ConwayLife.com

A crude idea for a new type of search program:

Between the low period spaceships that can be found by gfind and the like and the generic gemini construction there is a lot of leeway, so far only filled by corder ships and the caterpillar.

How about a search program for space ships only built from small common constellations and reactions betweeen them?

The first ingredient would be a database of collisions between these common constellations (e.g. b-heptomino, pi or r-pentomino), that lists the outcome and the relative spacing of these.

The hypothetic spaceship/puffer would consist of a sequence of several such collisions that support each other vice versa.

In each generation there would be partly normal evolution of the common constellations and partly collisions at work.

The search program would start with say a constellation that is the upper left end of the spaceship and then be trying to recursively find supporting reactions in the database that produce what is the current partial result.
It would be necessary to classify some of the collision results as debris to restrict the search space. The program would not try to follow up on this debris part.

All in all still a pretty vague idea. Could some of you give it some thought?

Any other ideas on how to find or construct something like a corder engine very welcome!

Thanks in advance for any efforts,
Hartmut

The simplest potential case seems like a ship composed of just two Bs in a ladder-like setup… one releases its first glider back at another one's final explosion stage; this produces a new B that then races forward to release its first glider at the other exploding. The timing constraints would be so tight that the probability of the existence of a B-ladder seems piss poor, though.

Another thing that seems like ought to be at least checked out is reactions between two parallel pi-heptominos, in case there are any that cleanly release the 9c/30 child pi of one (or even both?!)

A different basic mechanism entirely would be having a seed repeatedly crash into a junk constellation to recreate both some ways off.

An easier intermediate to search for might be finding more "tracker" reactions as used in the caterpillar, pi ship, etc. If some then turn out to produce upstream junk, these could be potentially chained together to assemble at least a larger tracker (most probably a growing of shrinking one). And some could of course be used as engineering fodder for new caterpillars.

Anyway, given an assembly kit of these, a loop might eventually be closeable by a glider or *WSS released parallel-ish to the stream (which would in no way need be orthogonal or diagonal).

Also, a simple sub-linear growth pattern idea: shoot a glider at some junk such that the junk ends up recreated and a glider is sent back (say, B guts + some extra stuff), at another similar stream growing at a 45° mirrored direction to the 1st. Basically this'd be reflector that shifts both itself (2N lanes) and the glider (N lanes). But again, we'd first want to know how to search for reflectors that shift itself *any* amount distinct from zero…

/€0.02

I don't know about a database of reactions that could be combined. It sounds good at first, but without some new idea, there are just too few useful reactions available, and too many miracles that would have to occur.

I think that searching for a new or overlooked reaction is probably a better bet than any of the specific reactions that have been suggested as the basis for a new spaceship.

I do still think the following idea of mine from a few years back may be workable. (I'm not sure whether it's been made public before.) It's not based on a continuous reaction, so it's not tied to a specific velocity.

------------

A glider could bounce back and forth between two small objects and form a spaceship, if only the collision reactions displaced the objects and the glider by just the right amount. That's too much to hope for, but what if the reaction just reflected the glider, produced an additional glider that escapes, and left behind some junk? For example:
Then, sometime later, a flotilla of spaceships flies by and repairs the junk:
(The patterns given here are just a proof of concept. You would probably not want to use these specific patterns.)

You also need a similar reaction to reflect the glider back:
Since the flotillas are unharmed, they can support an arbitrary number of these glider relays, each of which periodically releases a glider. The gliders will have to be arranged to construct (and destroy) the flotillas. So, the gliders will have to be able to collide -- if nothing else, we could make two parallel tracks like this that fire gliders toward each other.

The overall spaceship could perhaps move in any direction, but I think it's easiest if it moves in the same direction as the flotillas, as in my examples.

The point is that there is a lot of flexibility in the position of the glider relays. They can be moved up and down (by any multiple of 2 cells), and the glider can be advanced by any number of generations (except for a forbidden zone where the flotillas would get in the way).

I'm pretty sure that your concept will work if only someone puts enough energy into it.

What I imagine though is a puffer where few active areas run through a field of still lifes, found by a search program that plugs together known reactions using backtracking. If unknown collision reactions are recognized or the goal period/offset can no longer be reached.

Each reaction is stored as two or more inputs with relative offset and its result is an output that consists of two ore more constellations that can then be used as input for further reactions.

All the classic dicoveries in that direction were made by looking at the screen, finding some remarkable pattern that repeats itself and then trying to stabilize that reaction with eaters. This was surely the case for the switch engine, queen bee shuttle, ...

As I am not really sure how my idea would work manually, I don't see how to get this concept into a running program. I don't even have a library of reaction that could be used as a basis for the search.

Which is the point where I hope for clever ideas from other people.

Rgds,
Hartmut