Category talk:Patterns that can be constructed with 1000 or more gliders

From LifeWiki
Jump to navigation Jump to search

Observe that this can now be renamed to 'Patterns that can be constructed with 329 gliders'. Calcyman (talk) 22:46, 10 June 2018 (UTC)

Indeed, we should move the page eventually. I think we should wait at least a few days in case small optimizations further reduce the required number of gliders.
~Sokwe 06:28, 11 June 2018 (UTC)
Technical note --- category pages can't be moved, it'll have to be created anew under a new name. And yes, I agree, waiting for a little longer until the dust settles is sensible. (Also, for anyone who might be reading this and wondering, the above is in reference to this, primarily.) Apple Bottom (talk) 07:11, 11 June 2018 (UTC)
I'm finding it obscurely troublesome to mention 329 gliders in this context. The whole reverse-caber-tosser episode is pointing out the awkward fact that, from now on, there will be a distinction between "Patterns known to be constructible with X gliders" and "Patterns for which an X-glider construction can be exhibited" -- for all X >=329.
These "Patterns that can be constructed with X gliders" categories in the LifeWiki are really supposed to be in the second of these groups, but that group was invented before this crazy universal-constructor idea came along, so the wording accidentally implies that it's talking about the first group.
... On the other hand, 329 gliders is really a better cut-off than 1000 as far as creating categories is concerned. Any object that takes more than a few hundred gliders to construct will probably be the only object in its category. So really in practice I won't mind if the top glider-expense category turns into "Patterns that can be constructed with exactly 329 gliders" -- as long as there's a link to a reverse-caber-tosser article somewhere to explain why the actual glider recipes given for the patterns listed in that category are mysteriously much higher than 329. Dvgrn (talk) 21:55, 12 June 2018 (UTC)
That's sensible all around.
Looking at Category:Patterns constructible by a given number of gliders, there's no categories between "...276 gliders" and "...1000 or more gliders" anyway, so setting the cut-off to 329 wouldn't affect any articles other than the four macro-ships in the residual category (Demonoid, Gemini, Orthogonoid, parallel HBK).
We wouldn't need to edit any of the articles in question; the pattern infobox templates could be changed to default to 329 for the category if synthesis= was passed a higher number. (The infobox itself could still display the actual number passed, or it could also display 329 if the number passed exceeded that. Our call.)
I do agree that this would have to be explained well, with a prominent link (ideally rendered in large, friendly letters) to an explanatory article on reverse caber-tossers which would of course also have to be written. And said link/explanation should be both in Category:Patterns constructible by a given number of gliders and Category:Patterns that can be constructed with 329 gliders.
A separate explicit note in our documentation wouldn't be necessary if the templates defaulted to Category:Patterns that can be constructed with 329 gliders; users wouldn't see any redlinks to categories that shouldn't exist. But a note in the template documentation that this is the expected and intended behavior, again with a prominent link to an explanation of reverse caber-tossers, would be good.
Speaking of the number 329, perhaps that should be made stuck in a template somewhere deep in the bowels of the wiki as well, so as to allow easier tweaking when (not if!) this goes down in the future.
FWIW, re: the distinction between "known to be constructible with 329 gliders" and "explicitely constructible with X gliders", where X>329 --- the infobox templates could of course also be rigged to include a footnote remarking on this when X is indeed >329. Something along the lines of "†A 329-glider synthesis exists, but has not been explicitely constructed. See reverse caber-tosser for details" or so.
How's that sound? Apple Bottom (talk) 19:36, 13 June 2018 (UTC)
Yes, making the 329 defined in a single place would be useful if it ever gets reduced further (although I must say Goldtiger did an incredibly good job at this -- some of the quartets of blocks in the p256 gun are constructed by 6 gliders instead of 8, for instance, and there are a few cases where pairs of blocks are constructed by 3 gliders).
What is impressive is that 329 is so close to Gustavo's initial estimate of 386 gliders. Of course, using 2015 technology (no bumpers, no 2-engine Corderships) we wouldn't have been able to get anywhere near 386.
Does someone want to create the article for reverse caber-tosser or Formichev universal constructor? I believe the latter refers to the entire assembly, whereas the former doesn't include the shotgun and receding BLSE. Calcyman (talk) 08:35, 14 June 2018 (UTC)
I think it would still be useful to mention the explicit syntheses with >329 gliders, because those can actually be shown in RLE format and used in guns etc. Also, that should be "Fomichev" universal constructor (don't propagate Gustavo's typo into the wiki...). 77topaz (talk) 09:00, 14 June 2018 (UTC)
Having read the entire thread now, I can confidently say I feel thoroughly unqualified to write either article. ;) I'll have to leave these to the resident experts.
I do agree with 77topaz that listing explicit syntheses still has value. See Dave's observations above one the distinction between "known-to-be-constructible-with-X-gliders", and "having-an-explicit-X-glider-recipe"; also see my remarks on how this could be handled, template-/infobox-wise. Apple Bottom (talk) 15:57, 14 June 2018 (UTC)
I've created a page for reverse caber-tosser, including its application. It is probably wisest to redirect the other page to point to this; the topics are too similar to warrant distinct pages. Calcyman (talk) 18:15, 15 June 2018 (UTC)

A lower lower bound

The number of 329 appears to be on the verge of indeed being reduced, to something in the neighborhood of 65 -- a very impressive improvement indeed, BTW --, which makes me wonder how to best proceed regarding our category infrastructure.

329 was a "convenient" number insofar as that we didn't have any patterns not already in the residual category (currently still "1000 or more gliders"). 65, OTOH, would obsolete about one third of our construction categories.

There already seems to be consensus to keep the "best known/practical/reasonable" number of gliders needed to construct a pattern in the infobox, with a footnote explaining theoretical constructibility using fewer gliders if the number exceeds the current lower bound (let's call that "B", BTW). I'd like to propose that something similar be done for categories.

I don't think it would be useful to duplicate the category tree outright, separating "can be constructed using X gliders" and "has an explicit recipe using X gliders". What I'd instead like to suggest is repurposing our existing categories to refer to the latter, rather than the former, with an explanatory note at the top of each category if X exceeds B, the current lower bound. [Technical side note: this could be made to display automatically by rigging Template:SynthesisCategory, and if B had its own template, then again, we'd only need to flip one switch if a better bound was obtained, and everything would update automatically.]

The rationale for this is simple: our categories are meant to assist users in finding information they need. And I think that users scouring the category tree are more likely to be interested in practical recipes than they are in theoretical results --- those that were would hopefully take note of the prominent explanatory notes.

The alternative would be to a) duplicate the category tree, which I think isn't worth the effort, and which could lead to confusion; or b) keep the category tree as is, which (as per above) I think would be optimizing for the uncommon use case.

When going ahead with this we might want to rename our categories, so that instead of e.g. Category:Patterns that can be constructed with 78 gliders we'd have Category:Patterns with an explicit 78-glider construction or so. This would be a bit of work, but not much --- even though categories can't be moved, most of the actual work is done behind the scenes by the category template, so all we'd need to do is create the new categories with a template call and delete the old ones.

Thoughts? Apple Bottom (talk) 14:06, 22 June 2018 (UTC)

I agree with your way of doing it; it makes it the easiest to use the wiki. This might be a bit picky, but I'm not sure if "has an explicit recipe" is the right description. The glider syntheses for the reverse-caber-tosser will (I think) be "explicit" in the sense that you can give the exact coordinates of each glider. You'll be able to store them in a .rle or .mc file. You might even be able to open them in Golly and zoom in on each individual glider. The only thing you won't be able to do is run them to completion in Golly. I'm not quite sure what a better term would be. Perhaps "practical" vs "impractical"? --Macbi (talk) 16:07, 22 June 2018 (UTC)
We definitely shouldn't duplicate the categories, and even renaming them might be unnecessary. I think it's sufficient to include (transclude?), into the body of the category, a note explaining the existence of the reverse caber tosser. Oscar's suggestion in the forums of using a parenthetical N after the practical cost in the pattern template is the best suggestion I've seen. Calcyman (talk) 17:52, 22 June 2018 (UTC)
Macbi -- good point about "practical" perhaps being more appropriate than "explicit". Of course, what I meant by "explicit" was "already has a file showing the construction" (not necessarily on the wiki); so long as nobody actually constructed a B-glider synthesis using the RCT mechanism, it wouldn't be explicit. But you're right, of course: there's nothing that would keep people from explicitely constructing such syntheses. So yes, "practical" might be the better term.
Calcyman --- this seems like a sensible way to go forward for now; let's see if we can get away with making minor adjustments before committing to making major ones. :) I'll tweak the pattern templates/categories appropriately in a moment. Apple Bottom (talk) 12:08, 23 June 2018 (UTC)
EDIT: I've made the appropriate tweaks to Template:PatternDownload and Template:SynthesisCategory. The "best known lower bound" is pulled in from Template:Reverse caber tosser/Lower bound (this can be moved/renamed if necessary); I've put in the number 329 there, so please edit it once the dust settles and we've got a new, better lower bound. ;) Apple Bottom (talk) 12:32, 23 June 2018 (UTC)
The lower bound has improved to 59 thanks to Chris Cain. I've edited the reverse caber-tosser article, but the Template:Reverse caber tosser/Lower bound is transcluded into the Main Page via 'in the news' therefore is beyond my powers to edit.
This improves the glider count of surprisingly many syntheses. Even more surprisingly, Macbi noted that there provably exist knightships with smaller total population than Sir Robin. Calcyman (talk) 13:00, 28 June 2018 (UTC)
Apologies for the mix-up with the template, I didn't consider that this would happen. Fixed now, I hope. Apple Bottom (talk) 17:26, 28 June 2018 (UTC)