Spaceship Discussion Thread

wwei47 · Post by **wwei47** » February 27th, 2021, 9:45 am

Hunting wrote: ↑
February 27th, 2021, 1:06 am
Symetric searches and asymmetric searches of the same period are at the same difficulty. I don't see where are you coming from.

Example:
c/6 w12a
c/12 w12e

EvinZL · Post by **EvinZL** » February 27th, 2021, 10:38 am

Symmetric search is roughly as hard as asymmetric search at the same period but half the width

Hunting · Post by **Hunting** » February 27th, 2021, 11:49 am

wwei47 wrote: ↑
February 27th, 2021, 9:45 am

Hunting wrote: ↑
February 27th, 2021, 1:06 am
Symetric searches and asymmetric searches of the same period are at the same difficulty. I don't see where are you coming from.
Example:
c/6 w12a
c/12 w12e

What's e? Actually, symmetry options = different boundary handling, and it does not affect the number of unset cells at all.

MathAndCode · Post by **MathAndCode** » February 27th, 2021, 12:03 pm

Hunting wrote: ↑
February 27th, 2021, 11:49 am
Actually, symmetry options = different boundary handling, and it does not affect the number of unset cells at all.

Yes, but it pairs unset cells so that once you know which state one of them has, you know that the other must have the same state, so in terms of how many variables must be solved, it's like the number of unset cells has been halved.

wwei47 · Post by **wwei47** » February 27th, 2021, 2:30 pm

Hunting wrote: ↑
February 27th, 2021, 11:49 am
What's e?

Even symmetry. w12a means 12 unknown cells per row per generation. w12e means 6 unknown cells per row per generation, because it's reflected to the other aide.
EDIT: Fixed quote tags.

Hunting · Post by **Hunting** » February 27th, 2021, 11:00 pm

MathAndCode wrote: ↑
February 27th, 2021, 12:03 pm

Hunting wrote: ↑
February 27th, 2021, 11:49 am
Actually, symmetry options = different boundary handling, and it does not affect the number of unset cells at all.
Yes, but it pairs unset cells so that once you know which state one of them has, you know that the other must have the same state, so in terms of how many variables must be solved, it's like the number of unset cells has been halved.

wwei47 wrote: ↑
February 27th, 2021, 2:30 pm

Hunting wrote: ↑
February 27th, 2021, 11:49 am
What's e?
Even symmetry. w12a means 12 unknown cells per row per generation. w12e means 6 unknown cells per row per generation, because it's reflected to the other aide.
EDIT: Fixed quote tags.

I think you have trouble understanding symmetries. It's ran like an asymmetric search, and then reflected, so w9 u is actually 18 cells wide as asymmetric.

wwei47 · Post by **wwei47** » February 27th, 2021, 11:25 pm

Hunting wrote: ↑
February 27th, 2021, 11:00 pm
I think you have trouble understanding symmetries. It's ran like an asymmetric search, and then reflected, so w9 u is actually 18 cells wide as asymmetric.

I don't know what you're talking about. If I have to search w18e, I have to tell JLS to create an 18-by-whatever board with the last few rows unset to prevent edge effects at the bottom, then specify vertical symmetry.

bubblegum · Post by **bubblegum** » February 27th, 2021, 11:28 pm

wwei47 wrote: ↑
February 27th, 2021, 11:25 pm
I don't know what you're talking about. If I have to search w18e, I have to tell JLS to create an 18-by-whatever board with the last few rows unset to prevent edge effects at the bottom, then specify vertical symmetry.

w18e is width 36—in terms of free cell groups, it's equivalent to a w18a search.

MathAndCode · Post by **MathAndCode** » February 28th, 2021, 12:40 pm

Hunting wrote: ↑
February 27th, 2021, 11:00 pm
I think you have trouble understanding symmetries. It's ran like an asymmetric search, and then reflected, so w9 u is actually 18 cells wide as asymmetric.

I think that that's where we're getting confused. I meant the width after one half has been mirrored to the other side.

wwei47 · Post by **wwei47** » February 28th, 2021, 12:45 pm

MathAndCode wrote: ↑
February 28th, 2021, 12:40 pm
I think that that's where we're getting confused. I meant the width after one half has been mirrored to the other side.

Yeah, I use JLS a lot, especially where it's not really meant to be used (high-period spaceship searches and catalysts come to mind), and JLS needs you to put in the full width.

gfind, ntzfind, and qfind, on the other hand, only want the half width,

LaundryPizza03 · Post by **LaundryPizza03** » March 1st, 2021, 11:13 am

My computer does not have enough RAM to continue any further on c/8d with ikpx2. Unfortunately, the full results are too big to share even as a ZIP file, but I reached width 10, and just as the heap size had peaked at around 34.4M.

Post by **dvgrn** » March 1st, 2021, 12:08 pm

LaundryPizza03 wrote: ↑
March 1st, 2021, 11:13 am
Unfortunately, the full results are too big to share even as a ZIP file, but I reached width 10, and just as the heap size had peaked at around 34.4M.

Maybe multiple ZIP files? Or check in to a GitHub repo or something?

Last time I experimented with this, it seemed like moderators could successfully attach bigger ZIP files to posts than anyone else could, so let me know if I can help make files available (at least, if you think someone else might find them useful as a starting point.)

LaundryPizza03 · Post by **LaundryPizza03** » March 1st, 2021, 2:04 pm

dvgrn wrote: ↑
March 1st, 2021, 12:08 pm

LaundryPizza03 wrote: ↑
March 1st, 2021, 11:13 am
Unfortunately, the full results are too big to share even as a ZIP file, but I reached width 10, and just as the heap size had peaked at around 34.4M.
Maybe multiple ZIP files? Or check in to a GitHub repo or something?

Last time I experimented with this, it seemed like moderators could successfully attach bigger ZIP files to posts than anyone else could, so let me know if I can help make files available (at least, if you think someone else might find them useful as a starting point.)

The ZIP file is 660 MB. Not a good sign.

LaundryPizza03 · Post by **LaundryPizza03** » March 2nd, 2021, 9:48 am

This is my latest and longest (2,1)c/8 partial, and the last at width 20 for ikpx2. I've started getting "decision limit" warnings regarding cadical, however.