Gustavo Ramos Rehermann's patterns

[Gustavo6046]

I ran the following in ptbsearch:

Code: Select all

.a
..a
aaa
.
...b
..b.b
..b.b
...b

Invoked ptb2 with:

Code: Select all

ptb2 boatcol.life 2 120 1 1 >boatcol.out

And the file was empty!

[dvgrn]

Invoked ptb2 with:

Code: Select all

ptb2 boatcol.life 2 120 1 1 >boatcol.out

And the file was empty!

Ah, good -- you won't get into too much trouble with an empty file... and you can concentrate on finding that LWSS->G record-breaker. It's less than five minutes of work now, if you just take a fresh look at the walkthrough. Or Hersrch can find something in a few seconds, once you get it set up.

Seriously, though, read the ptbsearch README again, and run one of the examples to make sure that that works right:

Code: Select all

ptb2 picol.life ptb2.list 80 2 20 0 > ptbs.out

Notice you're missing "ptb2.list" in your version of the call to ptbsearch, which means you're essentially doing a search without any catalysts.

[Xenonquark]

You can do it Gustavo!

[Gustavo6046]

Is my Hersrch input right?

Code: Select all

hersrch.exe -d 0 -12 73 F -u 73 -l 6 -x 250 -n 10 -f gglog.log -o ggop.rle -m

I want a GFnG where n is anything and the glider changes color.

[dvgrn]

Is my Hersrch input right?

Code: Select all

hersrch.exe -d 0 -12 73 F -u 73 -l 6 -x 250 -n 10 -f gglog.log -o ggop.rle -m

I want a GFnG where n is anything and the glider changes color.

Right or wrong partly depends on which H-to-G you are planning to use.

This is a very good first try, actually. I'm impressed. Most people get scared away from Hersrch without even trying to make it work.

You got about half of the command-line switches right. I see some things that won't work:

"-l 6" doesn't do anything -- there's no such thing as a track length less than 64 anyway.
"-u 73" will keep you from getting get any results.
As you'll see, you'll need a few more than 73 ticks to get the signal to where you want it. Leave out -u and just use the default for now.
"-x 250" is too small -- you'll need more space than 16x16 or so. -x specifies the total number of cells in the bounding box, so the number should be pretty big. Just leave the -x switch out, too.
"-m" you won't need -- this isn't a very long or difficult search.

"-d 0 -12 73 F" -- Output in the same orientation (F) as the start orientation (0 0 0 F by default), so Hersrch won't bend the track around in a loop. I don't quite see why you picked the three numbers. I would expect something like "-d 51 0 100 F" which means "move the Herschel an odd number of cells (51) horizontally, stay on the same line vertically, and the conduit will take at least 100 ticks to get the Herschel signal to that output point".

So that gets us down to

Code: Select all

hersrch -p 1 -u 999 -d 51 0 100 F -n 10 -f gglog.log -o ggop.rle
pause

Theoretically I think this ought to give you some results. Unfortunately it doesn't (!).

For some reason Hersrch doesn't think that any of its conduits are compatible with the -p 1 setting. I asked Hersrch's author about this years ago, but never managed to get this kind of search working, and really I still think this is a bug.

I've never had to worry about it, because the alternate way of specifying multiple targets works really really well... so let's go on to that, because you're going to need to be able to search for multiple output locations anyway.

Here's the idea. You really don't just want to tell Hersrch one place where the signal has to end up -- e.g., in the sample above, exactly at (51,0) relative to the input. There won't be many results for that search, quite possibly none at all if the conduits don't happen to line up right.

Instead, you want to tell Hersrch all the possible places where a signal could go and still give you the right color glider. If there are a lot of possible places, then the odds go way way up that Hersrch will be able to find you a working Herschel track that takes the signal to one of those places.

-- Okay, so we write a script that says basically

put a Herschel on any chessboard square of the same color -- could be BLACK or WHITE depending on how you color it, or you could just say it's ODD cells (sum of coordinates is odd)... but anyway all the same-color squares are okay.

Here's how you set up that script in Hersrch:

Code: Select all

#CXRLE Pos=-30,-31
x = 95, y = 61, rule = LifeHistory
64.D$63.D.D$62.D3.D$61.D5.D$60.D7.D$59.D9.D$58.D11.D$57.D13.D$56.D15.
D$55.D17.D$54.D19.D$53.D21.D$52.D23.D$51.D25.D$50.D27.D$49.D29.D$48.D
31.D$47.D33.D$46.D35.D$4.D40.D37.D$5.D38.D39.D$3.3DB36.DA40.D$4.4B10.
A23.D.3A39.D$5.4B7.3A22.D5.A39.D$6.4B5.A24.D5.2A40.D$7.4B4.2A22.D6.4B
39.D$2A6.9B21.D9.3B19.A7.2D10.D$.A7.6B22.D10.4B16.3A6.DBD11.D$.A.2A5.
6B3.B2.2B2.2B8.D12.10B8.A8.3BD12.D$2.A2.A4.19BD5.D14.5BD4B7.2A6.4B14.
D$3.2AB3.20BDBD2.D16.2B3D4B2.B.5B5.4B16.D$4.14B2A9B3DB2.D14.3BDBD4B.
6B6.4B16.D$5.13B2A11BD4.D14.2BD14B4.4B16.D$6.24B7.D13.24B16.D$6.17B.B
.2B10.D14.21B16.D$7.15B4.3B10.D15.20B14.D$7.15B5.A2B.2A7.D15.20B12.D$
8.13B5.A.A2B.A8.D14.19B2A10.D$10.13B2.A.AB2.A10.D11.8B2.11B2A9.D$9.8B
4.2A.A.A3.A12.D10.2A3.B5.11B9.D$9.6B6.2ABA2.4A.A11.D10.A7.11B10.D$9.
5B8.B2.A.A3.A.A11.D6.3A8.2A5.4B9.D$9.B.B9.2A.2A2.A2.A.A12.D5.A11.A17.
D$10.3B9.A.A2.2A3.A14.D13.3A17.D$9.B2AB9.A.A23.D12.A18.D$10.2A11.A25.
D29.D$50.D27.D$51.D25.D$52.D23.D$53.D21.D$54.D19.D$55.D17.D$56.D15.D$
57.D13.D$58.D11.D$59.D9.D$60.D7.D$61.D5.D$62.D3.D$63.D.D$64.D!

Use Ctrl+Shift+O in Golly to open this template pattern from the clipboard. That will put the input Herschel's center at (0,0), so you can see that you want to move the input Herschel by (+24,-1) and flip it upside down (Fx). The batch file will look like this:

Code: Select all

hersrch -p 997 -o ggop.rle -n 10 -f ggoplog.log -s -e (t=0..300,j=-10..20,k=-10..20)F(0)[0,0]..Fx(t)[24+j+k,-1-j+k]
pause

Hersrch will try any value of t from 0 to 300, any value of j from -10 to 20, and any value of k also from -10 to 20 -- all combinations of those variables in one search.

Try putting in different numbers for j and k, and you'll see that the "Fx(t)[24+j+k,-1-j+k]" at the end of the script basically just means "inside the red diamond in the template, any cell that is the same color as the output Herschel's center cell".

Hersrch will make some RLE for you, showing every different string of conduits that it finds, that moves a Herschel to any correct-color cell inside that red diamond. Just pick the shortest conduit, connect the ends to the pieces in the template, and you're guaranteed to have a color-changing G-to-G.

See how all of that works? Ask questions if you don't. Hersrch is definitely not easy to figure out the first time around!

[Gustavo6046]

GF423G

Code: Select all

x = 115, y = 52, rule = LifeHistory
24.DB$9.A13.BDBD$9.3A10.2B2D$12.A8.4B$11.2A7.4B$11.5B3.4B$13.3B2.4B
51.2A$3.2A7.9B51.B2AB$3.A8.8B53.3B$2A.A.B3.10B41.2A12.B$A2.3AB.2B2A7B
42.A6.B4.3B$.2A2.BA3B2A7B42.A.AB.5B.5B$3.4A12B43.2AB.11B$3.A.2B3.7B.B
2A43.14B$4.3AB2.7B.BA.A42.14B$7.A4.4B5.A35.2A5.14B$2.5A5.4B5.2A26.A6.
B2AB5.13B$2.A10.4B32.3A5.3B5.13B$4.A9.4B34.A3.B.B6.12B$3.2A10.4B32.2A
2.6B5.13B$16.4B31.10B4.15B$17.4B22.2A8.27B$18.4B22.A8.28B$19.4B21.A.A
B6.27B3.2A$20.4B10.A10.2AB.3B2.28B3.A$21.4B7.3A12.36B2.A.AB$22.4B5.A
15.36B3.A.AB.B16.2A$23.4B4.2A15.34B5.A3B2A15.A$16.2A6.9B14.33B3.6B.B
2A12.BA.A$17.A7.6B14.31B9.3B3.B10.2A.B2A$17.A.2A5.6B3.B2.2B2.20B3.12B
.2B.3B.4B12.B2A2B$18.A2.A4.19BD15B5.30B6.4B$19.2AB3.20BDBD4B.7B6.13B
2A11BD4B3.6B$20.14B2A9B3D4B2.6B7.12B2A9B3D4B2.6B$21.13B2A11BD4B3.6B8.
21BDBD4B.7B$22.29B6.4B6.23BD15B$22.17B.B.2B12.B2A2B6.2A.11B2.2B2.21B$
23.15B4.3B12.2A.B2A5.A2.9B10.19B$23.15B5.A2B.2A11.BA.A.3A3.9B12.18B$
24.13B5.A.A2B.A14.A.A4.8B15.18B$26.13B2.A.AB2.A15.2A4.2AB.4B15.19B$
25.8B4.2A.A.A3.A21.A.AB.2B2AB14.19B$25.6B6.2ABA2.4A.A19.A6.2A13.2AB.
3B2.12B$25.5B8.B2.A.A3.A.A17.2A20.A.AB6.10B.B2A$25.B.B9.2A.2A2.A2.A.A
39.A8.10B2.BA.A$26.3B9.A.A2.2A3.A39.2A7.9B7.A$25.B2AB9.A.A55.10B7.2A$
26.2A11.A55.4B.6B$94.4B3.B.B$102.3B$101.B2AB$102.2A!

[dvgrn]

GF423G...

It works! But it's not a color-changer, and with the extra Snark at the beginning it's not even a reflector, more of a lane-changer. For most purposes you can do the same lane change with two Snarks.

If you choose your H-to-G carefully you should be able to make a much smaller color-changing one of these. Remember, there's no need to post all the possible things you can build with these pieces. Anyone who wants a G-to-G can put together a custom one.

By contrast, an actual record-breaking LWSS-to-G is something that nobody knows how to make yet, so that will probably get a bit more attention and interest. (I don't want to promise too much -- in my experience, a Conway's Life discovery doesn't exactly make you world famous or anything.)

[Gustavo6046]

I see no reason why can't us just use the following reaction

Code: Select all

x = 8, y = 10, rule = B3/S23
o$b2o$2o3$2b2o$3bo$3bob2o$4bo2bo$5b2o!

plus a Silver reflector and some snarks to finish this thing dandy.

[thunk]

I see no reason why can't us just use the following reaction

Code: Select all

x = 8, y = 10, rule = B3/S23
o$b2o$2o3$2b2o$3bo$3bob2o$4bo2bo$5b2o!

plus a Silver reflector and some snarks to finish this thing dandy.

There are three reasons.
a) it doesn't produce any sort of outbound signal.
b) Silver reflectors are pretty much deprecated at this point--snarks and syringes can do anything you can.
c) The reaction you found first is actually considerably better and more useful. The LWSS->G solution from that is a lot faster and more obvious--you just need a syringe and the two conduits dvgrn provided in step 3 of the walkthrough.

[biggiemac]

Suppose you have a factory, something that turns a glider into a still life. If you can find a way to hit that still life with an LWSS and get something interesting out without destroying the rest of the factory, then you have something interesting.

In this case, your factory is a boat-bit generator. It has been tried before (but wouldn't be too hard to try again if you wanted to convince yourself) that the boat-bit cannot be turned into anything interesting by an LWSS without destroying the base still life (in your case, a beehive with tail).

The H-to-boat fed by a syringe, effectively a G-to-boat, does allow you to hit the boat with an LWSS and get something interesting out without destroying the factory, because the boat is further from the other pieces. It's as simple as that.

[Gustavo6046]

Something more complicated, like a catalyst perturb the LWSS-boat collision so that the glider appears and a tub, then this tub is hit by that glider to produce a boat and another glider.
Anyway, is this HF117H interesting?

Code: Select all

x = 52, y = 25, rule = LifeHistory
28.2A$29.A$16.A11.A$16.3A9.2A$19.A9.B$6.2A10.2A9.3B$7.A10.5B5.6B$7.A.
AB9.4B3.10B$8.2AB.3B4.6B2.11B3.2B2.2B$10.7B.13B2A15BD$10.21B2A15BDBD$
11.37B3DB$10.40BD$8.41B$6.18B.B3.13B.B$6.2BD13B7.7B.B$5.3BDBD4B.6B$6.
2B3D4B2.B.5B$5.5BD4B7.2A$4.10B8.A$3.4B16.3A$2.2A2B19.A$.A.AB$.A$2A!

hersrch -c 1 -d 5 -36 150 Bx -t 20000 -x 200 -n 5 -f h2h.log -o h2h.rle -s

[Gustavo6046]

Code: Select all

x = 10, y = 14, rule = LifeHistory
3.A$2.A.A$2.A.A$.2A.2A2$.2A.2A$2.A.2A$A$2A2$8.2A$4.2A.2A$3.A.A3.A$4.A
!

[Gustavo6046]

This is possibly an smaller variant of the eater 2 (in terms of cells).

[dvgrn]

Something more complicated, like a catalyst perturb the LWSS-boat collision so that the glider appears and a tub, then this tub is hit by that glider to produce a boat and another glider.

That sounds like a nice theoretical conduit. But until you find an example that works in real Life... it doesn't actually exist. Among other things, there is no glider+tub collision that produces a boat and an output glider. This can easily be verified by experiment.

You might as well say, "Suppose a glider hits a simple constellation, and there's a huge explosion and the sparks spell out the letters 'GUSTAVO', and then the constellation reappears exactly the way it was, and a glider comes out." Sometimes your suggestions sound almost as unnecessarily complicated as that, and just about as unlikely to work out in practice.

Anyway, is this HF117H interesting?

It was very very interesting in the mid-1990s when it was first discovered. Look in the Herschel conduit stamp collection in Golly, and in Hersrch's conduit library. If there's something already in there with the same number of steps and the same output orientation, it's very likely going to be a known conduit.

[dvgrn]

This is possibly an smaller variant of the eater 2 (in terms of cells).

And it shouldn't surprise you that it's very well known already. It was well known back in 2001, and is plainly visible in hundreds of Life patterns that you should look at carefully before you post anything else like this.

Study all the guns in chris_c's new gun collection, for example -- you'll find this eater2 variant in there somewhere.

Just a fair warning: it's actually quite difficult for me to keep answering your posts when you insist on posting well-known things as if they were new, over and over again. I'll probably stop posting responses fairly soon... but I'm really excited about a new smaller faster LWSS-to-G, so I've been trying really hard to give you the tools you need to find it.

So:

Can you modify the Hersrch search script that looked for color-changing G-to-Gs, so that it finds solutions to the LWSS-to-G problem? Start with the template I posted yesterday. Or just use the walkthrough to solve it manually, as several people have suggested now.

You suggested the initial LWSS->G reaction, which apparently nobody else has looked at since the syringe was discovered... so it's your discovery, if you finish it and post it before anyone else.

I've been refusing to do all the work for you, though. You won't learn anything that way.

All your experimenting with LWSS-to-G signal connections recently may have given you a better sense of how very rare and wonderful it is to find Herschel-conduit pieces that fit together in a really tight loop and do exactly what you want.

Herschel conduits are kind of big and awkward, at best, and they usually get in each other's way and prevent tight connections. A complete loop that works in less than 550 ticks is actually quite elegant, as these things go!

[Gustavo6046]

I am searching for Pi reactions that can be made into a conduit with Bellman.
Say this one:

Code: Select all

x = 7, y = 9, rule = LifeHistory
5.A$4.A.A$5.A4$3E$E.E$E.E!

[Gustavo6046]

It is Bell'ing already, and I only get some still lifes that erase the output glider-- and worse, are not catalysts!

Code: Select all

#S max-active 20
#S first-encounter 10
#S last-encounter 90
#P 0 0
.
.
..????????????
..?????????????
..??????????????
..???????????????
.......??????????
.......@?????????
......@.@????????
.......@...??????
..
..
..
..@@@
..@.@
..@.@
#F 73 9 16
*.*
.**
.*

[gmc_nxtman]

How to use bellman.

[Gustavo6046]

Where is precompiled makegif for Bellman?

[Gustavo6046]

All the results were bull****, or at least the fifteen I saw of the dozens.

[Gustavo6046]

Double-R eater:

Code: Select all

x = 28, y = 28, rule = LifeHistory
23.A$22.A.A$22.A.A$21.2A.3A$22.B4.A$20.B2AB3A$13.B.B4.B2A.A$12.9B$11.
10B$11.9B$11.B2E5B$8.3B.B2E4B$7.3BEB.E3B$6.4B3E2.B$7.4BEB$6.8B$7.6B$
7.5B$7.5B$7.3B$5.4B$3.A.2A$.3AB2A$A4.B$.3A.2A$3.A.A$3.A.A$4.A!

[gmc_nxtman]

edited for redundancy

[gmc_nxtman]

Double-R eater:

Code: Select all

x = 28, y = 28, rule = LifeHistory
23.A$22.A.A$22.A.A$21.2A.3A$22.B4.A$20.B2AB3A$13.B.B4.B2A.A$12.9B$11.
10B$11.9B$11.B2E5B$8.3B.B2E4B$7.3BEB.E3B$6.4B3E2.B$7.4BEB$6.8B$7.6B$
7.5B$7.5B$7.3B$5.4B$3.A.2A$.3AB2A$A4.B$.3A.2A$3.A.A$3.A.A$4.A!

This is generation 1 of butterfly, and making an eater is fairly trivial because the output of a butterfly is just two beehives:

Code: Select all

x = 17, y = 17, rule = B3/S23
15b2o$15b2o3$5b3o$4bo$4bo$4bo8$2o$2o!

It is Bell'ing already, and I only get some still lifes that erase the output glider-- and worse, are not catalysts!

Code: Select all

#S max-active 20
#S first-encounter 10
#S last-encounter 90
#P 0 0
.
.
..????????????
..?????????????
..??????????????
..???????????????
.......??????????
.......@?????????
......@.@????????
.......@...??????
..
..
..
..@@@
..@.@
..@.@
#F 73 9 16
*.*
.**
.*

What do you mean by "not catalysts"? Do you mean that the tub is destroyed? This is normal. Any extraneous still lifes in the output file but not in the input file being destroyed is abnormal, and is a malfunction of Bellman as any of those should catalyze.

Bellman is a program that, when given an active or unstable object, attempts to find stable catalysts of that object - Bellman can't take a existing interaction between a stable and unstable object, and replace the stable object with a catalyst. In this example, the only thing that bellman will put out is catalysts that interact with the perturbation you showed - not ones in which the tub is transparent or restored.

In this example, you claimed that you only saw fifteen of dozens of results -

All the results were bull****, or at least the fifteen I saw of the dozens.

It's often worthwhile to check through all of the results if there aren't too many. Do you still know exactly how many outputs there were from Bellman? "Dozens" makes it seem like under 500 results - but when I was a noob with Bellman, I gave unworthwhile searches with over 2000 outputs.

Speaking of unworthwhile searches, I think that #S max-active 20 is a bit much - er, way too much. It takes me days to complete a max-active 10 search, so 20 is pushing the limits.

Even if the search had, say, max-active 5, the search still wouldn't be really worth the time because this reaction (A Pi-heptomino perturbing and destroying a tub to produce a glider which can be done in many ways and is trivial) isn't full of potential.

[Gustavo6046]

I asked Bellman to run a Eater 1 variant.
Here's the result:

Code: Select all

x = 27, y = 40, rule = LifeHistory
8.A2.A$8.4A2$.2A3.8A3.2A$A2.A.A8.A.A2.A$.3A.10A.3A2$.3A.9A.2A$A2.A.A
7.A.A.A$.2A3.6A.A.A.A$12.A3.A$6.3A.2A$5.A2.A.A$6.A.A.A$7.A2.A.2A$10.A
.2A$9.2A$10.A$10.A.AB$11.2AB$13.B2A$13.BA.A$16.A$16.2A$13.2A.A$13.2A.
A2.A$16.A.A.A$16.A.A2.A$15.2A.3A$10.A3.A$9.A.A.A.6A3.2A$9.A.A.A7.A.A
2.A$10.2A.9A.3A2$8.3A.10A.3A$7.A2.A.A8.A.A2.A$8.2A3.8A3.2A2$15.4A$15.
A2.A!

Ugly right?

[Gustavo6046]

Look what my Python induction-coil search program found:

Code: Select all

x = 19, y = 20, rule = LifeHistory
2.A$.A.A.2A$A2.A.2A$.3A$5.2A$3.2A2.A3.2A$2.A3.2A2.A.A$3.3A4.A$7.4A$3.
4A$2.A4.4A$2.2A2.A4.A.2A$5.A.3A.A.2A$6.A2.A.A$7.A.A.6A$8.A7.A$13.2A.A
.A$12.A.A2.2A$12.A.A$13.A!

It takes a initial unstable life (in this case two Eater 1s loosely connected by a long string),

Code: Select all

x = 13, y = 13, rule = LifeHistory
5.2A$4.A.A$4.A$.4A$A$.4A$5.A$5.A$5.A$5.6A$10.A$10.A.A$11.2A!

then randomly add still lifes around the cells that become on near the still life to be inducted. There is a limit of induction coils to be added specified in the command line. If the induction coil happens to stabilize the unstable life, then a solution is generated in a text file and the search continues, if else then the search will pass. Sometimes a induction coil results in another cell borning; in this case the search is smart enough to put other still lifes, up to the max_layer value.

Syntax is:

- python induct.py <inputrle> [outputrle] [max_layer] [max-coils] [coillist] (edit: [maxwelds] too)

Coil lists are text files with Python arrays of RLEs. By default the hardcoded list is used.
EDIT: The program may accidentally weld (as the sharp loaf and the tub in the center southwest) or you may specify a max number of welds used on conflicting coils. This takes time though.
EDIT 2: He also allows extensions from the hardcoded EXTENSIONLIST (I plan to enable a custom extension list the next release.), like here:

Code: Select all

x = 5, y = 7, rule = B3/S23
bo$b3o$4bo$b4o$o$obo$b2o!

Yes, he added the north tail from the initial explosion pattern:

Code: Select all

x = 5, y = 5, rule = B3/S23
4bo$b4o$o$obo$b2o!