step. If you typically start from the top, and typically avoid pressing the topmost corner, your parity fix would be to press the top corner at the very start. It'd be tough, though, since you'd have to essentially lookahead the entire first step.

EDIT: Hey, I managed to predict parity a couple times in a row (on level 3 which is almost certainly the maximum this is possible for). It wasn't always within 15 seconds, though, and I mispredicted it a few times after that. Just saying, it is definitely doable.

 

[Stefan]

474.83 for level 6 (and now I'll probably never want to do this again, I'm just glad I didn't have parity)

 

[emolover]

474.83 for level 6 (and now I'll probably never want to do this again, I'm just glad I didn't have parity)

You can only get parity on 3, 5 and I guess 1.

 

[ben1996123]

1:31.90 on n=4.

Edit: 1:17.27

 

[mr. giggums]

4 clicks on level 3 wtf o_0
And your other times are ridiculous too, what's your method?

Also: 174.73 level 6. Oddly enough I can't get anywhere near 1/3 of this time on level 5.
And 5.80 lucky level 3, plus some sub20s (lol) with parity.

I just realized that it was level two not three.

EDIT: 32.50 on the real level three.

 

[emolover]

15.90 on n=4

With average with solves without parity I average like 23.

 

[qqwref]

I did 100 level 3 solves with parity prediction Made some mistakes, but fortunately not too many.
(Yeah, this is WITHOUT choosing only parity-free solves - I solved 100 scrambles in a row.)

avg5: 7.48 (σ = 0.71)
6.50, 8.17, 7.77, (20.40), (6.47)

avg12: 8.48 (σ = 0.84)
(6.93), 8.80, (21.83), 9.40, 8.53, 8.37, 9.60, 9.50, 6.97, 7.77, 8.33, 7.50

avg100: 10.85 (σ = 3.55)

Spoiler

11.53, 20.90, 14.60, 12.20, 7.67, 9.57, 14.30, 12.00, 12.50, 13.47, 12.03, 9.90, 23.00, 8.40, 10.30, 10.13, 10.73, 8.20, 11.20, 10.67, 7.83, 9.97, 9.33, 7.83, 18.17, 13.17, 14.63, 9.37, 10.03, 7.60, 9.80, 8.43, 12.30, 9.60, 7.93, 18.73, 11.30, 11.13, 8.00, 9.93, 11.63, 10.67, 8.57, 10.30, 6.87, 9.27, 7.93, 11.00, 11.00, 8.53, 10.47, 11.53, 9.73, 6.93, 8.80, 21.83, 9.40, 8.53, 8.37, 9.60, 9.50, 6.97, 7.77, 8.33, 7.50, 14.23, 13.97, 9.83, 8.10, 9.23, 9.13, 17.03, 7.33, 9.57, 6.50, 8.17, 7.77, 20.40, (6.47), 8.40, 8.63, 15.43, 8.50, (24.90), 9.73, 7.07, 10.93, 10.07, 11.03, 11.00, 8.83, 17.40, 10.70, 8.23, 11.53, 12.03, 8.13, 11.10, 10.90, 22.97

EDIT:

15.90 on n=4

With average with solves without parity I average like 23.

Crazy. What's your technique? I'm getting like 40ish (probably because of the teeny tiny dots).

 

[emolover]

EDIT:

Crazy. What's your technique? I'm getting like 40ish (probably because of the teeny tiny dots).

Your right that is crazy. I truly don't even know that got there because I was doing 3x3 at the time/eating dinner. I don't even know...

Maybe I meant n=3? But I dont know because my best is 11.XX...

But I will make a video showing my technique.

 

[aronpm]

243.83 for n=6

Spoiler

my timer was running super slow it was actually like 15 minutes

 

[RNewms27]

243.83 for n=6

243.93/243 ~ 1 sec/puzzle

my timer was running super slow it was actually like 15 minutes

~900/243 ~ 3.7 sec/puzzle

 

[qqwref]

I'm not sure what you mean by "per puzzle". I think n=6 has 1098 dots (each is 3 times the previous one, minus 3), so maybe that's more useful.

Also: wow, this timer is spectacularly bad. I solved the n=6 and got a time of 153.67, but it was timed on qqtimer as well, and there it was a 6:49 (this is probably much more realistic). The js is heavily obfuscated, so I have no idea what was done for the timer, but I'd really suggest using something like the current system clock (which qqtimer uses).

 

[Meep]

I'm not sure what you mean by "per puzzle". I think n=6 has 1098 dots (each is 3 times the previous one, minus 3), so maybe that's more useful.

Also: wow, this timer is spectacularly bad. I solved the n=6 and got a time of 153.67, but it was timed on qqtimer as well, and there it was a 6:49 (this is probably much more realistic). The js is heavily obfuscated, so I have no idea what was done for the timer, but I'd really suggest using something like the current system clock (which qqtimer uses).

Yeah that was with an issue of me not knowing how, but Justin just showed me and I'll probably fix it when I get home later.

 

[qqwref]

Awesome, glad to hear you're updating it. While you're at it, would it be possible to increase the size of the dots on the bigsierpinski one (at least for n<6)? They're pretty hard to hit and I'd like to be able to get times comparable to the smaller app.

 

[Meep]

Yeah I can experiment and try to scale them with n.

 

[Meep]

I fixed the timer to use the system clock, and it now uses the big nodes for everything but n = 6.

I also removed the 'bigsierpinski.html' page and just made the regular one go up to 6. I made the big one just to watch my script generate higher iterations of the fractal, not expecting people to want to play them. =P

Link: http://meep.cubing.net/html5/sierpinski.html

 

[aaronb]

How do you fix it when you just have one corner still white on n=3?

 

[Stefan]

n = 4
43.26, (38.63), (48.61), 42.88, 47.11 => 44.42

Definitely my favorite size.

 

[Meep]

How do you fix it when you just have one corner still white on n=3?

See the previous posts for qq's parity fix/explanation. It happens on the odd-iterations (1,3,5).

 

[qqwref]

Yeah, n=4 is pretty fun. Not too ridiculously large, not too small, and no parity.

27.45, (24.76), 26.77, (30.63), 27.79, 26.02, 25.51, 27.97, 24.82, 27.18, 29.31, 25.92 => 26.87

EDIT: 84.72 on n=5.

n=6 is broken now...

 

[benskoning]

my PB 3.14