Glider Guns of large periods

For discussion of specific patterns or specific families of patterns, both newly-discovered and well-known.
User avatar
Kazyan
Posts: 1247
Joined: February 6th, 2014, 11:02 pm

Re: Glider Guns of large periods

Post by Kazyan » March 26th, 2014, 6:21 pm

The deal isn't that these guns aren't worthwhile or useful, but that it's very simple to use a 2-state machine to thin out a known gun. The p512 in the gun collection is "trivial" as well (it just points two p256s at each other in a standard way), but Herik's is smaller. It doesn't appear on the list because the 2-state machine wasn't known at the time.
Tanner Jacobi
Coldlander, a novel, available in paperback and as an ebook. Now on Amazon.

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Glider Guns of large periods

Post by dvgrn » March 26th, 2014, 10:44 pm

Herik Zorneck wrote:If they were so trivial, these gliders guns should be appear in the list and this is not the case.
Unfortunately Paul Callahan's web pages are just an archive copy -- that list hasn't been updated in well over a decade, I believe. Jason Summers has the most recent version of this glider gun collection, with no missing periods from 14 to 1000.

The larger collection, which isn't exhaustive, has an old-style p1024, but nobody ever bothered to build the p2048, p4096, or p8192. Here again this was because the higher periods were thought to be "trivial", not because nobody knew how to build one. More accurately, nobody had any ideas that could improve on the obvious solution, which at the time was to add a p8 glider-to-Herschel converter to a p256 gun, and attach as many Herschel period-doublers or period-quadruplers as you needed to get the period you wanted. [Or there were lots of other non-Herschel x2 and x4 tricks, including oscillating pulse dividers. But I don't remember there being a compact period-2^N version.]

Nowadays there are a lot more ways to try for a new bounding box record for high periods. For example, notice that the p1024 in Jason's collection also uses two p256 guns, but collides the glider streams to get an output Herschel and glider in a period-quadrupling reaction instead of a period-doubling one:

Code: Select all

#C p1024 gun = p256x4
x = 106, y = 52, rule = B3/S23
88b2o$88b2o5b2o$95b2o$16b2o$9b2o5b2o$9b2o53b2o8b2o17b2o$64b2o9bo17b2o$
75bobo21b2o$11b2o17b2o8b2o34b2o21b2o$11b2o17bo9b2o16b2o$5b2o21bobo27b
2o$5b2o21b2o32b2o$46b2o14b2o$46b2o$42b2o$42b2o$57b2o$57b2o41bo$98b3o$
47b2o5b2o41bo$5bo41b2o5b2o41b2o$5b3o$8bo$7b2o8$85bo18b2o$85b2o17b2o$
84bobo$2o$2o17b3o51bo$19bo51b2o26b2o$20bo10bo39b3o2b2o21b2o$31bo38bo5b
2o25b2o$5b2o26bo37bo4bo26b2o$5b2o21bo3b3o27b2o7bo3bo9b2o$b2o25b2obo3bo
26b2o9bo6b2o3bobo$b2o27b5o33b2o8bo2bo5bo9b2o$19b2o10b3o8b2o24b2o8bobo
6b2o8b2o$18bobo3b2o16b2o$7b2o9bo6b3o8b2o$7b2o8b2o7b2o8b2o28b2o$66b2o5b
2o$73b2o$38b2o20bo$31b2o5b2o21bo$31b2o26b3o!
For high enough periods it might be possible to string together semi-Snarks (i.e., "two-state machines") and Snarks on two or more different outputs of a single p256, in such a way that the outputs collide in some way similar to the above, and cause a further period-multiplication. Or, more likely, arrange a string of Snarks or semi-Snarks so that the output interferes with the original stream and produces a higher period that way.

For periods like 8192 it will be quite a trick to come up with the smallest possible solution by bounding box, though! Jason's version of the gun collection was very highly optimized, with all sorts of tricks at the edges to squeeze out a few more rows or columns from a more standard gun.
Herik Zorneck wrote:According to your definition of trivial, everything was built from something already known, using one block or fish hook, for example, would be a trivial pattern.
The definition of "trivial" is indeed very silly if you take it to that extreme. So don't take it to that extreme. Codeholic is trying to explain a subtle but real distinction between patterns that "a lot of people already know how to make", and patterns that contain "something new that nobody has seen before" -- i.e., a technological improvement, major or minor, that might be usable in other similar situations. In my reading he's not being rude about this -- unlike a few other people on the conwaylife.com forums in similar cases, by the way! -- just short and to the point.

While we're talking trivia, here's a trivial modification of Herik's p512 gun, which makes it 225 cells smaller measuring by bounding box. You can actually re-use one of the blocks from an R64 conduit as part of the semi-Snark reflector!

Code: Select all

x = 61, y = 54, rule = B3/S23
57bo$55b3o$54bo$46b2o6b2o$46bobo$31b2o14bo$31bo2bo3b2o$38b2o12b2o$27bo
24b2o5b2o$25bo7b2o3bo20b2o$7b2o8b2o12b2o4b2o$7b2o9bo5bo11b3o$18bobo3bo
3bo6b2o5b2o$19b2o7bo6bob2o3b2o$b2o22bobo8b3o$b2o54b2o$5b2o26b2o22bobo$
5b2o26bo25bo$34bo24b2o2$32bo$2o33bo$2o30bo2bo$34b3o$34b3o$14bo18bo2bo
23bo$13bo19bobo22bobo$13b3o17b3o23b2o6$7b2o$8bo$5b3o$5bo41b2o$47b2o4$
42b2o$42b2o$46b2o$46b2o$5b2o21b2o$5b2o21bobo$11b2o17bo9b2o$11b2o17b2o
8b2o3$9b2o$9b2o5b2o$16b2o!

User avatar
Herik Zorneck
Posts: 108
Joined: September 8th, 2013, 12:09 am

Re: Glider Guns of large periods

Post by Herik Zorneck » March 27th, 2014, 1:08 am

dvgrn wrote:
Herik Zorneck wrote:If they were so trivial, these gliders guns should be appear in the list and this is not the case.
Unfortunately Paul Callahan's web pages are just an archive copy -- that list hasn't been updated in well over a decade, I believe. Jason Summers has the most recent version of this glider gun collection, with no missing periods from 14 to 1000.

The larger collection, which isn't exhaustive, has an old-style p1024, but nobody ever bothered to build the p2048, p4096, or p8192. Here again this was because the higher periods were thought to be "trivial", not because nobody knew how to build one. More accurately, nobody had any ideas that could improve on the obvious solution, which at the time was to add a p8 glider-to-Herschel converter to a p256 gun, and attach as many Herschel period-doublers or period-quadruplers as you needed to get the period you wanted.

Nowadays there are a lot more ways to try for a new bounding box record for high periods. For example, notice that the p1024 in Jason's collection also uses two p256 guns, but collides the glider streams to get an output Herschel and glider in a period-quadrupling reaction instead of a period-doubling one:

Code: Select all

#C p1024 gun = p256x4
x = 106, y = 52, rule = B3/S23
88b2o$88b2o5b2o$95b2o$16b2o$9b2o5b2o$9b2o53b2o8b2o17b2o$64b2o9bo17b2o$
75bobo21b2o$11b2o17b2o8b2o34b2o21b2o$11b2o17bo9b2o16b2o$5b2o21bobo27b
2o$5b2o21b2o32b2o$46b2o14b2o$46b2o$42b2o$42b2o$57b2o$57b2o41bo$98b3o$
47b2o5b2o41bo$5bo41b2o5b2o41b2o$5b3o$8bo$7b2o8$85bo18b2o$85b2o17b2o$
84bobo$2o$2o17b3o51bo$19bo51b2o26b2o$20bo10bo39b3o2b2o21b2o$31bo38bo5b
2o25b2o$5b2o26bo37bo4bo26b2o$5b2o21bo3b3o27b2o7bo3bo9b2o$b2o25b2obo3bo
26b2o9bo6b2o3bobo$b2o27b5o33b2o8bo2bo5bo9b2o$19b2o10b3o8b2o24b2o8bobo
6b2o8b2o$18bobo3b2o16b2o$7b2o9bo6b3o8b2o$7b2o8b2o7b2o8b2o28b2o$66b2o5b
2o$73b2o$38b2o20bo$31b2o5b2o21bo$31b2o26b3o!
For high enough periods it might be possible to string together semi-Snarks (i.e., "two-state machines") and Snarks on two or more different outputs of a single p256, in such a way that the outputs collide in some way similar to the above, and cause a further period-multiplication. Or, more likely, arrange a string of Snarks or semi-Snarks so that the output interferes with the original stream and produces a higher period that way.

For periods like 8192 it will be quite a trick to come up with the smallest possible solution by bounding box, though! Jason's version of the gun collection was very highly optimized, with all sorts of tricks at the edges to squeeze out a few more rows or columns from a more standard gun.
Herik Zorneck wrote:According to your definition of trivial, everything was built from something already known, using one block or fish hook, for example, would be a trivial pattern.
The definition of "trivial" is indeed very silly if you take it to that extreme. So don't take it to that extreme. Codeholic is trying to explain a subtle but real distinction between patterns that "a lot of people already know how to make", and patterns that contain "something new that nobody has seen before" -- i.e., a technological improvement, major or minor, that might be usable in other similar situations. In my reading he's not being rude about this -- unlike a few other people on the conwaylife.com forums in similar cases, by the way! -- just short and to the point.

While we're talking trivia, here's a trivial modification of Herik's p512 gun, which makes it 225 cells smaller measuring by bounding box. You can actually re-use one of the blocks from an R64 conduit as part of the semi-Snark reflector!

Code: Select all

x = 61, y = 54, rule = B3/S23
57bo$55b3o$54bo$46b2o6b2o$46bobo$31b2o14bo$31bo2bo3b2o$38b2o12b2o$27bo
24b2o5b2o$25bo7b2o3bo20b2o$7b2o8b2o12b2o4b2o$7b2o9bo5bo11b3o$18bobo3bo
3bo6b2o5b2o$19b2o7bo6bob2o3b2o$b2o22bobo8b3o$b2o54b2o$5b2o26b2o22bobo$
5b2o26bo25bo$34bo24b2o2$32bo$2o33bo$2o30bo2bo$34b3o$34b3o$14bo18bo2bo
23bo$13bo19bobo22bobo$13b3o17b3o23b2o6$7b2o$8bo$5b3o$5bo41b2o$47b2o4$
42b2o$42b2o$46b2o$46b2o$5b2o21b2o$5b2o21bobo$11b2o17bo9b2o$11b2o17b2o
8b2o3$9b2o$9b2o5b2o$16b2o!
I researched the information about gliders guns on the Jason Summer’s page :

http://www.radicaleye.com/lifepage/patt ... nprev.html

According to the author, the information in the page was last modified on Dec/2012 .
I do not know whether exists a more complete list of glider guns elsewhere on, sometimes in the Game of Life is hard to know if something exists or not, or where the information is available.
I suggested these 512p, 1024p, 2048p, 4096p and 8192p gliders guns based in the list proposal for gliders guns , according to the author :

"The goal of this collection is to include the smallest glider gun for each
period . Size is Measured by the area of the bounding rectangle of the gun .
Every cell that is a part of the rotor or stator of the gun must be
included in the bounding rectangle , excluding only the output gliders . It
Should be Noted That the population of the gun is irrelevant , and it is
Often easy to reduce the population at the cost of increase increasing the bounding
rectangle .

Also this collection serves as a way to showcase new technology Life . in
particular, if you find something que can be used to make a smaller gun
for some period than otherwise Could be built , then you know you've found
something new and useful .

In this collection , the glider gun must emit gliders in exactly one uniform
stream . The period of the gun itself is irrelevant , the only effective
period of the output gliders is considered.

If you manage to construct a smaller gun than one in this collection ,
please send me the pattern file , and I'll add it to the collection .
( Please send the full pattern , as opposed to instructions on how to make
one . ) If you construct a gun for a period that is not currently in the
collection , I'll include it if it is reasonably well-built , and optimized
for bounding -box area .”


***

“The definition of "trivial" is indeed very silly if you take it to that extreme. So don't take it to that extreme. Codeholic is trying to explain a subtle but real distinction between patterns that "a lot of people already know how to make", and patterns that contain "something new that nobody has seen before" -- i.e., a technological improvement, major or minor, that might be usable in other similar situations. In my reading he's not being rude about this -- unlike a few other people on the conwaylife.com forums in similar cases, by the way! -- just short and to the point.”

Ok. I igree about the definition. I too do not mean to be rude, just defending my point of view.
I will adopt the definition of "trivial" suggested . note that several gliders guns on the list are built with well-known parts
For example, look at the 180p glider gun in the list: basically it's composed with a Gosper 60p more a 3 ^ N multiplicator .nor for it he's off the list as "trivial"

***

“but nobody ever bothered to build the p2048, p4096, or p8192.”

but if it is a list that should contain all possible glider guns until the 9999 glider period someone should have worried... for example, someone bothered to make a glider gun 8370p, despite the fact that he is built from know components.


“While we're talking trivia, here's a trivial modification of Herik's p512 gun, which makes it 225 cells smaller measuring by bounding box. You can actually re-use one of the blocks from an R64 conduit as part of the semi-Snark reflector!”

It really is a good improvement. The idea is this. Make the glider guns for all possible periods between 1 and 9999 in the smallest possible area. But I'm just an apprentice... if all this serve to draw you attention to the need to review the design of gliders guns i think the Game of Life community's (maybe there are other learners like me?) can have the list of glider guns optimized and updated.

Thank you for your explanations

oblique
Posts: 122
Joined: July 16th, 2013, 1:30 pm

Re: Glider Guns of large periods

Post by oblique » March 27th, 2014, 2:07 am

Herik Zorneck wrote:
dvgrn wrote:
Herik Zorneck wrote:If they were so trivial, these gliders guns should be appear in the list and this is not the case.
Unfortunately Paul Callahan's web pages are just an archive copy -- that list hasn't been updated in well over a decade, I believe. Jason Summers has the most recent version of this glider gun collection, with no missing periods from 14 to 1000.
It really is a good improvement. The idea is this. Make the glider guns for all possible periods between 1 and 9999 in the smallest possible area. But I'm just an apprentice... if all this serve to draw you attention to the need to review the design of gliders guns i think the Game of Life community's (maybe there are other learners like me?) can have the list of glider guns optimized and updated.
As for "possible periods": it's easy to show that 14 *is* the shortest possible period. (Paste a glider, run to gen 13 and paste anotherone in the original position - they will collide).

Anything above this is "possible".

User avatar
codeholic
Moderator
Posts: 1147
Joined: September 13th, 2011, 8:23 am
Location: Hamburg, Germany

Re: Glider Guns of large periods

Post by codeholic » March 27th, 2014, 2:36 am

Herik Zorneck wrote:
codeholic wrote:Of course they are trivial, because they are built from known components.
According to your definition of trivial, everything was built from something already known, using one block or fish hook, for example, would be a trivial pattern
It depends on how you define a component. If you define a component as a cell pattern, then, say, the semisnark should be called trivial, that we would probably like to avoid. But cell patterns themselves don't really matter (unless they have special properties, like Gardens of Eden, but that's a different story), but it's reactions that do.

For instance, if you replaced an eater1 with another one, that does the same job, in a compound catalyst, the catalyst would stay the same, though the cell pattern would have changed, because that's a trivial change.
Ivan Fomichev

User avatar
codeholic
Moderator
Posts: 1147
Joined: September 13th, 2011, 8:23 am
Location: Hamburg, Germany

Re: Glider Guns of large periods

Post by codeholic » March 27th, 2014, 3:49 am

Well, there is yet another case. It might be that a compound pattern made of known components is so complex, that this bare fact makes it non-trivial, I mean something like the OTCA megapixel or Pi calculator - these examples are bad, because AFAIK they used some novel reactions. But even if we assumed that there had been no novel reactions used in such a pattern, we would still have to admit that the pattern is not trivial. But that should be orders of magnitude more complex than combining the Gosper's glider gun and a couple of Guam's semi-snarks.

On a side note, we proved that 31c/240 spaceships exist, and they can be build with known technology, and even most reactions we will probably use are known (probably except some spaceship edge-shooting seeds), but there is still shitload of work to be done to actually assemble the ship.
Ivan Fomichev

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Glider Guns of large periods

Post by dvgrn » March 27th, 2014, 3:59 pm

Herik Zorneck wrote:I do not know whether exists a more complete list of glider guns elsewhere on, sometimes in the Game of Life is hard to know if something exists or not, or where the information is available.
Yes, generally it's more fun to work on interesting new stuff than to keep all the endless documentation up to date. Someone really needs to do a new version of the Life Lexicon, too, for example... I don't think anyone has done any systematic work on the gun collection since 2012.
Herik Zorneck wrote:I will adopt the definition of "trivial" suggested . note that several gliders guns on the list are built with well-known parts. For example, look at the 180p glider gun in the list: basically it's composed with a Gosper 60p more a 3 ^ N multiplicator.

but if it is a list that should contain all possible glider guns until the 9999 glider period someone should have worried... for example, someone bothered to make a glider gun 8370p, despite the fact that he is built from know components.
Yup, somebody built the p8370 variant using that high-period crystal reaction, it and matched Jason's "reasonably optimized" criterion, so he threw it into the collection. Especially in the smaller, comprehensive p14-p999 collection, quite often a trivial combination of known pieces happens to have the smallest bounding box -- that's not a problem. You just document the "current best known", like the p180 gun, and then see if you can figure out how to beat it.

A decade ago no one had enough ambition to consider making the p1000-p9999 collection comprehensive. It would be a lot of work to do a good job with that, since each period is a different mathematical problem. For period 7777, do you start with a period-1111 base gun (which can include p3 oscillators) and look for a way to multiply the period by 7 -- or will the final result be smaller if you start with a couple of glider streams from a base p707 gun and see if you can collide them to get a factor-of-11 multiplication?

Now that we have the semi-Snark, it's suddenly easier to take weird intermittent guns -- such as catalyzed period-N crossed streams that absorb all the gliders except for four out of every eleven -- and turn them into clean period-11N guns. And of course the Snark is also a huge help in closing small glider-to-Herschel-to-glider loops of various lengths.

A simple Hersrch batch file could be written to come up with endless candidate guns for each period. Hersrch would take a long time to find the best Herschel loop option for 9000 different periods, though, and then most of those could be improved by period multiplying or loop-reset tricks.

The prime periods are probably the most interesting ones to tackle. If we can come up with enough timing variations for a resettable Herschel/glider loop, we can hit every possible period with a relatively small (O(log N) sized) gun. Then we can avoid Hersrch entirely, and just write a Golly script to generate period-N guns on demand. Maybe include a lookup table for known guns that are smaller than the default script-generated ones.

User avatar
simsim314
Posts: 1823
Joined: February 10th, 2014, 1:27 pm

Re: Glider Guns of large periods

Post by simsim314 » March 27th, 2014, 5:48 pm

Well I suggest to add in the first post of this thread, all the improvements we find here of Jason's p14-1000 gun collection, because after p1000 it's too much even for me...

Although with obvious design, Herik's p512 is improvement on the existing model in the collection, so I will add his gun first. Anyone with better smaller design than in the collection will also be added.

NOTE: The bounding box size is defined by the LARGEST axis it spreads on. For example a bounding box of size (10,50) will be considered of size 50, so something of size (40,40) will be considered smaller although it's total area is larger. I think this criteria is more suitable for most of life applications.

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Glider Guns of large periods

Post by dvgrn » March 27th, 2014, 6:32 pm

simsim314 wrote:NOTE: The bounding box size is defined by the LARGEST axis it spreads on. For example a bounding box of size (10,50) will be considered of size 50, so something of size (40,40) will be considered smaller although it's total area is larger. I think this criteria is more suitable for most of life applications.
As soon as you change that rule, there's suddenly a "smaller" version of at least 95% of the guns in Jason's current collection. This is because Jason's (admittedly arbitrary) rule was based on the total size of the bounding box -- i.e., the number of cells -- and an (N+1) by (N-1) bounding box is one cell smaller than an N-by-N bounding box. So basically all of the current collection is highly optimized to be short and wide as opposed to square.

You're certainly free to make this decision if you're absolutely sure you want to restart the gun collection pretty much from ground zero -- and take over the maintenance of the new collection, too: you won't be able to contribute a lot of the new guns to Jason's collection, since they'll be "bigger" according to his rules.

Here's a Golly Python script that gives the bounding box cell count for the currently selected pattern. With a few reminders, I could adapt this into a "compare-bbox.py" tool that lets you select two patterns in succession to find out which one is larger. Still a bit awkward, I admit, but it might help make the bounding-box measurement a little more palatable.

bbox-numcells.py:

Code: Select all

# bbox-numcells.py
# Displays the bounds and bounding box cell count of the selected pattern in Golly's status bar.
# The selection is adjusted to the smallest possible size before the calculation is made.
# bbox-numcells.py
import golly as g

if len(g.getselrect())==0:
  g.select(g.getrect())
if len(g.getselrect())==0:
  g.exit("Universe is empty.")
g.shrink()
r=g.getselrect()
if len(g.getcells(r))>0:
  g.show("Bounding box size = " + str(r[2]*r[3]) + ":  " +str(r[2]) + "x" + str(r[3]))
else:
  g.show("Nothing in selection.")
Jason also wrote a C++ utility at one point that compared the bounding boxes of a set of new guns against the current collection -- not sure if that code is worth resuscitating at this point, though!

User avatar
Herik Zorneck
Posts: 108
Joined: September 8th, 2013, 12:09 am

Re: Glider Guns of large periods

Post by Herik Zorneck » March 27th, 2014, 9:06 pm

simsim314 wrote:Well I suggest to add in the first post of this thread, all the improvements we find here of Jason's p14-1000 gun collection, because after p1000 it's too much even for me...

Although with obvious design, Herik's p512 is improvement on the existing model in the collection, so I will add his gun first. Anyone with better smaller design than in the collection will also be added.

NOTE: The bounding box size is defined by the LARGEST axis it spreads on. For example a bounding box of size (10,50) will be considered of size 50, so something of size (40,40) will be considered smaller although it's total area is larger. I think this criteria is more suitable for most of life applications.
I thank but to update the collection I think the 512p gun model made by dvgrn is best suited because it is smaller than the model I had drawn . by the way I'll update the others guns I have developed incorporating the design suggested by him (1024p, 2048p, 4096p, 8192p) despite knowing that they will not enter the list 14-1000p, but I think worth it for anyone who is working with guns with periods of binary multiples.
dvgrn wrote:
While we're talking trivia, here's a trivial modification of Herik's p512 gun, which makes it 225 cells smaller measuring by bounding box. You can actually re-use one of the blocks from an R64 conduit as part of the semi-Snark reflector!

Code: Select all

x = 61, y = 54, rule = B3/S23
57bo$55b3o$54bo$46b2o6b2o$46bobo$31b2o14bo$31bo2bo3b2o$38b2o12b2o$27bo
24b2o5b2o$25bo7b2o3bo20b2o$7b2o8b2o12b2o4b2o$7b2o9bo5bo11b3o$18bobo3bo
3bo6b2o5b2o$19b2o7bo6bob2o3b2o$b2o22bobo8b3o$b2o54b2o$5b2o26b2o22bobo$
5b2o26bo25bo$34bo24b2o2$32bo$2o33bo$2o30bo2bo$34b3o$34b3o$14bo18bo2bo
23bo$13bo19bobo22bobo$13b3o17b3o23b2o6$7b2o$8bo$5b3o$5bo41b2o$47b2o4$
42b2o$42b2o$46b2o$46b2o$5b2o21b2o$5b2o21bobo$11b2o17bo9b2o$11b2o17b2o
8b2o3$9b2o$9b2o5b2o$16b2o!
thanks,
Herik

User avatar
Herik Zorneck
Posts: 108
Joined: September 8th, 2013, 12:09 am

Re: Glider Guns of large periods

Post by Herik Zorneck » March 28th, 2014, 3:27 am

Although with obvious design, Herik's p512 is improvement on the existing model in the collection, so I will add his gun first. Anyone with better smaller design than in the collection will also be added.
832p glider gun :P

Code: Select all

x = 83, y = 68, rule = B3/S23
42bo$42b3o$45bo$44b2o6b2o$51bobo$52bo2$46b2o$39b2o5b2o$39b2o3$56b2o$
56b2o2$41b2o$40bobo$40bo$39b2o$52b3o$52bo$53bo$64bo8b2o$64b3o6b2o$67bo
$66b2o13b2o$81bo$79bobo$79b2o$43b2o8b2o19b2o$42bo2bo3b2o4b4o15b2o$41bo
bobo3b2o2b2ob3o$40b3obo8bo$40b3o$78bo$77bobo$78b2o2$44bo8b2o4b2o$44bo
8b2o3bobo8b2o$60bo8b2o2$12b2o$11bo2bo$12b2o$47bobo$3b2o8bo26bo6bo2b2o$
3bobo6bob2o23bobo5b3o$4bo7bo3bo23bo15bo$13bo32bo8bobo$14bo3bo25b3o9bo$
8bo5bo4bo24b2o2b3o$6b2ob2o6b3o24b2o3b2o$9b2o28b2o9bo$2b2o34bobo6bob2o$
bobo34bo8b3o$bo35b2o9bo$2o13b2o25b2o$15bo26b2o$8b2o6b3o$8b2o8bo3$54b3o
$42b2o8bob3o$38b2o4b4o3bobobo$38b2o2b2ob3o3bo2bo$42bo9b2o!
or alternatively ( i think it is even smaller box )

Code: Select all

x = 70, y = 61, rule = B3/S23
57bo$55b3o8b2o$48bo5bo11b2o$47bobo4b2o$33b3o12bo$31bob3o15b2o5b2o6b3o$
30bobobo15bobo5b2o7b2o$30bo2bo18bo11b2o$31b2o31b3o$65bobo$66b2o$52b2o$
31b2o19bobo2bo$31b2o20b2o3bo$53b6o7b2o$55b3o7bo2bo$31b3o5b2o23bobobo$
32b2o5b2o22b3obo$29b2o19bo12b3o$29b3o17bobo$30bobo17bo$31b2o7$33bo$33b
obo$33b2o2$54b2o$36b2o16b2o$35b2o$37bo4$12b2o17b2o$11bo2bo16b2o17bo$
12b2o35bobo12bo$31bo18bo12bob2o$3b2o8bo16bobo6b2o26bo$3bobo6bob2o14bo
2bo5b2o22bo2bobo$4bo7bo3bo14bo2bo30bo2bo$13bo46bo5b2o$14bo3bo12bo2bo
20bo3bobo$8bo5bo4bo11b2o21bo4bobo$6b2ob2o6b3o34bo2bo2b2o$9b2o37bo6b2o
9b2o$2b2o45b2o5bo4b2o3b2o$bobo27b2o15b2o$bo28bo2bo32bo$2o13b2o13bobo2b
o22bo6bobo$15bo15bo26b2o5bo2bo$8b2o6b3o13b2obo12bo17bo2bo$8b2o8bo15bo
12bobo4b2o$48bo5bo11bo2bo$55b3o8b2o$57bo!


obs.: I'm not sure if I should post all glider guns that I want to suggest here, because I feel that this may annoy more advanced users.
if this is not the right place please let me know.

P.S. - :idea: on second thought I think I already gave my idea of ​​how to improve the guns
I think any of you using a script or gencol (I design all my patterns by hand) can improve glider guns that "trivial" way (using semi-snarks) or maybe even better (i hope)

User avatar
simsim314
Posts: 1823
Joined: February 10th, 2014, 1:27 pm

Re: Glider Guns of large periods

Post by simsim314 » March 28th, 2014, 4:46 am

An even smaller p832

Code: Select all

x = 64, y = 53, rule = B3/S23
38bo$36b3o$35bo$27b2o6b2o$27bobo$28bo3$40b2o$32b2o6b2o$32b2o2$23b2o$
18b2o3b2o$6b2o10b3o$5bo2bo3b2o2bo2bobo16b2o$6bobo3b2o2b2o2b2o16bobo$4b
2obo8b2o22bo$4b3o33b2o$4b2o$12b2o$9b2o3bo$9b2o5bo$9b2o6bo$16b2o$12bo
37b2o$10bobo37b2o$9bo2bo$10b2o$59b2o$58bobo$3bo55bo$2bobo$3bo8b3o4bo$
14bo3bobo13b2o17b2o$13bo5bo14b2o17b2o2$60b2o$2b2o21b2o33bobo$bobo21bob
o34bo$bo24bo20b2o13b2o$2o46bo$5b2o38b3o6b2o$5b2o24b2o12bo8b2o$31b2o2$
24b2o$17b2o4bobo$7b2o7b3o4bo$7b3o5bob2o3b2o13b2o$b2o2bo2bobo3bobo20bo$
b2o2b2o2b2o3bo2bo12b2o6b3o$5b2o8b2o13b2o8bo!
It would be also nice to see someone improves on existing best guns, using new herschel technology.

User avatar
simsim314
Posts: 1823
Joined: February 10th, 2014, 1:27 pm

Re: Glider Guns of large periods

Post by simsim314 » March 28th, 2014, 5:19 am

@dvrgn - oops... I thought this is the accepted rule (for me it seems to be the most natural).

Due to the fact I didn't wanted to maintain another list of "smallest guns" by some another arbitrary criteria, lets for now accept Jason rule.

Considering the fact that Jason has one of the most linked life pages, I think it's better for now to concentrate on his criteria. It's also nice we have a script for that.

But anyone who wants to post an improvement that accepted by my criteria and not Jason's is welcome to do so, I will post his gun in the front page in different section, with no promises for further maintenance.

@Herik - If you have a huge list of gun improvements you may as well attach a zipped folder, placing each gun in separate file with name of it's period. Thus not being annoying and still passing your discoveries in a convenient way.

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Glider Guns of large periods

Post by dvgrn » March 28th, 2014, 8:20 am

simsim314 wrote:@dvrgn - oops... I thought this is the accepted rule (for me it seems to be the most natural).

Considering the fact that Jason has one of the most linked life pages, I think it's better for now to concentrate on his criteria. It's also nice we have a script for that.
Every size metric has its weaknesses, it seems. With your rule, a 101x20 gun wouldn't be an improvement over a 100x100 gun, and that does seem a little strange. On the other hand, Jason's rule doesn't take into account the fact that diamond-shaped glider guns "feel" a lot smaller than square ones (they pack closer together, have smaller populations, cover a smaller area, etc.)

Really the "most natural" rule might be something like the area of the convex hull -- but then you'd really have to run a script to compare guns most of the time, and that's just plain awkward. With Jason's rule you can usually just look at the first line of the RLE.
simsim314 wrote:An even smaller p832

x = 64, y = 53, rule = B3/S23...
Case in point -- here's a way to save two rows:

Code: Select all

#C p832 = p104 x8, with three semi-Snarks and a Snark
x = 62, y = 53, rule = B3/S23
28b2o3b2o$26b3obo2b2o$25bo4bo$25bo2b2ob4o$24b2obobobo2bo$25bobobobo9b
2o$25bobob2o10bobo$26bo16bo2b2ob2o2b2o$42b2obobobobo2bo$39b2o4bobobobo
bo$30b2o7bo5bo5bob2o$30b2o5bobo6bo3bo$37b2o8b3o$54b2o$48bo4bo2bo$12b2o
33b3o3bobobo$12b2o40bob3o$56b3o$19bo$3b2o14bobo5b2o$3bobo13b2o7bo$4bo
20b3o$25bo18b2o7bo$44b2o7bo$9b2o42b2o$9b2o18b2o$29b2o$2b2o$bobo40b2o4b
o$bo36b2o4bobo3b3o$2o13b2o20bobo4bo6bo6bo$15bo22bo18bobo$8b2o6b3o23bo
15bo$8b2o8bo22bobo$13b2o27bo$13b2o19b2o$34b2o$39b2o17b2o$4b2o33bobo16b
obo$4bobo34bo18bo$5bo20b2o13b2o17b2o$27bo27b2o$24b3o6b2o20b2o$24bo8b2o
$8b2o$8b2o$3b2o$2bobo37b3o$2bo39b3obo8b2o$b2o13b2o25bobobo3b4o4b2o$16b
o10bo16bo2bo3b3ob2o2b2o$9b2o6b3o8bo16b2o9bo$9b2o8bo6b3o!
You may think this is a little unnatural, too. You'd be perfectly right -- it is fairly ugly! I've added an extra Snark reflector, replaced one of the oscillators at the corner of the p104 with a bigger oscillator, and then had to customize one side of the oscillator to let the Snark fit one cell closer. But that's the consequence of accepting Jason's rule -- population doesn't matter, as long as you can fit the extra cells in a smaller rectangle.

I found that oscillator in the p52 gun, by the way -- there's one in the p26 gun that allows much more clearance, but I didn't need that much space, and it would have added two rows back again (!)

A few more technical points:
  • The bounding boxes of the guns in Jason's collection are never taller than they are wide.
  • The gliders always exit to the southeast (down and to the right).
  • The guns are always in the phase where an output glider will leave the bounding box on the next tick.
  • Edge sparks count. If there are no permanent ON cells at a bounding box edge, it might be good to add a single ON cell at some appropriate point to mark the true bounding box.
These aren't really requirements for getting a gun added to Jason's collection -- but the guns get oriented and adjusted before they go into the big ZIP file (makes it easier to automatically verify new submissions against the previous records). All other things being equal, everybody might as well just build guns so they match all the criteria from the start.

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Glider Guns of large periods

Post by dvgrn » March 28th, 2014, 11:55 am

Jason's gun-collection instructions wrote:If you construct a gun for a period that is not currently in the
collection, I'll include it if it is reasonably well-built , and optimized for bounding -box area .
Herik Zorneck wrote:The idea is this. Make the glider guns for all possible periods between 1 and 9999 in the smallest possible area. But I'm just an apprentice...
I have to be careful not to get stuck optimizing bounding boxes for too long, since there are a lot of other Life projects I should be working on first. But I did get pretty good at this business, about a decade ago. Maybe I can just be a glider-gun optimization consultant.

Here's another random sample. The p8192 gun can be reduced to something like this -- the height is now 61 instead of 87:

Code: Select all

#C p8192 gun = p256 x 32, with smaller bounding box
x = 100, y = 61, rule = B3/S23
57bo$55b3o18bo$54bo19b3o$46b2o6b2o17bo$46bobo16b2o6b2o$31b2o14bo17bobo
$31b2o5b2o26bo$38b2o$59b2o10b2o$51b2o6b2o10b2o5b2o$7b2o8b2o17b2o13b2o
25b2o$7b2o9bo17b2o$18bobo21b2o$19b2o21b2o17b2o$b2o58b2o$b2o54b2o$5b2o
50bobo16b2o$5b2o52bo16bobo$59b2o17bo$78b2o2$2o26b3o$2o26bo$29bo$69b2o
15b2o$69bo16b2o$67bobo$67b2o$95b2o$41b2o9b2o40bobo$41b2o9b2o41bo3$7b2o
60b2o$8bo32b2o19b2o5b2o20b2o$5b3o32b3o19b2o27b2o$5bo36bo4b2o47b2o$33b
2o4b3o5b2o8bo38bobo$31b2ob2o4b2o14bobo39bo$31bo2bo4b3o14b2o6b2o17b2o
13b2o$31bo2bo4b2ob2o20bo19bo$32b2o5b2ob2o21b3o13b3o6b2o$41bo19b2o4bo
13bo8b2o$46b2o13b2o$46b2o$5b2o21b2o$5b2o21bobo21b2o$11b2o17bo9b2o10bob
o$11b2o17b2o8b2o11bo3$9b2o$9b2o5b2o38b2o$16b2o38b2o$51b2o$50bobo$50bo$
49b2o13b2o$64bo10bo$57b2o6b3o8bo$57b2o8bo6b3o!
Possibly this could be cut down a little further by reinstating the stream crossing, with four semi-Snarks turning the output counterclockwise -- just hide those four semi-Snarks better in the "shadow" to the right of the p256 gun, without adding any extra rows. I'll leave that experiment as an exercise for the reader.

Of course you can take one semi-Snark off to get a p4096 gun, but it ends up looking a little different in standard form:

Code: Select all

#C p4096 gun = p256 x 16, with smaller bounding box
x = 90, y = 54, rule = B3/S23
16b2o$9b2o5b2o$9b2o3$11b2o17b2o8b2o$11b2o17bo9b2o$5b2o21bobo$5b2o21b2o
$46b2o$46b2o$42b2o23bo$42b2o21b3o18bo$21b3o40bo19b3o$23bo32b2o6b2o17bo
$22bo33bobo16b2o6b2o$47b2o8bo17bobo$5bo41b2o27bo$5b3o54b2o$8bo53b2o5b
2o10b2o$7b2o60b2o10b2o5b2o$88b2o2$52b2o$52b2o17b2o$71b2o$67b2o$67bobo
16b2o$69bo16bobo$69b2o17bo$88b2o$2o$2o2b2o13bob2o$4b2o5b2o6b4o$5b3o3b
2o6bo3bo$6b3o11b3o36b2o18b2o$5b5o10b3o36bo19bo$2bobo3bobo9bo30bo5bobo
17bobo9bo$bo7bo39bobo5b2o18b2o8bobo$bo3bo44b2o36b2o$2bo16b2o21b2o18b2o
$18bobo21b2o18b2o$7b2o9bo17b2o$7b2o8b2o17b2o$59b2o18b2o$52b2o5b2o11b2o
5b2o$38b2o12b2o18b2o$31b2o5b2o$31b2o14bo19bo$46bobo17bobo$46b2o6b2o10b
2o6b2o$54bo19bo$55b3o17b3o$57bo19bo!
Run these in LifeHistory to confirm that there are no obvious places where semi-Snarks can be moved a step or two closer to each other. There don't seem to be many obvious targets for "welding" two semi-Snarks together, which simplifies things (with regular Snarks there are all kinds of ways to do this, so often it's hard to be sure if you really have the optimal distances between reflectors.)

If it happens to fit better, you could also experiment with throwing in a p8 2G-to-2H-to-G in place of one of the semi-Snarks --

Code: Select all

#C p8 glider-to-Herschel-to-glider period doubler -- one of many
x = 49, y = 46, rule = LifeHistory
12.A$11.A.A$10.A3.A$A8.A3.A$3A5.A3.A$3.A3.A3.A$2.2A4.A.A33.A$9.A34.3A
$5.A41.A$4.A.A39.2A$2A2.2A28.2A$2A9.A7.A15.A$11.2A4.3A15.A.A$3.3A5.2A
3.A19.2A$5.A6.A3.2A$4.A5$36.B$36.B.B8.2D$25.2A9.3B8.2D$25.2A11.B$13.
2A$12.A2.A$7.2A4.2A$6.A.A$6.A$5.2A$15.2A$15.A29.2A$16.3A27.A$18.A11.
2A11.3A$26.2A2.2A11.A$25.A.A9.2A$22.A3.A6.2A2.2A$22.2A9.2A$21.2A.A3.
2A$20.A2.3A2.A$19.A.A.A5.3A$18.A.A.A8.A$16.3A2.A$17.A.2A$18.2A$19.A!
I've included a p8 glider reflector at the minimum distance -- it may sometimes fit in places where a Snark won't, and of course for a period-2^N pattern there's no harm in using p2 or p4 or p8 oscillators wherever they come in handy.

There are a lot of other Herschel-based period multipliers at 2^N periods, many of which have Herschel outputs and could be strung together to get higher multiples. Here's a sample x4 component:

Code: Select all

#C p8 glider-to-Herschel-to-glider period quadrupler -- one of many
x = 246, y = 167, rule = B3/S23
221bo$159bo61b3o$158bobo63bo$157bo3bo51b2o8b2o$147bo8bo3bo53bo$147b3o
5bo3bo54bobo$150bo3bo3bo44b2o10b2o$149b2o4bobo46bo$156bo34bo11bo$152bo
38b3o9b2o20b2o$151bobo40bo30b2o$147b2o2b2o28b2o10b2o$147b2o9bo7bo15bo
61b2o$158b2o4b3o15bobo59b2o$150b3o5b2o3bo19b2o$152bo6bo3b2o41b2o$151bo
54b2o7$172b2o$172b2o23b2o$160b2o35bo$159bo2bo35b3o$154b2o4b2o38bo33b2o
$153bobo54b2o22b2o$153bo55bobo$152b2o55bo$162b2o44b2o$162bo$163b3o$
165bo11b2o$173b2o2b2o$172bobo9b2o$169bo3bo6b2o2b2o$169b2o9b2o$168b2obo
3b2o$167bo2b3o2bo$166bobobo5b3o$165bobobo8bo$163b3o2bo$164bob2o$165b2o
$166bo18$100b3o$102bo$101bo48$50b3o$52bo$51bo48$3o$2bo$bo!
Obviously, even without the extra reflector, this is still quite a bit bigger than two semi-Snarks strung together, so you'd generally only use this G-to-H conversion when you need an output Herschel instead of a glider -- or maybe sometimes when you need the output to go in a different direction -- or when there's some reason you need multiple glider outputs instead of one.

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Glider Guns of large periods

Post by dvgrn » March 28th, 2014, 4:42 pm

Decided to split the last post into two -- now they're both just much-too-long instead of ridiculous. The questions at the bottom of this message are intended to introduce a beginner to the wearisome and strangely fascinating art of Conway's Life glider-gun construction.

You can "bootstrap" a H loop by feeding an extra glider back in to the G-to-H converter. Below is a sample period 1408+16N adjustable loop along these lines. p1408 already has a smaller entry in Jason's list -- which can be made much smaller with semi-Snarks, by the way -- but that gun is not +/-8N adjustable. As usual, I'd suggest running patterns like these in Golly's built-in LifeHistory rule to get a sense of where the active cells are.

I'm not claiming that this is the smallest way to build a p1408+16N loop -- this is again left as an exercise for the reader. As an introduction to Herschel/glider loops, though, it's pretty good sample:

Code: Select all

#C G-to-H-to-G loop, p704+8N x2
x = 118, y = 74, rule = B3/S23
31b2o2$29bo3bo$29bo4bo$31bobobo8bo$32bobobo5b3o$33bo4bo2bo$34bo3bo2b2o
2$35b2o2bo$38bobo$39b2o2b2o60bo$31bo11b2o60b3o$29b3o76bo$28bo27bo9bo
40b2o$28b2o26b3o5b3o28b2o$18b2o39bo3bo13b2o17bo$19bo38b2o3b2o13bo17bob
o$19bobo55bo19b2o$20b2o4b2o49b2o$25bo2bo$26b2o$38b2o$38b2o2$80b2o$80b
2o26b2o$108b2o4$24b2o3b2o$26bo2bo19b2o$23bo6b3o15bobo11b2o6bo$24b2o6bo
15bo13b2o6b3o2b2o5b2o$4bo42b2o7bo16bo2bo5b2o$2b5o14b2o32bobo12b3o3bobo
27b2o$bo5bo13bo32bo3bo11bo6b2o28bo$bo2b3o12bobo13b2o18bo3bo8bobo10b2o
21b3o10bo$2obo15b2o14b2obo17bo3bo5b3ob2o8bobo21bo10bobo$o2b4o32bo8bo8b
o3bo3bo15bo11b2o5b2o14b2o$b2o3bo3b2o24bo9b3o9bobo4b2o10bo11bo3b2o5bo$
3b3o4b2o25bob2o4bo13bo16b3o8bobo8bobo$3bo35b2o4b2o16bo11bob3o8b2o8b2o$
2obo58bobo9bo3bo15b2o$2ob2o38bo19b2o2b2o4bo3bo16bobo$42bobo22b2o3b3obo
18bo$43b2o2b2o24b3o23bo$11b2o34b2o25bo23bobo$12bo84bo3bo$9b3o86bo3bo$
9bo89bo3bo$100bo3bo$101bobo$102bo7$56b2o$37b2o13b2o2b2o21b2o$33b2o2b2o
12bobo21b2o2b2o$32bobo13bo3bo21bobo$33bo14b2o25bo$29bo17b2obo3b2o15bo$
35b2o9bo2b3o2bo15b3o4b2o$27b3obo3bo9bobobo5b3o11bob3o3bo$26bob2o6b3o5b
obobo8bo10bo3bo5b3o$26b2obo8bo3b3o2bo19bo3bo8bo$24bob3o14bob2o19b3obo$
44b2o21b3o$26bo18bo22bo!
Apprentice glider-gun builders might want to consider these practice questions:
  • Part 1 -- glider colors and phases
  • Why can't the Snark be replaced by a p8 reflector, or a p8 reflector be replaced by a Snark?
  • Why can't a pair of p8 reflectors be replaced by a pair of Snarks?
    (The paths can be made to line up, but the construction still fails. Why?)
  • Why can't the seven p8 reflectors be replaced with one p8 reflector?
    (Same answer as for the previous question, really. Never mind about the bounding box getting bigger.)

    Part 2 -- loop timing and oscillator rephasing
  • What are two different pairs of reflectors you can move to get a p1424 gun?
    (Period 1424 is not in the current collection.)
  • How far do you have to move the two reflectors to get p1424? Why?
  • What do you have to do to the two reflectors after you move them, to make the pattern work again?
  • Why wouldn't you have to change the reflectors if you move them twice as far, to get a p1440 gun?

    Part 3 -- multi-signal loops, alternate H-to-Gs, overclocking:
  • Where can you add one glider to a p1440 gun to get a p720 gun?
  • Where can you add two gliders to a p1440 gun to get a p480 gun?
  • Why can't you add three gliders to get a p360 gun?
  • Why can't you use the multi-signal trick to get p704 from p1408, or p712 from p1424?
    (Same answer as for the add-three-gliders question.)
  • What new problem appears when you try adding four gliders to get a p288 gun?
  • How can you fix that and get an 'overclocked' p288 gun after all?
    (Hint: reposition a key p8 reflector, and add two more reflectors -- Snarks will do.)
  • What else could you do to get a p704 or p712 with this basic design?
    (Hint: if you don't want the period doubled, don't use a period doubler.)

    Bonus questions:
  • Can you make the p720 x2 base loop work with more than five signals in it?
  • How about a p728 x2 base loop with more than six signals?
See also this stable adjustable loop gun if you'd rather work with a loop that doesn't have any headache-inducing oscillators. With just Snarks as reflectors, you can't adjust glider phases, and you're also more or less out of luck if your input glider is the wrong color. Known p4/5/6/7/8 reflectors change the phase of a reflected glider by one tick. So generally you'd have to add more Herschel conduits and/or H-to-G converters to get the right phase and color combination of output glider.

Another possible pure-stable adjustable template gun pattern is the Snark-assisted p43 gun. The base gun shown there is p86+8N, but with enough adjustments you can get pretty much any period you want. P648 seems to be a good starting point:

Code: Select all

#C p648 Snark-assisted bootstrapped Herschel base loop
x = 129, y = 126, rule = B3/S23
44b2o3b2o$44b2o2bob3o$48bo4bo$44b4ob2o2bo$44bo2bobobob2o$47bobobobo$
48b2obobo$52bo2$38b2o$39bo7b2o$39bobo5b2o$40b2o7$50b2o$50bo$40b2o9b3o$
41b2o10bo$40bo2$27bo$25b5o14b2o$24bo5bo13bo$24bo2b3o12bobo$23b2obo15b
2o$23bo2b4o$24b2o3bo3b2o$26b3o4b2o$26bo$23b2obo$23b2ob2o$88b2obo$88bob
2o$34b2o$35bo20bo24b2o$32b3o3b2o14b5o22b2o$32bo6bo13bo5bo$39bobo12b3o
2bo$40b2o15bob2o$54b4o2bo$49b2o3bo3b2o$49b2o4b3o$57bo13b2o$57bob2o11bo
$56b2ob2o11bobo$73b2o2$48b2o$48bo$49b3o$51bo17bobo$69b2obo$10b2o60b3o$
10b2o42b2o13b2o4bo$55bo13bob5o$55bobo13bo$56b2o12bo3b2o$69bo3bobo$69b
2o3bo$48bo$46b3o44b2o$45bo47b2o$32b2o11b2o$32b2o2$88b2o$88b2o$2o90b2o$
2o90b2o3$57b2o27b2o$21b2o34b2o11b2o14b2o36bo$21bobo46bo35b2o14b5o$23bo
47b3o33bo13bo5bo$23b2o48bo33bobo12b3o2bo$108b2o15bob2o$113b3o6b4o2bo$
31b2o82b4o3bo3b2o$30bobo81bob3o4b3o$30bo94bo$29b2o94bob2o$124b2ob2o3$
116b2o$116bo$117b3o$119bo9$86bo$84b3o$83bo$83b2o7$73b2o$72bobo5b2o$72b
o7b2o$71b2o2$85bo$81b2obobo$80bobobobo$77bo2bobobob2o$77b4ob2o2bo$81bo
4bo$46bo30b2o2bob3o$47bo29b2o3b2o$45b3o!
Doesn't look as if this will set any records, though things start getting almost competitive with eight signals in the loop (p81). It would be a clear winner if only we were measuring guns by convex hull or by population -- or if edge shooters got lots of bonus points!

Notice that the current gun collection's p648 gun manages to fold a similar-sized adustable loop into a much smaller space, by using both p8 and p6 reflectors.

User avatar
Herik Zorneck
Posts: 108
Joined: September 8th, 2013, 12:09 am

Re: Glider Guns of large periods

Post by Herik Zorneck » March 28th, 2014, 10:17 pm

wow! I think the guns are getting very good
I think I could not make a better contribution as my only idea is to improve the guns using the semi snarks
at least my insistence served to irritate you to the point of making a list update, this is a good thing (I think) :wink:
I have to be careful not to get stuck optimizing bounding boxes for too long, since there are a lot of other Life projects I should be working on first. But I did get pretty good at this business, about a decade ago. Maybe I can just be a glider-gun optimization consultant.

Here's another random sample. The p8192 gun can be reduced to something like this -- the height is now 61 instead of 87:

CODE: SELECT ALL
#C p8192 gun = p256 x 32, with smaller bounding box
x = 100, y = 61, rule = B3/S23
57bo$55b3o18bo$54bo19b3o$46b2o6b2o17bo$46bobo16b2o6b2o$31b2o14bo17bobo
$31b2o5b2o26bo$38b2o$59b2o10b2o$51b2o6b2o10b2o5b2o$7b2o8b2o17b2o13b2o
25b2o$7b2o9bo17b2o$18bobo21b2o$19b2o21b2o17b2o$b2o58b2o$b2o54b2o$5b2o
50bobo16b2o$5b2o52bo16bobo$59b2o17bo$78b2o2$2o26b3o$2o26bo$29bo$69b2o
15b2o$69bo16b2o$67bobo$67b2o$95b2o$41b2o9b2o40bobo$41b2o9b2o41bo3$7b2o
60b2o$8bo32b2o19b2o5b2o20b2o$5b3o32b3o19b2o27b2o$5bo36bo4b2o47b2o$33b
2o4b3o5b2o8bo38bobo$31b2ob2o4b2o14bobo39bo$31bo2bo4b3o14b2o6b2o17b2o
13b2o$31bo2bo4b2ob2o20bo19bo$32b2o5b2ob2o21b3o13b3o6b2o$41bo19b2o4bo
13bo8b2o$46b2o13b2o$46b2o$5b2o21b2o$5b2o21bobo21b2o$11b2o17bo9b2o10bob
o$11b2o17b2o8b2o11bo3$9b2o$9b2o5b2o38b2o$16b2o38b2o$51b2o$50bobo$50bo$
49b2o13b2o$64bo10bo$57b2o6b3o8bo$57b2o8bo6b3o!

Possibly this could be cut down a little further by reinstating the stream crossing, with four semi-Snarks turning the output counterclockwise -- just hide those four semi-Snarks better in the "shadow" to the right of the p256 gun, without adding any extra rows. I'll leave that experiment as an exercise for the reader.
I tried to improve their solution to making a 8192p glider gun with a smaller box (I am not sure that I have achieved) but take a look in this pattern :

Code: Select all

x = 110, y = 62, rule = B3/S23
92b2o$92b2o5b2o$99b2o2$19bo$19b3o46b2o8b2o17b2o$22bo45b2o9bo17b2o$21b
2o6b2o48bobo6bo14b2o$28bobo49b2o5bob2o12b2o$29bo20bo11b2o23bo3bo$48b3o
11b2o23b2o2bo$23b2o22bo18b2o19b2ob2o$16b2o5b2o14b2o6b2o17b2o16bo3b2o$
16b2o21bobo40b2o$40bo42b2o2$33b2o9b3o14b2o$33b2o17b2o7b2o41bo$45b2o5b
2o48b3o$18b2o26bo54bo$17bobo16bo64b2o$17bo17bobo2b2o$16b2o17bobo2b2o$
12b2o22bo3bo2b2o$12b2o27b3o6b2o$41b3o6bobo$43b2o7bo$3b2o39b2o6b2o$3bob
o39bo$4bo2$108b2o$108b2o$7b2o$7b2o$2b2o$bobo99b2o$bo101b2o$2o13b2o90b
2o$15bo20b2o69b2o$8b2o6b3o17bo29b2o21b2o$8b2o8bo15bobo29b2o21bobo$34b
2o36b2o17bo9b2o$41b2o29b2o17b2o8b2o$19b2o21bo$19b2o21bobo$43b2o25b2o$
70b2o5b2o$36b2o20b2o17b2o$29b2o5b2o20b2o$29b2o$49b2o$24bo16b2o6b2o$23b
obo15b2o$23b2o6b2o$31bo$32b3o19bo$34bo18bobo$46b2o6b2o$47bo$44b3o$44bo
!
this was done by connecting the glider gun with a 5X semi snark system pattern
unfortunately this 5X semi snark system does not work well in any input period less than 180p
here a others examples , a 5760p glider gun

Code: Select all

x = 110, y = 58, rule = B3/S23
19bo$19b3o$22bo62bo$21b2o6b2o54bobo$28bobo57b2o6b2o$29bo20bo23b2o12b2o
4bo3bo$48b3o23b2o12b2o3bo5bo$23b2o22bo21b3o13bobo4b2obo3bo8b2o$16b2o5b
2o14b2o6b2o20b3o13bo7bo5bo8b2o$16b2o21bobo28bo23bo3bo$40bo29bo25b2o$
70bo$33b2o9b3o22bobo12b2o17b2o$33b2o17b2o30b2o17b2o$45b2o5b2o$18b2o26b
o22bobo31bo$17bobo16bo33bo31bobo$17bo17bobo2b2o28bo19b2o10bobo$16b2o
17bobo2b2o28bo18b2o12bo$12b2o22bo3bo2b2o24b3o5bobo11bo$12b2o27b3o6b2o
17b3o5b2o$41b3o6bobo25bo21b2obob2o$43b2o7bo47bo5bo$3b2o39b2o6b2o18b2o
27bo3bo$3bobo39bo25bobo24bo3b3o$4bo67bo24b2o$85b2o10bobo$84bo3bo$83bo
5bo$7b2o74bo3bob2o2b2o$7b2o74bo5bo3b2o$2b2o80bo3bo$bobo81b2o14bo$bo99b
o$2o13b2o83bobo$15bo20b2o61b2ob2o$8b2o6b3o17bo61bo5bo$8b2o8bo15bobo33b
2o3b2o24bo$34b2o36b3o23b2o3b2o$41b2o28bo3bo15b2o$19b2o21bo6bo22bobo15b
obo$19b2o21bobo3b3o22bo6b2o7b3o4b2o3bo$43b3ob5o28b2o6b3o4bo2b3obo$45b
3o3b2o36b3o4b2ob2obo$36b2o7b3o3b3o4b2o30bobo8bo$29b2o5b2o8b2o3b2o5b2o
13b2o16b2o$29b2o16b5o21b2o$48b3o$24bo16b2o6bo$23bobo15b2o$23b2o6b2o$
31bo$32b3o19bo$34bo18bobo$46b2o6b2o$47bo$44b3o$44bo!
and a p7680 glider gun :

Code: Select all

x = 95, y = 66, rule = B3/S23
88b2o$88b2o7$19bo69bo$19b3o54b3o9b3o$22bo54bo9b5o$21b2o6b2o46bo8bobobo
bo$28bobo45b3o7b2o3b2o$29bo20bo$48b3o25b3o$23b2o22bo28b3o10bo$16b2o5b
2o14b2o6b2o39bobo$16b2o21bobo34b3o9bob2o$40bo36bo11b3o$77bo11bo2bo$33b
2o9b3o29b3o11b3o$33b2o17b2o35bo3bo$45b2o5b2o34bo5bo$18b2o26bo36bo5bo3b
o$17bobo16bo39bo5bo7b3o$17bo17bobo2b2o32b2o6b3o$16b2o17bobo2b2o33b2o$
12b2o22bo3bo2b2o$12b2o27b3o6b2o$41b3o6bobo$43b2o7bo$3b2o39b2o6b2o$3bob
o39bo$4bo$90b2o$90b2o$82bobo$7b2o71bo3bo$7b2o71bo$2b2o75bo4bo$bobo76bo
$bo78bo3bo6b2o$2o13b2o56bo8bobo6bobo$15bo20b2o34b3o18bo$8b2o6b3o17bo
35b3o18b2o$8b2o8bo15bobo$34b2o34b2o3b2o$41b2o27b2o3b2o$19b2o21bo39bo$
19b2o21bobo37b2o$43b2o28bo3b2o4b2o8b2o$72bobo2b2o4b3o7b2o$36b2o20b2o
12bobo2b2o4b2o$29b2o5b2o20b2o13bo8b2o$29b2o51bo$49b2o$24bo16b2o6b2o$
23bobo15b2o$23b2o6b2o$31bo$32b3o19bo$34bo18bobo$46b2o6b2o$47bo$44b3o$
44bo!
thanks,
Herik
Last edited by Herik Zorneck on March 29th, 2014, 1:38 am, edited 1 time in total.

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Glider Guns of large periods

Post by dvgrn » March 28th, 2014, 10:41 pm

Herik Zorneck wrote:I tried to improve their solution to making a 8192p glider gun with a smaller box (I'm not sure) but take a look in this pattern :

x = 110, y = 62, rule = B3/S23...
Just check the first line of the RLE pattern code. Your bounding box is 110x62. You quoted my p8192 gun pattern just before that:
dvgrn wrote: #C p8192 gun = p256 x 32, with smaller bounding box
x = 100, y = 61, rule = B3/S23
Your revised gun is larger in both the X and Y dimensions, so unfortunately it's not an improvement.

As it happens, an extension of the p4096 gun I posted turns out to cut several more rows off of the p8192 bounding box:

Code: Select all

#C p8192 gun = p256 x 32, with shorter wider bounding box
x = 101, y = 54, rule = B3/S23
57bo19bo$55b3o17b3o$54bo19bo22bo$46b2o6b2o10b2o6b2o19b3o$46bobo17bobo
25bo$31b2o14bo19bo18b2o6b2o$31b2o5b2o46bobo$38b2o32b2o13bo$59b2o11b2o
5b2o$51b2o6b2o18b2o11b2o$7b2o8b2o17b2o13b2o39b2o5b2o$7b2o9bo17b2o61b2o
$18bobo21b2o18b2o$19b2o21b2o18b2o$b2o79b2o$b2o54b2o18b2o3b2o$5b2o50bob
o17bobo$5b2o52bo19bo17b2o$59b2o18b2o16bobo$99bo$99b2o$2o$2o2$69b2o$69b
o20b2o$35bo31bobo20bo$12bo21b2o31b2o19bobo9bo$11bo24bo51b2o8bobo$11b3o
19b3o16b2o45b2o$34bo17b2o19b2o$73b2o2$7b2o60b2o$8bo53b2o5b2o19b2o$5b3o
54b2o19b2o5b2o$5bo41b2o34b2o$47b2o8bo$56bobo19bo$56b2o6b2o11bobo$64bo
12b2o6b2o$42b2o21b3o17bo$42b2o23bo18b3o$46b2o40bo$46b2o$5b2o21b2o$5b2o
21bobo$11b2o17bo9b2o$11b2o17b2o8b2o3$9b2o$9b2o5b2o$16b2o!
It's one cell wider than the old version, so this would not be an improvement under simsim314's first proposed size metric. But 101x54=5454, meaning a lot fewer cells in the bounding box than 100x61=6100 -- according to the rules of Jason Summers' gun collection, this would be the current winner.

Anyone see a configuration that beats 101x54? After a while it starts getting a lot harder to shrink the bounding box any further...!

For the record, here's a version of the p832 gun that beats the previous x = 62, y = 53:

Code: Select all

#C p832 = p104 x8 (3 semi-Snarks)
x = 55, y = 48, rule = B3/S23
22b2o$20bo2bo8bo$20b3o7b3o18bo$7b2o2b2ob2o2b2o9bo19b3o$7bo2bobobobo2bo
b2o6b2o17bo$8bobo5bobo2bobo16b2o6b2o$7b2obob3obob2o2bo17bobo$11bo3bo
25bo2$6b2o3bo3bo18b2o10b2o$5bo2bo2bo3bo10b2o6b2o10b2o5b2o$6bobo17b2o
25b2o$4b2obo4b3o$4b3o10b2o$4b2o11b2o17b2o$36b2o$32b2o$32bobo16b2o$8bo
7b2o16bo16bobo$7bobo6b2o16b2o17bo$7bobo10b3o30b2o$8bo13bo$21bo$9bo$8b
4o$7bo2b2o32b2o$3bo3bo4bo31bo$2bobo2b2obob2o28bobo9bo$3bo7b4o4bo22b2o
8bobo$11b4o3bobo32b2o$10b5o4bo7b2o$9b2o2bo13b2o$10b3o$2b2o7bo$bobo40b
2o$bo35b2o5b2o$2o35b2o$5b2o$5b2o25bo$31bobo$31b2o6b2o$39bo$17b2o21b3o$
7b2o7b3o23bo$7b3o5bob2o$b2o2bo2bobo3bobo$b2o2b2o2b2o3bo2bo$5b2o8b2o!

User avatar
Herik Zorneck
Posts: 108
Joined: September 8th, 2013, 12:09 am

Re: Glider Guns of large periods

Post by Herik Zorneck » March 29th, 2014, 1:53 am

dvgrn wrote:

Code: Select all

#C p8192 gun = p256 x 32, with shorter wider bounding box
x = 101, y = 54, rule = B3/S23
57bo19bo$55b3o17b3o$54bo19bo22bo$46b2o6b2o10b2o6b2o19b3o$46bobo17bobo
25bo$31b2o14bo19bo18b2o6b2o$31b2o5b2o46bobo$38b2o32b2o13bo$59b2o11b2o
5b2o$51b2o6b2o18b2o11b2o$7b2o8b2o17b2o13b2o39b2o5b2o$7b2o9bo17b2o61b2o
$18bobo21b2o18b2o$19b2o21b2o18b2o$b2o79b2o$b2o54b2o18b2o3b2o$5b2o50bob
o17bobo$5b2o52bo19bo17b2o$59b2o18b2o16bobo$99bo$99b2o$2o$2o2$69b2o$69b
o20b2o$35bo31bobo20bo$12bo21b2o31b2o19bobo9bo$11bo24bo51b2o8bobo$11b3o
19b3o16b2o45b2o$34bo17b2o19b2o$73b2o2$7b2o60b2o$8bo53b2o5b2o19b2o$5b3o
54b2o19b2o5b2o$5bo41b2o34b2o$47b2o8bo$56bobo19bo$56b2o6b2o11bobo$64bo
12b2o6b2o$42b2o21b3o17bo$42b2o23bo18b3o$46b2o40bo$46b2o$5b2o21b2o$5b2o
21bobo$11b2o17bo9b2o$11b2o17b2o8b2o3$9b2o$9b2o5b2o$16b2o!
It's one cell wider than the old version, so this would not be an improvement under simsim314's first proposed size metric. But 101x54=5454, meaning a lot fewer cells in the bounding box than 100x61=6100 -- according to the rules of Jason Summers' gun collection, this would be the current winner.

Anyone see a configuration that beats 101x54? After a while it starts getting a lot harder to shrink the bounding box any further...!
I think it is better done in this way because this system can easily be used in a modular form
for example a glider gun p1920 using a p60 glider gun with input and your 5X semi snark combo

Code: Select all

x = 77, y = 59, rule = B3/S23
33bo19bo$31b3o17b3o$30bo19bo22bo$22b2o6b2o10b2o6b2o19b3o$22bobo17bobo
25bo$23bo19bo18b2o6b2o$27b4o31bobo$27b3obo16b2o13bo$28b2o5b2o11b2o5b2o
$35b2o18b2o10b3o$75b2o$68b2o5b2o$18b2o18b2o29bo$18b2o18b2o19bo$58bobo
2b2o$33b2o18b2o3bobo2b2o$24b2o7bobo17bobo3bo3bo2b2o$25b2o8bo19bo8b3o6b
2o$24bo10b2o18b2o7b3o6bobo$66b2o7bo$2b2o63b2o6b2o$2b2o64bo$26bo$24b3o
5bobo$23bo9b2o10b2o$4b2o17b2o8bo11bo20b2o$43bobo20bo$21bo21b2o19bobo$
20b3o41b2o$19bo3bo4b2o$2o3b2o14bo6b2o19b2o$b5o3b2o7bo5bo24b2o$b2ob2o4b
2o6bo5bo12b2o$b2ob2o3bo9bo3bo13b2o6b2o$2b3o15b3o22b2o19b2o$59b2o5b2o$
16b2o41b2o$15b2o16bo$17bo14bobo19bo$4b3o25b2o6b2o11bobo$4b3o33bo12b2o
6b2o$3bo3bo33b3o17bo$2bo5bo34bo18b3o$3bo3bo56bo$4b3o12bo$18b3o$18b3o2$
16b2o3b2o$16b2o3b2o3$19bo$18bobo$4b2o14b2o$4b2o14b2o$19b3o$18bobo$18b
2o!

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Glider Guns of large periods

Post by dvgrn » March 29th, 2014, 8:58 am

Herik Zorneck wrote:
dvgrn wrote:Anyone see a configuration that beats 101x54? After a while it starts getting a lot harder to shrink the bounding box any further...!
I think it is better done in this way because this system can easily be used in a modular form
for example a glider gun p1920 using a p60 glider gun with input and your 5X semi snark combo

x = 77, y = 59, rule = B3/S23...
In this case the bounding box can be reduced to under 4000 cells -- 91x43=3913 is an improvement on 77x59=4543. Anyone who is paying attention might well start worrying again about using Jason's bounding-box size metric to define the "smallest" gun -- or might even reconsider getting mixed up in this strange and arbitrary pastime at all. The optimizations often start to seem rather silly. Here I can improve the overall size by increasing the footprint of the base p60 gun...!

Here's the gun in canonical form, with the output glider advanced to the edge of the new bounding box. A careful look through the known p60N guns would very likely turn up some tricks that could cut another column or two off of this design:

Code: Select all

#C p1920 = p60 x32 with 5 semi-Snarks
x = 91, y = 43, rule = B3/S23
48bo19bo$46b3o17b3o18bo$45bo19bo19b3o$37b2o6b2o10b2o6b2o17bo$24bo12bob
o17bobo16b2o6b2o$24b4o10bo19bo17bobo$8bo16b4o48bo$7bobo5b2o8bo2bo5b2o
27b2o$2o3b2o3bo14b4o5b2o14b2o11b2o5b2o10b2o$2o3b2o3bo4bobob2o3b4o14bo
7b2o18b2o10b2o5b2o$5b2o3bo5b2o3bo2bo17b2o2bo42b2o$7bobo10bo$8bo8bo2bo
12b2o18b2o$33b2o18b2o17b2o$72b2o$48b2o18b2o$27bo20bobo17bobo16b2o$28bo
21bo19bo16bobo$26b3o21b2o18b2o17bo$89b2o5$31b3o26b2o18b2o$33bo26bo19bo
$32bo25bobo17bobo9bo$53b2o3b2o18b2o8bobo$52bo2bo33b2o$15bobo25b2o6b2ob
2o7b2o$13bo3bo5b3o17b2o7bobo8b2o$13bo7bobo2bo2b2o22bo$6b2o4bo4bo7b2o2b
o2bo$6b2o5bo19bo26b2o18b2o$13bo3bo2b3o10bo6b2o18b2o11b2o5b2o$15bobo15b
o6b2o31b2o$29bo2bo$29b2o17bo19bo$47bobo17bobo$47b2o6b2o10b2o6b2o$55bo
19bo$56b3o17b3o$58bo19bo!
While optimizing this one I noticed that I'd been careless about the original p8192 -- you can save another column by tightening up the semi-Snarks. At least this means that the new version is now a contender by either the longest-diameter or bounding-box method:

Code: Select all

p8192 gun = p256 x 32 with 5 semi-Snarks
x = 100, y = 54, rule = B3/S23
57bo19bo$55b3o17b3o18bo$54bo19bo19b3o$46b2o6b2o10b2o6b2o17bo$46bobo17b
obo16b2o6b2o$31b2o14bo19bo17bobo$31b2o5b2o46bo$38b2o32b2o$59b2o11b2o5b
2o10b2o$33bo17b2o6b2o18b2o10b2o5b2o$7b2o8b2o14bo2b2o13b2o45b2o$7b2o9bo
17b2o$18bobo21b2o18b2o$19b2o21b2o18b2o17b2o$b2o78b2o$b2o31bo22b2o18b2o
$5b2o26b2o22bobo17bobo16b2o$5b2o26b2o24bo19bo16bobo$59b2o18b2o17bo$98b
2o2$2o$2o$32b2obo$32b5o32b2o18b2o$33b4o32bo19bo$13bo53bobo17bobo9bo$
12bo20b3o31b2o18b2o8bobo$12b3o18b3o62b2o$34bo17b2o18b2o$52b2o18b2o3$7b
2o60b2o18b2o$8bo53b2o5b2o11b2o5b2o$5b3o54b2o18b2o$5bo41b2o$47b2o8bo19b
o$56bobo17bobo$56b2o6b2o10b2o6b2o$64bo19bo$42b2o21b3o17b3o$42b2o23bo
19bo$46b2o$46b2o$5b2o21b2o$5b2o21bobo$11b2o17bo9b2o$11b2o17b2o8b2o3$9b
2o$9b2o5b2o$16b2o!
But yes, it's definitely easy to extend this design to get higher powers of two. Each semi-Snark is placed in the same position relative to the previous one, except that the very last semi-Snark can be moved one step closer.

User avatar
Kazyan
Posts: 1247
Joined: February 6th, 2014, 11:02 pm

Re: Glider Guns of large periods

Post by Kazyan » March 29th, 2014, 1:22 pm

dvgrn wrote:#C p832 = p104 x8 (3 semi-Snarks)
x = 55, y = 48, rule = B3/S23
Trivial improvements shave off a row and a column.

Code: Select all

x = 54, y = 47, rule = B3/S23
31bo$7b2o7b2o11b3o18bo$7bo2b2ob2o2bo10bo19b3o$8b2obobob2o3b2o6b2o17bo$
9bo5bo4bobo16b2o6b2o$9bob3obo5bo17bobo$10bo3bo25bo2$5b2o3bo3bo18b2o10b
2o$4bo2bo2bo3bo10b2o6b2o10b2o5b2o$5bobo17b2o25b2o$3b2obo4b3o$3b3o10b2o
$3b2o11b2o17b2o$35b2o$31b2o$31bobo16b2o$7bo7b2o16bo16bobo$6bobo6b2o16b
2o17bo$6bobo10b3o30b2o$7bo13bo$20bo$8bo$7b4o$6bo2b2o32b2o$2bo3bo4bo31b
o$bobo2b2obob2o28bobo9bo$2bo7b4o4bo22b2o8bobo$10b4o3bobo32b2o$9b5o4bo
7b2o$8b2o2bo13b2o$9b3o$2b2o6bo$3bo39b2o$3o33b2o5b2o$o35b2o$4b2o$4b2o
25bo$30bobo$30b2o6b2o$38bo$16b2o21b3o$6b2o7b3o23bo$6b3o5bob2o$2o2bo2bo
bo3bobo$2o2b2o2b2o3bo2bo$4b2o8b2o!
Tanner Jacobi
Coldlander, a novel, available in paperback and as an ebook. Now on Amazon.

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Glider Guns of large periods

Post by dvgrn » March 29th, 2014, 6:40 pm

simsim314 wrote:Here is a general concept of how an o(log(n)) glider gun can be built for any period. The main idea is that the "vibrating" blocks that actually represent the binary code of "current state", can be manipulated after the glider is "shoot" out to fit any needed period.
I've been thinking off and on about this design, and last week I collected the pieces needed to do this same kind of thing with stable circuitry. It doesn't take much: Silver reflectors to split glider streams and also merge them, a Snark as a color-preserving reflector (a Silver reflector would work as well, but we might as well use the smallest fastest thing we have available) -- and semi-Snarks for the two-state switches as usual.

Stringing the pieces together produces a construction that's roughly the same size as the p60 version:

EDIT: fixed initial states of some of the semi-Snarks

Code: Select all

#C stable version of binary multiplier, base period around 6900
x = 521, y = 752, rule = B3/S23
96b2o$96b2o20b2o$118bo$116bobo10bo$116b2o11b3o$132bo14bo$131b2o12b3o$
144bo$144b2o3$143b2o$124b2o17b2o$124b2o2$86b2o$21b4o4b3o4b3o4b3o4b3o
33b2o$21bo3bo2bo3bo2bo3bo2bo3bo2bo3bo$21bo3bo2bo6bo3bo2bo6bo2b2o6b2o$
13b2o6bo3bo2b4o4b4o2b4o3bobobo6b2o65b2o$13b2o6b4o3bo3bo6bo2bo3bo2b2o2b
o53b2o19bo$21bo6bo3bo2bo3bo2bo3bo2bo3bo53bobo15b3o$21bo7b3o4b3o4b3o4b
3o56bo15bo$109b2o19b2o$131bo$128b3o$128bo6$121b2o$122bo$34bobo85bobo
15b2o$29bo4bo2bo85b2o15b2o$18b2o10b2o5b2o113b2o$18b2o5b2o8bo3b2o5bo
105bobo$25b2o10b2o6bobo106bo$34bo2bo6bo3bo105b2o$19bo14bobo8b3o$18b3o
7b2o13b2o3b2o$17bo3bo4bo2bo$19bo$16bo5bo2bo$16bo5bo$17bo3bo4b2o20bo$
18b3o7bo19bo92b2o$44b3obo92bobo$143bo$25b2o3b2o11bo99b2o$25b2o3b2o11b
2obo$21bo4b5o14b2o$27bobo2$21bo5b3o15b2o3b2o$19bobo23b2o3b2o$20b2o24b
5o$39b2o6bobo$39b2o90b2o$47b3o81b2o$14b2o3b2o$28bo$15bo3bo9bo$16b3o8b
3o$16b3o$47b2o93b2o$47b2o72b2o19bo$39b3o15b2o63bo17bobo$39bo17b2o63bob
o15b2o$17b2o17bo3bo82b2o4bo$17b2o15bobo91bobo$35b2o91bobo$12b2o115bo
10b2o$12b2o126bobo$47b2o8b3o40bobobo37bo$47bobo6b2ob2o37bo7bo35b2o$3b
2o42bo8b2ob2o66b2o21bo$3bobo50b5o36bo4b2o3bo20bo19b3o$4bo50b2o3b2o39bo
bo21b3o19bo19b2o$97bo3bo5bo17bo21b2o18bo$100b2o63bobo10bo$55bo41bo9bo
57b2o11b3o$7b2o20b2o22b2obo124bo14bo$7b2o20b2o25bo41bo7bo73b2o12b3o$2b
2o96bobobo88bo$bobo49bob2o136b2o$bo36b2o15bo2bo$2o13b2o20bobo16bobo$
15bo22bo153b2o$8b2o6b3o154b2o17b2o$8b2o8bo154b2o$13b2o$13b2o19b2o19b3o
77b2o$34b2o18bo3bo76b2o$39b2o12bo5bo$4b2o33bobo12bo3bo$4bobo34bo13b3o
118b2o$5bo20b2o13b2o12b3o13bo84b2o19bo$27bo43b3o82bobo15b3o$24b3o6b2o
39bo83bo15bo$10b2o12bo8b2o38b2o83b2o19b2o$10b2o43b2o123bo$55b2o120b3o$
3b2o172bo$2bobo$2bo101b2o$b2o13b2o86bo$16bo85bobo$9b2o6b3o40bo37b2o2b
2o$9b2o8bo14bo9bo15b3o35b2o70b2o$34b3o5b3o18bo107bo$37bo3bo20b2o11b2o
94bobo15b2o$36b2o3b2o32b2o95b2o15b2o$201b2o$201bobo$203bo$203b2o4$50b
2o52b2o$50b2o34b2o16bobo$38b2o45bobo18bo$37bo2bo44bo20b2o$32b2o4b2o44b
2o4b2o98b2o$31bobo55bobo98bobo$31bo57bo102bo$30b2o56b2o7b2o93b2o$40b2o
33b2o20b2o$40bo34bo$41b3o32b3o$43bo34bo2$109bo$107b3o$106bo$106b2o72b
2o$180b2o$31b2o$31b2o3$22b2o63bo$22bobo62b3o101b2o$23bo66bo79b2o19bo$
89b2o80bo17bobo$171bobo15b2o$172b2o4bo$26b2o149bobo$26b2o149bobo$21b2o
155bo10b2o$20bobo166bobo$20bo128bobobo37bo$19b2o13b2o111bo7bo35b2o$34b
o141b2o21bo$27b2o6b3o71b2o35bo4b2o3bo20bo19b3o$27b2o8bo71b2o39bobo21b
3o19bo19b2o$146bo3bo5bo17bo21b2o18bo$149b2o63bobo10bo$146bo9bo57b2o11b
3o$230bo14bo$147bo7bo73b2o12b3o$149bobobo88bo$242b2o2$94b2o$94b2o145b
2o$222b2o17b2o$222b2o2$184b2o$184b2o3$225b2o$120bo84b2o19bo$120b3o82bo
bo15b3o$123bo83bo15bo$122b2o83b2o19b2o$229bo$226b3o$226bo2$153b2o$153b
o$151bobo$109bo37b2o2b2o$83bo9bo15b3o35b2o70b2o$83b3o5b3o18bo107bo$86b
o3bo20b2o11b2o94bobo15b2o$85b2o3b2o32b2o95b2o15b2o$250b2o$250bobo$252b
o$252b2o4$99b2o52b2o$99b2o34b2o16bobo$87b2o45bobo18bo$86bo2bo44bo20b2o
$81b2o4b2o44b2o4b2o98b2o$80bobo55bobo98bobo$80bo57bo102bo$79b2o56b2o7b
2o93b2o$89b2o33b2o20b2o$89bo34bo$90b3o32b3o$92bo34bo2$158bo$156b3o$
155bo$155b2o72b2o$229b2o$80b2o$80b2o3$71b2o63bo$71bobo62b3o101b2o$72bo
66bo79b2o19bo$138b2o80bo17bobo$220bobo15b2o$221b2o4bo$75b2o149bobo$75b
2o149bobo$70b2o155bo10b2o$69bobo166bobo$69bo128bobobo37bo$68b2o13b2o
111bo7bo35b2o$83bo141b2o21bo$76b2o6b3o71b2o35bo4b2o3bo20bo19b3o$76b2o
8bo71b2o39bobo21b3o19bo19b2o$195bo3bo5bo17bo21b2o18bo$198b2o63bobo10bo
$195bo9bo57b2o11b3o$279bo14bo$196bo7bo73b2o12b3o$198bobobo88bo$291b2o
2$143b2o$143b2o145b2o$271b2o17b2o$271b2o2$233b2o$233b2o3$274b2o$169bo
84b2o19bo$169b3o82bobo15b3o$172bo83bo15bo$171b2o83b2o19b2o$278bo$275b
3o$275bo2$202b2o$202bo$200bobo$158bo37b2o2b2o$132bo9bo15b3o35b2o70b2o$
132b3o5b3o18bo107bo$135bo3bo20b2o11b2o94bobo15b2o$134b2o3b2o32b2o95b2o
15b2o$299b2o$299bobo$301bo$301b2o4$148b2o52b2o$148b2o34b2o16bobo$136b
2o45bobo18bo$135bo2bo44bo20b2o$130b2o4b2o44b2o4b2o98b2o$129bobo55bobo
98bobo$129bo57bo102bo$128b2o56b2o7b2o93b2o$138b2o33b2o20b2o$138bo34bo$
139b3o32b3o$141bo34bo2$207bo$205b3o$204bo$204b2o72b2o$278b2o$129b2o$
129b2o3$120b2o63bo$120bobo62b3o101b2o$121bo66bo79b2o19bo$187b2o80bo17b
obo$269bobo15b2o$270b2o4bo$124b2o149bobo$124b2o149bobo$119b2o155bo10b
2o$118bobo166bobo$118bo128bobobo37bo$117b2o13b2o111bo7bo35b2o$132bo
141b2o21bo$125b2o6b3o71b2o35bo4b2o3bo20bo19b3o$125b2o8bo71b2o39bobo21b
3o19bo19b2o$244bo3bo5bo17bo21b2o18bo$247b2o63bobo10bo$244bo9bo57b2o11b
3o$328bo14bo$245bo7bo73b2o12b3o$247bobobo88bo$340b2o2$192b2o$192b2o
145b2o$320b2o17b2o$320b2o2$282b2o$282b2o3$323b2o$218bo84b2o19bo$218b3o
82bobo15b3o$221bo83bo15bo$220b2o83b2o19b2o$327bo$324b3o$324bo2$251b2o$
251bo$249bobo$207bo37b2o2b2o$181bo9bo15b3o35b2o70b2o$181b3o5b3o18bo
107bo$184bo3bo20b2o11b2o94bobo15b2o$183b2o3b2o32b2o95b2o15b2o$348b2o$
348bobo$350bo$350b2o4$197b2o52b2o$197b2o34b2o16bobo$185b2o45bobo18bo$
184bo2bo44bo20b2o$179b2o4b2o44b2o4b2o98b2o$178bobo55bobo98bobo$178bo
57bo102bo$177b2o56b2o7b2o93b2o$187b2o33b2o20b2o$187bo34bo$188b3o32b3o$
190bo34bo2$256bo$254b3o$253bo$253b2o72b2o$327b2o$178b2o$178b2o3$169b2o
63bo$169bobo62b3o101b2o$170bo66bo79b2o19bo$236b2o80bo17bobo$318bobo15b
2o$319b2o4bo$173b2o149bobo$173b2o149bobo$168b2o155bo10b2o$167bobo166bo
bo$167bo128bobobo37bo$166b2o13b2o111bo7bo35b2o$181bo141b2o21bo$174b2o
6b3o71b2o35bo4b2o3bo20bo19b3o$174b2o8bo71b2o39bobo21b3o19bo19b2o$293bo
3bo5bo17bo21b2o18bo$296b2o63bobo10bo$293bo9bo57b2o11b3o$377bo14bo$294b
o7bo73b2o12b3o$296bobobo88bo$389b2o2$241b2o$241b2o145b2o$369b2o17b2o$
369b2o2$331b2o$331b2o3$372b2o$267bo84b2o19bo$267b3o82bobo15b3o$270bo
83bo15bo$269b2o83b2o19b2o$376bo$373b3o$373bo2$300b2o$300bo$298bobo$
256bo37b2o2b2o$230bo9bo15b3o35b2o70b2o$230b3o5b3o18bo107bo$233bo3bo20b
2o11b2o94bobo15b2o$232b2o3b2o32b2o95b2o15b2o$397b2o$397bobo$399bo$399b
2o4$246b2o52b2o$246b2o34b2o16bobo$234b2o45bobo18bo$233bo2bo44bo20b2o$
228b2o4b2o44b2o4b2o98b2o$227bobo55bobo98bobo$227bo57bo102bo$226b2o56b
2o7b2o93b2o$236b2o33b2o20b2o$236bo34bo$237b3o32b3o$239bo34bo2$305bo$
303b3o$302bo$302b2o72b2o$376b2o$227b2o$227b2o3$218b2o63bo$218bobo62b3o
101b2o$219bo66bo79b2o19bo$285b2o80bo17bobo$367bobo15b2o$368b2o4bo$222b
2o149bobo$222b2o149bobo$217b2o155bo10b2o$216bobo166bobo$216bo128bobobo
37bo$215b2o13b2o111bo7bo35b2o$230bo141b2o21bo$223b2o6b3o71b2o35bo4b2o
3bo20bo19b3o$223b2o8bo71b2o39bobo21b3o19bo19b2o$342bo3bo5bo17bo21b2o
18bo$345b2o63bobo10bo$342bo9bo57b2o11b3o$426bo14bo$343bo7bo73b2o12b3o$
345bobobo88bo$438b2o2$290b2o$290b2o145b2o$418b2o17b2o$418b2o2$380b2o$
380b2o3$421b2o$316bo84b2o19bo$316b3o82bobo15b3o$319bo83bo15bo$318b2o
83b2o19b2o$425bo$422b3o$422bo2$349b2o$349bo$347bobo$305bo37b2o2b2o$
279bo9bo15b3o35b2o70b2o$279b3o5b3o18bo107bo$282bo3bo20b2o11b2o94bobo
15b2o$281b2o3b2o32b2o95b2o15b2o$446b2o$446bobo$448bo$448b2o4$295b2o52b
2o$295b2o34b2o16bobo$283b2o45bobo18bo$282bo2bo44bo20b2o$277b2o4b2o44b
2o4b2o98b2o$276bobo55bobo98bobo$276bo57bo102bo$275b2o56b2o7b2o93b2o$
285b2o33b2o20b2o$285bo34bo$286b3o32b3o$288bo34bo2$354bo$352b3o$351bo$
351b2o72b2o$425b2o$276b2o$276b2o3$267b2o63bo$267bobo62b3o101b2o$268bo
66bo79b2o19bo$334b2o80bo17bobo$416bobo15b2o$417b2o4bo$271b2o149bobo$
271b2o149bobo$266b2o155bo10b2o$265bobo166bobo$265bo128bobobo37bo$264b
2o13b2o111bo7bo35b2o$279bo141b2o21bo$272b2o6b3o71b2o35bo4b2o3bo20bo19b
3o$272b2o8bo71b2o39bobo21b3o19bo19b2o$391bo3bo5bo17bo21b2o18bo$394b2o
63bobo10bo$391bo9bo57b2o11b3o$475bo14bo$392bo7bo73b2o12b3o$394bobobo
88bo$487b2o2$339b2o$339b2o145b2o$467b2o17b2o$467b2o2$429b2o$429b2o3$
470b2o$365bo84b2o19bo$365b3o82bobo15b3o$368bo83bo15bo$367b2o83b2o19b2o
$474bo$471b3o$471bo2$398b2o$398bo$396bobo$354bo37b2o2b2o$328bo9bo15b3o
35b2o70b2o$328b3o5b3o18bo107bo$331bo3bo20b2o11b2o94bobo15b2o$330b2o3b
2o32b2o95b2o15b2o$495b2o$495bobo$497bo$497b2o4$344b2o52b2o$344b2o34b2o
16bobo$332b2o45bobo18bo$331bo2bo44bo20b2o$326b2o4b2o44b2o4b2o98b2o$
325bobo55bobo98bobo$325bo57bo102bo$324b2o56b2o7b2o93b2o$334b2o33b2o20b
2o$334bo34bo$335b3o32b3o$337bo34bo2$403bo$401b3o$400bo$400b2o72b2o$
474b2o$325b2o$325b2o3$316b2o63bo$316bobo62b3o101b2o$317bo66bo79b2o19bo
$383b2o80bo17bobo$465bobo15b2o$466b2o4bo$320b2o149bobo$320b2o149bobo$
315b2o155bo10b2o$314bobo166bobo$314bo128bobobo37bo$313b2o13b2o111bo7bo
35b2o$328bo141b2o$321b2o6b3o71b2o35bo4b2o3bo20bo32b2o$321b2o8bo71b2o
39bobo21b3o33b2o$440bo3bo5bo17bo$443b2o27bo$440bo9bo19b3o$454bo14bo$
441bo7bo4b3o12b2o$443bobobo9bo$456b2o2$388b2o$388b2o67b2o$457b2o17b2o
16bo$476b2o14b2ob2o$495b5o$496bo2bo14b2o$496bo2bo14b2o$497b2o2$473b2o$
414bo58bo19b2o$414b3o57b3o15bobo$417bo58bo15bo$416b2o52b2o19b2o$470bo$
471b3o$473bo2$447b2o$447bo$445bobo6b2o$403bo37b2o2b2o6bobo$377bo9bo15b
3o35b2o10bo25b2o$377b3o5b3o18bo45b2o25bo$380bo3bo20b2o11b2o40b2o15bobo
$379b2o3b2o32b2o40b2o15b2o8$393b2o52b2o$393b2o34b2o16bobo$381b2o45bobo
18bo$380bo2bo44bo20b2o$375b2o4b2o44b2o4b2o24b2o$374bobo55bobo23bobo$
374bo57bo25bo$373b2o56b2o7b2o15b2o$383b2o33b2o20b2o$383bo34bo$384b3o
32b3o$386bo34bo2$452bo$450b3o$449bo$449b2o18b2o$469b2o$374b2o$374b2o3$
365b2o63bo$365bobo62b3o25b2o$366bo66bo25bo19b2o$432b2o25bobo17bo$460b
2o15bobo$464b2o6bo4b2o$369b2o92bo2bo4bobo$369b2o93b2o5bobo$364b2o94b2o
10bo$363bobo93bobo$363bo95bo$362b2o13b2o79b2o$377bo95b2o$370b2o6b3o71b
2o19bo$370b2o8bo71b2o20b3o$476bo8$437b2o$437b2o5$515bo$513bobo$514b2o$
517b2o$517bobo$519bo$519b2o5$427bo$425b3o$424bo$424b2o7$414b2o$413bobo
5b2o$413bo7b2o$412b2o2$426bo$422b2obobo$421bobobobo$418bo2bobobob2o$
418b4ob2o2bo$422bo4bo$418b2o2bob3o$418b2o3b2o!
You can replace the p6960 gun with any higher-period gun, and then multiply its period by any factor from 1 to 256. At some point I can probably rearrange the pieces to get the minimum period down around 4000, but this will do as a proof of concept.

To program the adjustable circuit to multiply any gun's period by N, simply translate (N-1) into an eight-bit binary value -- e.g., "00010010" for 18, to produce an x19 multiplier. Starting from the southeast corner, retain the circled eaters for every "1" and remove them for every "0".

To get higher multipliers, of course, just extend the chain to add more bits. And really of course if you're multiplying by a factor of less than 129, you can shorten the chain appropriately, and then you'll be able to use slightly lower-period base guns.

Now I have to do some serious thinking. It must be possible to make this device much faster and more compact, probably by using Herschels to travel up and down the chain instead of gliders... but Herschel splitters and Herschel merges tend to be a bit of a headache, so it will take some work to find the best configuration. Snarks should come in very handy to reset Herschel receivers where necessary.

And maybe it's worth figuring out how to pack the chain into a nice spiral, while I'm at it, to keep the limit bounding box a bit lower...!

User avatar
simsim314
Posts: 1823
Joined: February 10th, 2014, 1:27 pm

Re: Glider Guns of large periods

Post by simsim314 » March 30th, 2014, 4:08 am

I was thinking about a new design to the O(log(n)) glider gun.

First of all, all current designs are multipliers, this making an arbitrary prime period an issue. I previously thought to make p60 multiplier and then add 60 such guns to unite their periods - but it's an ugly solution.

On the other hand what is nice about "circular" guns that they use addition, instead of multiplication. I was thinking how to combine the two.

So my solution is: use circular gun of size O(log(n)), to adjust a binary state of some binary (it's size is also less then log(n)). Then after shooting out a glider from this configuration, kill the circulating glider inside the gun, adjust the binary state AND use another glider to delay the period to the needed prime number, the delaying mechanism will also be of O(log(n)) size, approximation of the gun period, then re-fuse the gun again with another back shoot. Thus being both circular, and O(log(n)).

It's a bit tricky mechanism, which uses three back shoots - one to adjust the binary state, one to kill the circulation in glider gun, and one to re-fuse the same circulation, thus allowing adding a delay mechanism to adjust to any needed prime period.

It will take me time to construct it...but this is the idea.

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Glider Guns of large periods

Post by dvgrn » March 30th, 2014, 9:14 am

simsim314 wrote:So my solution is: use circular gun of size O(log(n)), to adjust a binary state of some binary (it's size is also less then log(n)). Then after shooting out a glider from this configuration, kill the circulating glider inside the gun, adjust the binary state AND use another glider to delay the period to the needed prime number, the delaying mechanism will also be of O(log(n)) size, approximation of the gun period, then re-fuse the gun again with another back shoot. Thus being both circular, and O(log(n)).
Yup, we're thinking along the same lines -- that was the next item on my list, too.

Stoppable/restartable glider loops aren't terribly difficult. Here's an old sample (from Golly's Life/Guns/Oscillators) with a base period of 1536:

Code: Select all

#C p103079214841 (prime: 4^13*1536-263) oscillator/gun --
#C  p1536 base loop, 13 quadruplers, and a 263-step glider advancer.
#C  Based on previous p97307852711 and p97307852687 oscillators;
#C  original design by Gabriel Nivasch, 7 August 2003.
#C The first three quadruplers are set (an L112 quadrupler, a Fx70
#C  quadrupler plus a required F166 dependent conduit to suppress the
#C  following Herschel's first glider, and another L112 after that),
#C  so the circuit counts down from 64. Use an 8^2 or 10^2 step size.
#C When all circuits are empty, a signal gets through to activate
#C the 263-step glider advancer -- after which the circuit counts down
#C again, this time from 4^13.
#C Dave Greene, 31 Mar 2006; checked and corrected by Tomas Rokicki.
x = 297, y = 227, rule = B3/S23
193b2o72b2o$193bo73bo$194bo73bo$182bo10b2o61bo10b2o$182b3o71b3o$185bo
73bo$172bo11b2o60bo11b2o$170b3o71b3o$169bo40bo32bo40bo$169b2o19b2o16b
3o32b2o37b3o$155b2o33b2o15bo21b2o50bo$156bo50b2o21bo50b2o$156bobo71bob
o$157b2o72b2o2$176b2o72b2o$176b2o72b2o6$173b2o36b2o34b2o36b2o$77b2o94b
o19b2o16bobo33bo19b2o16bobo$77b2o5b2o88b3o15bobo18bo34b3o15bobo18bo$
84b2o90bo15bo20b2o35bo15bo20b2o$48b2o97bob2o19b2o19b2o4b2o22bob2o19b2o
19b2o4b2o$49bo97b2obo19bo25bobo22b2obo19bo25bobo$36bo11bo14b2o17b2o87b
3o22bo48b3o22bo$36b3o9b2o14bo17b2o89bo21b2o7b2o41bo21b2o7b2o$39bo24bob
o21b2o114b2o72b2o$26b2o10b2o25b2o21b2o$27bo$27bobo$13b2o13b2o$13b2o36b
2o$51b2o$195bo18b2o53bo18b2o$7b2o140b2o43bobo17bo8b2o43bobo17bo$7b2o
139bobo44bo16bobo7bobo44bo16bobo$11b2o135bo63b2o8bo63b2o$11b2o134b2o
44b5o23b2o44b5o$192bo4bo68bo4bo$61b2o99b2o27bo2bo41b2o27bo2bo$42b2o18b
o28b2o69b2o24bo2bob2o41b2o24bo2bob2o$6b2o34bo16b3o29bo95bobobo5bo63bob
obo5bo$6b2o35b3o13bo29bobo77bob2o15bo2bo4bobo44bob2o15bo2bo4bobo$45bo
43b2o78b2obo18b2o2bo2bo44b2obo18b2o2bo2bo$26bo3b2o120b2o42b2o28b2o42b
2o$25bobo3bo120b2o72b2o$25b2o3bo111b2o72b2o$29bo113bo73bo$25b5obo111bo
bo71bobo$25bo4b2o112b2o72b2o$26b3o$9bo18bob2o$7b3o19bobo$6bo$6b2o186b
2o16b2o54b2o16b2o$9b2o77b2o104b2o16b2o54b2o16b2o$4b5obo61b2o14b2o$4bo
21b2o45bo22b2o$6bob2o16bobo41b3o23bo51b2o72b2o$5b2ob2o18bo41bo23bobo
50bobo39b2o30bobo39b2o$28b2o64b2o51bo42bo30bo42bo$20b2o124b2o42bobo27b
2o42bobo$20b2o169b2o72b2o$200b2o72b2o$4b2o70bo123b2o72b2o$4b2o70b3o$
79bo$78b2o83b2o72b2o$163b2o72b2o$154b2o72b2o$153bobo27b2o42bobo27b2o$
153bo30bo42bo30bo$152b2o30bobo39b2o30bobo$96b2o87b2o72b2o$21b2o73bo$
21bobo70bobo$23bo70b2o61b2o16b2o54b2o16b2o$23b2o132b2o16b2o54b2o16b2o
2$77b2o$77b2o15b2o$94bobo$5b2o89bo84b2o72b2o$4bobo23bo65b2o82bobo71bob
o$4bo23b3o149bo73bo$3b2o22bo151b2o72b2o$11b2o14b2o160b2o72b2o$11b2o64b
2o80b2o28b2o42b2o28b2o$77bo76b2o2bo2bo7bo36b2obo18b2o2bo2bo44b2obo$75b
obo73bo2bo4bobo6bobo35bob2o15bo2bo4bobo44bob2o$75b2o73bobobo5bo6b2ob2o
52bobobo5bo$151bo2bob2o9bo2bo28b2o24bo2bob2o41b2o$154bo2bo8bo3bo28b2o
27bo2bo41b2o$155bo4bo9bo58bo4bo$92b2obo60b5o23b2o44b5o$92bob2o79b2o8bo
63b2o$158bo16bobo7bobo44bo16bobo$85b2o70bobo17bo8b2o43bobo17bo$85b2o
71bo18b2o53bo18b2o$10b2o13b2o236b2o18bo$9bobo13b2o237bo17bobo$9bo254bo
bo16bo$8b2o255b2o$27bo253b5o$26bobo252bo4bo$27bo47b2o91bo72b2o41bo2bo$
28b3o45bo81b2o7b2o41bo21b2o7b2o41b2obo2bo$30bo45bobo80bo48b3o22bo47bo
5bobobo$77b2o80bobo22b2obo19bo25bobo44bobo4bo2bo$154b2o4b2o22bob2o19b
2o19b2o4b2o44bo2bo2b2o$136b2o17bo20b2o35bo15bo20b2o29b2o$20b2o115bo17b
obo18bo34b3o15bobo18bo$20b2o74bo27bo11bo19b2o16bobo33bo19b2o16bobo$94b
3o27b3o9b2o36b2o34b2o36b2o$93bo33bo$11b2o80b2o19b2o10b2o$11b2o84b2o16b
o$46b2o50bo16bobo$46b2o48bo19b2o$96b2o41b2o72b2o$139b2o72b2o50b2o16b2o
$12b2o152b2o97b2o16b2o$13bo152b2o26b2o$10b3o3b2o175bobo$10bo4bobo152b
2o21bo50b2o43b2o$15bo83b2o69bo21b2o50bo44bo$14b2o64b2o17b2o70b3o32b2o
37b3o39bobo$13bo66b2o91bo32bo40bo39b2o$13b3o25b2o87b2o75b3o$16bo24bo
88bo16b2o60bo11b2o14b2o38b2o$15b2o25b3o34b2o50b3o14bo73bo15bo38b2o$44b
o35bo52bo11b3o71b3o15bo$77b3o12b2o51bo10b2o61bo10b2o5b2o10bo$16b2o19b
2o38bo14bo64bo73bo15b3o$16b2o19b2o54b3o60bo73bo15bo$89b2o4bo60b2o72b2o
14b2o13b2o$89bobo170bo$91b3o127bo39bo31bo$90bo3bob2o123b3o37b2o28b3o$
90b2o2b2obo126bo65bo$223b2o65b2o4$254b2o$29b2o223b2o$29bo$30b3o260b2o$
32bo260bobo$295bo$295b2o$162b2o55b2o36b2o$8b2o3b2o103b2o42b2o54bobo16b
2o19bo$9bo2bobo103bo99bo18bobo15b3o25b2o3b2o$8bo3bo106bo97b2o20bo15bo
6bo11b2o7b2o3b2o$8b5o105b2o48b2o63b2o4b2o19b3o11bo$168b2o63bobo23bo15b
3o$6b5o153b2o69bo23b2o16bo$5bo4bo153b2o60b2o7b2o$4bo2bo218b2o$bo2bob2o
135b2o$obobo5bo63b2o39b2o26b2o$bo2bo4bobo9b2o51b2o39b2o52b2o$4b2o2bo2b
o9b2o146b2o$9b2o19b2o37b2o$28bo2bo37b2o$28b2o186b2o$33b2o182bo$33bo
111b2o70bobo$31bobo71b2o38bobo70b2o$30bobo72bo41bo$26b2o3bo20bo53b3o
11b2o25b2o$26b2o23bobo54bo11b2o$50bo2bo$8b2o41b2o183b2o$8b2o56bo67b2o
100bo$62b2o2bobo65bo99bobo$62b2o2b2o67b3o96b2o$137bo11b2o$124b2o22bobo
23bo$125bo22bo23b3o$122b3o22b2o22bo$122bo32b2o14b2o$155b2o$75b2o$9b2o
64b2o$8bobo60b2o$8bo62bobo$7b2o64bo$73b2o158b2o$60b2o155b2o14b2o$60b2o
156bo22b2o$215b3o23bo$12b2o3b2obo194bo23bobo$13bo3b2ob3o35b2o109b2o2b
2ob2o61b2o$10b3o10bo34b2o94b2o13bobo2bobo$10bo6b2ob3o130bobo16b2o3bo7b
o$18bobo132bo19bob2o6b3o$18bobo131b2o16bo2bobo6bo38bo$19bo149bobobobo
6b2o37b3o$170b2ob2o49bo$202bo20b2o$200b3o16b2o$199bo19bo$186b2o11b2o
20bo$186b2o32b2o6$155b2o60b2o$155b2o60b2o17b2o$161b2o73b2o$161b2o$180b
2o$176b2o2bo2b2o52b2o$159b2o16bo3bob2o52bo$159b2o5b2o9bobobo42b2o12b3o
$166b2o10b2obob2o22b2o16bo14bo$181bo2bo4b2o16bo14b3o$181b2o6b2o17b3o
11bo$210bo!
That pattern combines the glider insertion with the glider suppression, which is probably not the most efficient design these days. It might make more sense to lay out a loop using mostly Snarks, with a section of Herschel track to split off a signal for the binary counter and trigger an edge-shooter (or transparent H-to-G, anyway) to complete the 2^N loop.
simsim314 wrote:It's a bit tricky mechanism, which uses three back shoots - one to adjust the binary state, one to kill the circulation in glider gun, and one to re-fuse the same circulation, thus allowing adding a delay mechanism to adjust to any needed prime period.
Here's the design that I thought of, which should allow a Golly script to build a reasonably efficient gun for any input number. The biggest problem is that most reasonable-sized glider suppression/glider insertion mechanisms will fail to work for some of the possible phases of the base loop.

However, if the base loop has two suppression points on opposite edges, at least one of them is guaranteed to work for any given phase of the loop. The gun-construction script can route the signal to the appropriate suppression point for the specific period being constructed, and also split off a restart signal that enters the loop through the transparent H-to-G, one cycle later:

Code: Select all

#C Sample p1218 base loop with two "glider stoppers".
#C A signal in the base loop (Herschel) is suppressed by a glider entering
#C one of the two glider-stopper components.  A new glider can enter
#C the loop at any later time and restart the gun.  The output at the left
#C can be fed into a binary multiplier unit.
x = 147, y = 189, rule = B3/S23
42bo$43bo$41b3o28$78b2o$77bo2bo$78b2o8$82b3o$83bo$81b3o13$62bo$60b3o$
36bo22bo$24bo11b3o20b2o20b2o$22b3o14bo32b2o7b2o$6bo14bo16b2o33bo$6b3o
12b2o50bobo$9bo58b2o4b2o$8b2o59bo20b2o$69bobo18bo24b2o$70b2o16bobo24b
2o$9b2o77b2o$9b2o17b2o$28b2o$118bo9b2o$117bo10bo$117b3o6bobo$126b2o2$
25b2o32b2o$25bo20b2o11b2o$26b3o18bo60b2o$28bo15b3o35b2o8bo15b2o$22b2o
20bo37b2o2b2o4b3o$22bo63bobo6bo30b2o$23b3o62bo5b2o30bobo$25bo62b2o38bo
8bo$128b2o5b3o$134bo$134b2o$6b2o$5bobo109b2o$5bo25b2o84bobo22b2ob2o$4b
2o25bo87bo23bob2o$12b2o15bobo87b2o22bo$12b2o15b2o104b2o4b3o$135b2o3bo
3b2o$140b4o2bo$126b2o15bob2o$97b2o26bobo12b3o2bo$96bobo26bo13bo5bo$96b
o27b2o14b5o$95b2o45bo5$11b2o$10bobo$10bo$9b2o9$21b2o$21b2o$33b2o$33bob
o$35bo$35b2o2$10b2o$11bo19b2o$11bobo17bo$12b2o15bobo$24bo4b2o$23bobo$
23bobo$12b2o10bo$11bobo$11bo$10b2o$25b2o$4bo20bo$2b5o14b2o3b3o$bo5bo
13bo6bo$bo2b3o12bobo31b2o$2obo15b2o32bo$o2b4o44bobo$b2o3bo3b2o39b2o$3b
3o4b2o$3bo$2obo$2ob2o$29b2o$30bo$11b2o17bobo$12bo18b2o$9b3o$9bo2$20b2o
$21bo$21bobo30b2o$22b2o30bo$55b3o$40b2o15bo$40b2o3$33bo$22b2o$21bobo6b
3o$21bo10bo$20b2o9bo2$50bo$50b3o$33b2o18bo$33b2o17b2o7$62b2o$55b2o5bob
o$55b2o7bo$64b2o2$51bo$50bobob2o$50bobobobo$49b2obobobo2bo$50bo2b2ob4o
$50bo4bo$51b3obo2b2o$53b2o3b2o!
The biggest remaining detail is adjusting the timing of the binary multiplier's output, modulo the base loop's period, to achieve all possible gun periods. This can be done inside a fixed bounding box, using a 180-degree reflector plus a set of eight different Herschel circuits, with the exact same (X, Y) offsets but timing in the range (T, T+1, ... T+7). I think I have a set of these lying around somewhere, and if not, Hersrch can come up with something pretty quick.

EDIT: Sorry, my mental list of useful converters seems to be a bit out of date. Here's a base loop that can go as low as p512 (shown), though p1024 (or thereabouts) is more likely unless someone can dig up a more compact "glider stopper" component -- the receiver-based one in the previous pattern doesn't fit at p512. The loop glider can fairly easily be shot down directly by a 90-degree glider, of course, but that takes some extra synchronization circuitry -- seems better to use a stable switch if possible.

Code: Select all

#C Sample p512 base loop (really p392+8N) and Herschel factory.
#C Glider insertion via H-to-G is shown.
x = 124, y = 112, rule = B3/S23
92b2o$91bo2bo$92b2o26$95b2o$86b2o7b2o17bo$87bo$40b2o3b2o40bobo26bo$40b
2o2bob3o39b2o$44bo4bo54b2o12bo$40b4ob2o2bo54bo$40bo2bobobob2o51bobo15b
2o$43bobobobo52b2o16bobo$44b2obobo72bo$48bo38bo34b2o$87b3o$34b2o42bo
11bo$35bo7b2o7bo9bo15b3o8b2o14bo$35bobo5b2o7b3o5b3o18bo22bobo$36b2o17b
o3bo20b2o23bo$54b2o3b2o$110bo$108b3o$107bo$107b2o2$46b2o$46bo68b2ob2o$
47b3o18b2o46bob2o$49bo18b2o23b2o21bo$56b2o35bo14b2o4b3o$55bo2bo35b3o
11b2o3bo3b2o$50b2o4b2o38bo16b4o2bo$49bobo47b2o15bob2o$49bo48bobo12b3o
2bo$48b2o48bo13bo5bo$58b2o37b2o14b5o$58bo56bo$59b3o$61bo5$85bo$85b3o$
9bo78bo$7b5o14b2o59b2o$6bo5bo13bo$6bo2b3o12bobo$5b2obo15b2o$5bo2b4o$6b
2o3bo3b2o$8b3o4b2o$8bo88b2o$5b2obo81b2o5bobo$5b2ob2o80b2o7bo$99b2o2$
16b2o17bo50bo$17bo17b3o47bobob2o$14b3o21bo46bobobobo$14bo22b2o45b2obob
obo2bo$85bo2b2ob4o$9b2o74bo4bo$10bo75b3obo2b2o$10bobo75b2o3b2o$11b2o2$
47b2o$40b2o5bobo$32b2o6b2o7bo$32b2o15b2o2$11bo24bo$11bobo21bobob2o$11b
3o21bobobobo$13bo20b2obobobo2bo$35bo2b2ob4o$35bo4bo$36b3obo2b2o$38b2o
3b2o2$2b2o$bobo$bo$2o20b2o$22b2o!

User avatar
dvgrn
Moderator
Posts: 10612
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Glider Guns of large periods

Post by dvgrn » March 30th, 2014, 10:02 am

dvgrn wrote:

Code: Select all

#C Sample p512 base loop (really p392+8N) and Herschel factory.
For anyone looking to build small adjustable-period base guns, that 386-recovery H-to-G converter also allows for a p934+8N loop --

Code: Select all

#C p934+8N base glider loop and Herschel factory
x = 128, y = 119, rule = B3/S23
70b2o$69bo2bo$70b2o26$73b2o$64b2o7b2o$65bo$18b2o3b2o40bobo$18b2o2bob3o
39b2o$22bo4bo54b2o$18b4ob2o2bo54bo$18bo2bobobob2o51bobo$21bobobobo52b
2o$22b2obobo$26bo38bo$65b3o$12b2o42bo11bo49bo$13bo7b2o7bo9bo15b3o8b2o
14bo32b3o$13bobo5b2o7b3o5b3o18bo22bobo30bo$14b2o17bo3bo20b2o23bo31b2o$
32b2o3b2o$88bo$86b3o34b2ob2o$85bo38bob2o$85b2o37bo$33bo82b2o4b3o$24b2o
8bo81b2o3bo3b2o$24bo7b3o58b2ob2o23b4o2bo$25b3o18b2o46bob2o9b2o15bob2o$
4bo22bo18b2o23b2o21bo11bobo12b3o2bo$2b5o14b2o11b2o35bo14b2o4b3o11bo13b
o5bo$bo5bo13bo11bo2bo35b3o11b2o3bo3b2o8b2o14b5o$bo2b3o12bobo6b2o4b2o
38bo16b4o2bo25bo$2obo15b2o6bobo47b2o15bob2o$o2b4o20bo48bobo12b3o2bo$b
2o3bo3b2o14b2o48bo13bo5bo$3b3o4b2o24b2o37b2o14b5o$3bo32bo56bo$2obo33b
3o$2ob2o34bo3$11b2o$12bo$9b3o51bo$9bo53b3o$66bo$65b2o7$75b2o$68b2o5bob
o$68b2o7bo$77b2o2$64bo$63bobob2o$63bobobobo$62b2obobobo2bo$63bo2b2ob4o
$63bo4bo$64b3obo2b2o$66b2o3b2o6$60bo$60b3o$63bo$62b2o7$72b2o$65b2o5bob
o$65b2o7bo$74b2o2$61bo$60bobob2o$60bobobobo$59b2obobobo2bo$60bo2b2ob4o
$60bo4bo$61b3obo2b2o$63b2o3b2o!
-- and any number of other timings using other H-to-G components (see the lower right corner of the "elementary conduits" pattern).

Some of those H-to-Gs would make it easy to connect back to the G-to-H with a short counterclockwise loop instead of a long clockwise one, and maybe pick up a few more [1..7]+8N timings, too. But half of the output gliders will be the wrong color, of course; this can be corrected by adding more Herschel tracks, but then you're halfway to using Hersrch. Another exercise for the reader...!

Post Reply