Reflectorless Rotating Oscillator Discussion Thread in Life

For discussion of specific patterns or specific families of patterns, both newly-discovered and well-known.
Post Reply

User avatar
BlinkerSpawn
Posts: 1992
Joined: November 8th, 2014, 8:48 pm
Location: Getting a snacker from R-Bee's

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by BlinkerSpawn » January 27th, 2018, 11:49 am

I'm certain we have more than sufficient technology to make a self-constructing/destroying loop.
LifeWiki: Like Wikipedia but with more spaceships. [citation needed]

Image

User avatar
Macbi
Posts: 903
Joined: March 29th, 2009, 4:58 am

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by Macbi » January 27th, 2018, 11:55 am

The problem is that any kind of Geminoid construction won't work, because the two pieces would move by different amounts.

User avatar
BlinkerSpawn
Posts: 1992
Joined: November 8th, 2014, 8:48 pm
Location: Getting a snacker from R-Bee's

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by BlinkerSpawn » January 27th, 2018, 12:53 pm

Macbi wrote:The problem is that any kind of Geminoid construction won't work, because the two pieces would move by different amounts.
I was thinking more along the lines of a loop a la the newest spiral-growth except I don't know how to put new elbows at the opposing corners since you can't just leave them there.
LifeWiki: Like Wikipedia but with more spaceships. [citation needed]

Image

User avatar
dvgrn
Moderator
Posts: 10655
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by dvgrn » January 28th, 2018, 1:32 am

BlinkerSpawn wrote:
Macbi wrote:The problem is that any kind of Geminoid construction won't work, because the two pieces would move by different amounts.
I was thinking more along the lines of a loop a la the newest spiral-growth except I don't know how to put new elbows at the opposing corners since you can't just leave them there.
This stuff is slightly tricky but all definitely known technology, a mix of mechanisms previously seen in the 10hd and 0hd Demonoids and in the linear propagator. Latest progress on a potential RRO design is posted in the Self-Synthesizing Spaceship thread.

It may not be terribly obvious, except maybe with hindsight, that a self-synthesizing spaceship can be configured to travel in three different ways, with pretty much the same amount of design work:
  • diagonally (optionally with a slight oblique shift for extra exotic flavor)
  • orthogonally, by mirror-reflecting after each cycle and thus traveling in a zigzag
  • in a loop, either four steps of 90 degrees or two steps with a 180-degree turn, making a reflectorless rotating oscillator.
EDIT: Now that you mention it, though, it's not necessarily true that "any kind of Geminoid construction" wouldn't work! There's a possible RRO design that's also a mix of a Demonoid and the linear propagator, that would probably be somewhat simpler than a self-synthesizing spaceship memory loop design -- it wouldn't require two copies of the construction data.

Let's see if it still looks good after I run through a summary:
  • start with a Demonoid, but replace let's say the SW end with a constructor arm attached to a 90-degree reflector, not the usual 180-degree one. This end of the RROnoid will pretty much rotate in place.
  • All construction is done from this 90-degree construction-arm end -- the 180-degree end will be just a simple reflector
  • First the constructor builds and launches a Cordership seed. The Cordership heads back in the direction that the recipe gliders are coming from.
  • Next the constructor sends a Cordership-stopping recipe either in the opposite direction, or in the 90 degrees counterclockwise direction, to catch a Cordership that was launched one or two cycles ago. (Which of these allows the timing, or rather the spacing, to work out correctly? It's too late at night for me to sort it out, but I think one of them is right.)
  • The Cordership-stopping recipe builds a new 180-degree reflector end in the correct location.
  • The recipe builds a new 90-degree construction arm.
  • The recipe sends gliders to shoot down the previous 90-degree construction arm _and_ the previous 180-degree reflector, using either long-distance sniping or, more likely, an elbow block left near the old 180-degree reflector for this purpose.
  • The recipe gliders reflected by the 90-degree construction arm bounce off an existing 180-degree reflector (constructed in a previous cycle) and return to feed into the newly constructed 90-degree construction arm.
  • Done! The cycle repeats.
  • To make cleanup easier, the 90-degree and 180-degree circuits could be easily wired to self-destruct after use, with a single trigger glider coming in from somewhere -- lots of options there.

User avatar
77topaz
Posts: 1496
Joined: January 12th, 2018, 9:19 pm

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by 77topaz » January 28th, 2018, 3:46 am

dvgrn wrote:in a loop, either four steps of 90 degrees or two steps with a 180-degree turn, making a reflectorless rotating oscillator.
Would a loop of two steps with a 180-degree leave enough room for a second copy of the oscillator to cycle at the same time, though? Because, to have a proper RRO (of period p), it needs to be possible to have n copies loop the same circuit to create a single oscillator of period p/n, for some n greater than one. This should be possible with a four-step loop, but I'm not sure if there would be enough space in a two-step loop without the copies interfering with each other.

User avatar
dvgrn
Moderator
Posts: 10655
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by dvgrn » January 28th, 2018, 8:42 am

77topaz wrote:
dvgrn wrote:in a loop, either four steps of 90 degrees or two steps with a 180-degree turn, making a reflectorless rotating oscillator.
Would a loop of two steps with a 180-degree leave enough room for a second copy of the oscillator to cycle at the same time, though?
Maybe not, you're right. Was thinking about the multiplicity problem after posting -- I don't think any of the pieces get in each other's way for the 90-degree case, but would have to do a detailed layout to make sure.

We could patch up the 180-degree case by building two 180-degree reflectors next to each other at the Cordership shoot-down point, and a 180-degree reflector plus a 0-degree constructor arm in the center. Send the recipe back and forth twice. That would create enough space, I believe, though current versions of HashLife would be even less happy with it.

GUYTU6J
Posts: 2200
Joined: August 5th, 2016, 10:27 am
Location: 拆哪!I repeat, CHINA! (a.k.a. 种花家)
Contact:

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by GUYTU6J » June 23rd, 2019, 11:11 pm

Sorry for necroposting.Could we just construct a RRO entirely composed of gliders with this naive idea?
1.Start with the all-glider phase.The 1st glider hits the 2nd glider, which creates an elbow.
2.The following slow salvo constructs the freeze-dried salvo of the whole RRO.
3.The last glider triggers the seed, regenerating the original all-glider phase but at another orientation.
Could we make more with less in step 2? And would it be easier to regenerate at another orientation than at original orientation(possibly with displacement)?
EDIT:The step 3 could have trouble with the order of seed actication.

User avatar
Moosey
Posts: 4306
Joined: January 27th, 2019, 5:54 pm
Location: here
Contact:

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by Moosey » June 24th, 2019, 8:13 am

GUYTU6J wrote:Sorry for necroposting.Could we just construct a RRO entirely composed of gliders with this naive idea?
1.Start with the all-glider phase.The 1st glider hits the 2nd glider, which creates an elbow.
2.The following slow salvo constructs the freeze-dried salvo of the whole RRO.
3.The last glider triggers the seed, regenerating the original all-glider phase but at another orientation.
Could we make more with less in step 2? And would it be easier to regenerate at another orientation than at original orientation(possibly with displacement)?
EDIT:The step 3 could have trouble with the order of seed actication.
I think that’s probably possible, but it would be large in terms of cell count. The alternative would, of course, be a very large (in bounding box) 175-cell RRO, which would certainly be possible and would also be smaller, but it’s also not as interesting an idea. However, your idea is better, in my opinion, than just calling on RCT-magic. I think
I wrote:Dvgrn probably has a better answer that’s also conveniently divided up into sections.
not active here but active on discord

User avatar
dvgrn
Moderator
Posts: 10655
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by dvgrn » June 24th, 2019, 8:32 am

GUYTU6J wrote:1.Start with the all-glider phase.The 1st glider hits the 2nd glider, which creates an elbow.
2.The following slow salvo constructs the freeze-dried salvo of the whole RRO.
3.The last glider triggers the seed, regenerating the original all-glider phase but at another orientation.
Could we make more with less in step 2?
Not in any simple way, I'm afraid. With current technology you'll need at least five or ten gliders per freeze-dried still life, and let's say two still lifes per output glider. That adds up to making less with more, not the other way around.

It's hard to imagine an encoding system that can get to "more with less", without building very complex temporary memory loops and switching systems to re-use subrecipes many times. That would be an impressive feat of engineering if someone managed to put it all together; it's still highly non-trivial to come up with a mechanism that ends up reconstituting the exact same recipe that came in.

It's convenient to have two copies of the early part of the recipe, at least -- the part that does all the initial construction. It's hard to see how to encode that initial recipe data to get any kind of decent compression.

Once you have a memory-loop mechanism you could catch the later part of the recipe, make a copy of it while also using it to send out a new copy of the early part of the recipe. Then send the later part of the recipe directly from the memory loop while also triggering a self-destruct sequence on the memory loop mechanism.

But it seems much, much simpler to avoid most of the complicated reconstitution machinery by just having two complete copies of the recipe follow one another. This may be just a failure of imagination on my part, though. If so, I'll be very interested to see what someone constructs to improve on the two-copies-of-recipe design.

User avatar
Gustone
Posts: 744
Joined: March 6th, 2019, 2:26 am

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by Gustone » June 25th, 2019, 11:21 am

What about ELEMENTARY rro?

User avatar
Moosey
Posts: 4306
Joined: January 27th, 2019, 5:54 pm
Location: here
Contact:

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by Moosey » June 25th, 2019, 11:29 am

Gustone wrote:What about ELEMENTARY rro?
Currently it looks like (small ones) don't exist-- they'd be large statorless high-period oscillators.
not active here but active on discord

User avatar
Goldtiger997
Posts: 762
Joined: June 21st, 2016, 8:00 am

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by Goldtiger997 » August 21st, 2021, 5:00 am

I made an RRO, which is quite similar to the recently constructed self-synthesising oblique-loopships. However, the RRO is not quite self-synthesising, though it would possible to make it so with some fairly small modifications. The following RRO has a period of 289487508:

RRO.mc
(1.3 MiB) Downloaded 161 times

Here's the above RRO but with 6 oscillators placed into the loop, which reduces the period to 48247918. Unfortunately, this pattern runs very slowly on my computer:

RRO-6-loop.mc
(9.77 MiB) Downloaded 93 times

As can be seen in the above pattern, there's still a substantial amount of room between each RRO, so it should be possible to fit more RROs into the loop.

User avatar
calcyman
Moderator
Posts: 2936
Joined: June 1st, 2009, 4:32 pm

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by calcyman » August 21st, 2021, 5:41 am

Goldtiger997 wrote:
August 21st, 2021, 5:00 am
I made an RRO, which is quite similar to the recently constructed self-synthesising oblique-loopships. However, the RRO is not quite self-synthesising, though it would possible to make it so with some fairly small modifications. The following RRO has a period of 289487508:
Wow!!! (Remind me to vote for that in POTY 2021.)
Goldtiger997 wrote:
August 21st, 2021, 5:00 am
Here's the above RRO but with 6 oscillators placed into the loop, which reduces the period to 48247918. Unfortunately, this pattern runs very slowly on my computer:
I'd recommend running at a step size of 8^5; that runs relatively smoothly on my laptop.

The period can be increased by adjusting the time before igniting the crabstretcher, right? If we know how the period of the RRO varies as a function of recipe length, then that should allow us to calculate the theoretical maximum loopability of a Goldtiger RRO.
What do you do with ill crystallographers? Take them to the mono-clinic!

User avatar
dvgrn
Moderator
Posts: 10655
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by dvgrn » August 21st, 2021, 8:36 am

Here's a 7-fold version of the RRO -- period 41354700, base period 289482900. To get a period that was a multiple of 7, I reduced the gaps between recipes by two cells. Even that small change reduces the period by 4608 ticks.
RRO-7-loop-p41354700.mc.gz
7-fold version of RRO with slightly reduced memory loop size
(3.26 MiB) Downloaded 72 times
This can be set up as a Golly timeline, which can be animated very quickly once the frames have been recorded. If you don't know about Golly timelines, they're worth experimenting with for patterns like this: View > Show Timeline, pick a reasonable step size like 2^16 (seems good for this RRO), hit the Record button at the bottom of the screen, let it run for a while, then hit the same button to stop recording.

Timeline files can be saved in .mc or .mc.gz format, which means you don't have to re-do the calculations -- just load the file and it can be run forwards and backwards as fast as you want. I won't try attaching a timeline file here, though -- they turn out to be over 10MB even for a single cycle of this 7-fold RRO, probably because too many new hashtiles are needed to handle the burning crabstretcher fuses.

User avatar
calcyman
Moderator
Posts: 2936
Joined: June 1st, 2009, 4:32 pm

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by calcyman » August 21st, 2021, 10:06 am

dvgrn wrote:
August 21st, 2021, 8:36 am
Here's a 7-fold version of the RRO -- period 41354700, base period 289482900. To get a period that was a multiple of 7, I reduced the gaps between recipes by two cells. Even that small change reduces the period by 4608 ticks.
Thanks! I was able to copy your adjustment technique and, with the help of lifelib, assemble an 11-fold version of the RRO (period 74605116, base period 820656276). This is the limit of loopability possible with this design.
Attachments
rro11b.mc.gz
11-fold RRO
(5.13 MiB) Downloaded 76 times
What do you do with ill crystallographers? Take them to the mono-clinic!

User avatar
Goldtiger997
Posts: 762
Joined: June 21st, 2016, 8:00 am

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by Goldtiger997 » November 17th, 2023, 12:39 pm

I've been looking into making a RRO with higher loopability. The way I'm thinking of doing it is with a design much like the SSOL except that every nth cycle the RRO makes a turn rather than continuing straight. For the RRO to be able to decide between these different actions we need there to be some sort of variable state. My plan is to have the state encoded in the position of the ringmaster glider. After each cycle, the ringmaster glider is advanced. This process continues until a critical point is reached. The result of reaching the critical point is that the single-channel stream is sent out in a different direction for that cycle and the ringmaster glider is delayed. The RRO should then be back to its initial phase except rotated and translated. With that high-level overview out of the way, here's my current progress on the design:

Code: Select all

x = 1382, y = 1592, rule = LifeHistory
503.2A$503.A.A$505.A4.2A$501.4A.2A2.A2.A$491.A9.A2.A.A.A.A.2A$490.A.A
11.A.A.A.A$491.2A12.2A.A.A$509.A2$495.2A$496.A7.2A$496.A.A5.2A$497.2A
7$507.2A$507.A$508.3A$510.A167$701.A$699.3A$698.A$698.2A3$706.2A$707.
A$707.A.2A$678.A20.2A4.3A2.A$677.A.A19.2A3.A3.2A$677.2A25.4A$690.2A
15.A$689.A.A12.3A$689.A13.A$688.2A14.5A$708.A$706.A$706.2A70$64.2A$
51.2A11.2A$50.A.A$51.A9$7.2D$7.2D$68.3D$68.D$69.D$25.A5.2A$24.A.A4.A.
A$24.2A6.A11$97.2A$98.A$98.A.A$99.2A7$95.A$95.3A$98.A$8.2A87.2A$8.A.A
$9.A6.A$15.A.A$15.2A$118.2A$118.A.A$50.2A68.A$20.2A27.A2.A67.2A$2A18.
2A28.2A$2A$115.2A$115.A.A$49.2A66.A$49.2A66.2A2$45.2A2.2A12.2A$45.2A
2.2A12.A.A128.A$64.A.A126.A.A$65.A127.2A$106.2A$106.2A$98.2A$99.A$96.
3A89.A$96.A90.A.A$53.2A132.A.A$53.A.A41.A90.A$54.A.A39.A.A$55.A40.A.A
$94.3A.2A$93.A$94.3A.2A93.2A$96.A.2A92.A2.A$186.A5.A.A$106.2A77.A.A5.
A$106.2A7.2A68.2A$115.A$67.2A4.2A38.A.A$29.A38.A4.2A38.2A$28.A.A34.3A
$29.2A34.A2$93.2A$93.2A5$109.A$108.A.A$108.A.A56.A11.2A$109.A56.A.A
10.2A$110.3A20.2A31.A.A$112.A21.A25.2A2.3A.2A$134.A.A23.A2.A$61.D73.
2A21.A.A3.3A.2A$61.D.D94.2A6.A.2A$61.2D37.3D$100.D$101.D$47.2A9.2A49.
A$47.2A8.A.A48.A.A$58.A49.A2.A$109.2A45.2A13.2A$156.2A13.2A$106.2A78.
2A$105.A.A77.A2.A$105.A80.2A.A$104.2A83.A$189.2A$174.2A$175.A$147.2A
23.3A$48.2A97.A24.A$48.A.A42.2A53.3A$49.A43.A.A5.2A38.2A7.A$94.2A4.A
2.A37.A$100.A.A39.3A$101.A42.A8$98.2A$98.A.A$99.A.A$100.A3$98.2A$97.A
2.A$98.2A23$59.2D$59.2D63$356.A$355.A.A$355.A.A$356.A3$361.2A$360.A2.
A$361.A.A$362.A9$371.A$369.3A$368.A$367.A.A$367.A.A$368.A5$352.2A$
352.2A4$372.2A$372.A.A$374.A$365.2A7.2A$365.2A2$355.A.2A$353.3A.2A$
352.A$353.3A.2A$355.A.A$355.A.A$356.A2$355.A$355.3A$358.A$357.2A$365.
2A$365.2A6$376.2A$376.A$374.A.A$374.2A4$376.2A$376.2A6$356.2A$357.A$
354.3A$354.A10$276.2A$276.A.A$277.A10$389.3A$389.A$390.A122$442.2D$
441.2D$443.D40$484.2C$483.2C$485.C132$689.3A$689.A$690.A25$717.2A$
716.2A$718.A20$740.A$739.2A$739.A.A22$762.3A$762.A$763.A20$786.A$785.
2A$785.A.A21$808.2A$807.2A$809.A22$831.3A$831.A$832.A20$855.A$854.2A$
854.A.A21$877.2A$877.A.A$877.A33$911.3A$911.A$912.A37$951.2A$950.2A$
952.A22$975.2A$974.2A$976.A20$998.A$997.2A$997.A.A28$1028.A$1027.2A$
1027.A.A21$1050.2A$1050.A.A$1050.A22$1075.A$1074.2A$1074.A.A25$1101.
2A$1101.A.A$1101.A21$1123.3A$1123.A$1124.A20$1146.2A$1146.A.A$1146.A
21$1170.A$1169.2A$1169.A.A45$1217.A$1216.2A$1216.A.A26$1243.3A$1243.A
$1244.A20$1266.2A$1266.A.A$1266.A21$1290.A$1289.2A$1289.A.A21$1312.2A
$1312.A.A$1312.A21$1334.3A$1334.A$1335.A20$1357.2A$1357.A.A$1357.A21$
1379.3A$1379.A$1380.A!
Running the above pattern as given, after the copy of the single-channel stream is emitted to the NW, the relative positioning of the leading tape glider and the ringmaster glider has the ringmaster glider advanced by 4 ticks. On the other hand, if the white glider is made red and the nearby red glider is made white (effectively advancing the ringmaster glider by 168 ticks), the copy of the single-channel stream is instead released to the SW and the ringmaster glider's relative positioning is back to the original positioning.

The glider that is emitted to the NE needs to bounce around the loop, destroying the reflectors as it goes. When it reaches the final corner of the loop, seeds of destruction need to be rigged so that the three gliders marked in red are fired. These gliders destroy the remaining constellation NW of the Scorbie splitter pair. Importantly, depending on the initial position of the ringmaster glider, a different constellation results, but those three red gliders always cleanly destroy the constellation regardless. This was the trickiest part of the design -- making it so that all the one-time circuitry is used up irrespective of which "action" the RRO takes. I ended using quite a lot of still-lifes, so I'd be interested if anyone has some simplification ideas.

Another thing I should explain is the red blocks. Without the red blocks, child ringmaster gliders are released to both the NW and the SW, but we only want a child in the same direction as the output single-channel stream. The idea is to block the child ringmaster gliders by default with a block, but use the construction arm that is unblocked to fire a 90 degree glider to delete the block, making it so that the child with matching direction is unblocked.

User avatar
dvgrn
Moderator
Posts: 10655
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by dvgrn » November 17th, 2023, 1:12 pm

Goldtiger997 wrote:
November 17th, 2023, 12:39 pm
My plan is to have the state encoded in the position of the ringmaster glider. After each cycle, the ringmaster glider is advanced. This process continues until a critical point is reached. The result of reaching the critical point is that the single-channel stream is sent out in a different direction for that cycle and the ringmaster glider is delayed. The RRO should then be back to its initial phase except rotated and translated.
That should work! And it's a very good design pattern to have in the toolkit.

Could a mechanism along these lines be set up to produce various angles of oblique self-constructing spaceship? Let's say there are N possible positions for the ringmaster glider.

If (N-2) of the positions result in the single-channel stream continuing in the same direction, one of the remaining two positions turns the single-channel stream 90 degrees, mirror-reflects the single-channel target, but does not delay the ringmaster glider, and the last position turns the single-channel stream 90 degrees again (this will turn in the other direction because of the previous mirror-reflection) and again mirror-reflects the single-channel target but this time also delays the ringmaster glider ... then that would produce some number of moves in one direction followed by a "side-step", instead of an RRO.

Did I get that summary right, or is there something that wouldn't work properly there? Come to think of it, I guess maybe it's just as easy to not muck around with mirror-reflecting the target, and just arrange for single-channel outputs in three possible directions?

User avatar
Goldtiger997
Posts: 762
Joined: June 21st, 2016, 8:00 am

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by Goldtiger997 » November 19th, 2023, 5:34 am

I built my high-loopability RRO design. Each "cycle" takes 21436526 ticks with the exception of the cycles that turn the corner which take an additional 37 ticks. The RRO repeatedly performs 42 going forward cycles and then 1 turning cycle, so the period is (21436526*43 + 37)*4 = 3,687,082,620. I think it has maximum loopability 87, but I suspect it will be possible to increase that by modifying the size of the memory loops. It takes a long time to watch the RRO go round the whole loop, so to minimise the tediousness of viewing both the going forward and turning actions I've attached the RRO in a phase just before it starts to turn.
loopability 87.mc
(1.28 MiB) Downloaded 17 times
dvgrn wrote:
November 17th, 2023, 1:12 pm
Could a mechanism along these lines be set up to produce various angles of oblique self-constructing spaceship? Let's say there are N possible positions for the ringmaster glider.

If (N-2) of the positions result in the single-channel stream continuing in the same direction, one of the remaining two positions turns the single-channel stream 90 degrees, mirror-reflects the single-channel target, but does not delay the ringmaster glider, and the last position turns the single-channel stream 90 degrees again (this will turn in the other direction because of the previous mirror-reflection) and again mirror-reflects the single-channel target but this time also delays the ringmaster glider ... then that would produce some number of moves in one direction followed by a "side-step", instead of an RRO.

Did I get that summary right, or is there something that wouldn't work properly there?
That sounds like it would work except that I think you would need the first of the remaining two positions to modify the positioning of the ringmaster glider. If the positioning is not modified, then the state doesn't change, but we'd need it to change since the spaceship needs to something different on the next cycle, namely delay the ringmaster glider.

Having three different actions sounds like it will require quite a lot of one-time circuitry. It's quite tricky ensuring all the one-time circuitry is used up regardless of which action is taken. Maybe the same sort of approach can be used as I did in my design in my previous post. The critical piece of one-time circuitry was the following constellation which emits a glider on the same lane for two different lanes (which are 5hd apart) of input glider:

Code: Select all

x = 16, y = 16, rule = DoubleB3S23
5.2C$4.C2.C$5.2C4$4.2C$4.2C$14.A$2C2.2C7.2A$2C2.2C7.A.A3$14.B$13.2B$13.
B.B!

User avatar
dvgrn
Moderator
Posts: 10655
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Reflectorless Rotating Oscillator Discussion Thread in Life

Post by dvgrn » November 20th, 2023, 7:20 pm

Goldtiger997 wrote:
November 19th, 2023, 5:34 am
Having three different actions sounds like it will require quite a lot of one-time circuitry. It's quite tricky ensuring all the one-time circuitry is used up regardless of which action is taken.
There are quite a lot of known tricks for dealing with that kind of thing, these days. Inspection of 0E0P metacell circuitry will produce a lot of ideas along these lines.

The problem usually isn't designing something that will get used up regardless of which action is taken. The problem is coming up with something that is close to optimal, or at least ... looks like it might be close to optimal.

The idea that will probably always work, but might not end up looking particularly optimal, is to build a network of splitters and clean "detonators", connected to every one-time splitter/turner that might or might not end up getting used.

It usually seems to be possible to find a fairly cheap detonator even for periodic objects, and of course for stable objects it's easier. The slowest glider + block mutual annihilation seems to be really good at cleaning up optional objects nearby, so that would be an easy choice at least for a first-draft design. No idea how much might be saved on average by a totally customized design that somehow avoids adding extra detonators.

Post Reply