Haycat2009's *find searches (Re: B35/S23)

[Haycat2009]

No worries - my roommate works at a research lab, which has a supercomputer. I will ask him to run gfind for 24 hours.

My understanding is that supercomputers rely on parallel architectures for their speed, but gfind is single threaded; it is not written for parallel computing. I imagine a gfind search on a supercomputer wouldn't be much faster than on a home desktop.

The programs I'm aware of that utilize multithreading are LSSS, LLSSS, ikpx2, and qfind. Of these, only some variant of LLSSS has been configured to look for ships that work in two desired rules at once. It wouldn't be hard to modify qfind for this purpose, but qfind will always be at a fixed width, while Amling's LLSSS searches are for arbitrary-width ships. Thus these searches can eliminate the possibility of any ships of the chosen period and speed (if Keith has enough RAM to complete the search).

Other than qfind, are there any other one-letter finds that work? I heard that dfind, jfind and kfind work for parallel computing and OCA, but dfind is prone to finding still lives (due to the fact that it runs a periodicity check), jfind is optimised for quantum computing (Which it is not), kfind is meant to find ships that work slightly differently in 2 different rules and thus it works weirdly.

Which should I use? (qfind only works in Life)

[confocaloid]

I heard that dfind, jfind and kfind [...]

AFAIK neither of those three exists in a non-vapourware form.

Which should I use?

The simplest and most general suggestion is "learn and try everything, and see which tools work for which problems".

(qfind only works in Life)

No, qfind 2.2 supports two-state isotropic rules via the usual Hensel notation. See here for a spaceship partial I just found using qfind, and here for spaceships found by another person in another rule.

[Haycat2009]

I heard that dfind, jfind and kfind [...]

AFAIK neither of those three exists in a non-vapourware form.

Which should I use?

The simplest and most general suggestion is "learn and try everything, and see which tools work for which problems".

(qfind only works in Life)

No, qfind 2.2 supports two-state isotropic rules via the usual Hensel notation. See here for a spaceship partial I just found using qfind, and here for spaceships found by another person in another rule.

These are already up and running, but they are not published yet on the forums. I guess not everyone has heard of it… Anyway, dfind, jfind and kfind have their issues. I told my roommate to run dfind first, as any partials would be way too hard to complete.

Priority:

1. Dfind
2. Kfind
3. Qfind

(Jfind only works on quantum computer, that’s why it is not published)

[confocaloid]

I guess not everyone has heard of it… Anyway, dfind, jfind and kfind have their issues.

Just to be clear: do you really want to derail a thread dedicated to a rule you seem to care about, with an offtopic discussion?

[Haycat2009]

I guess not everyone has heard of it… Anyway, dfind, jfind and kfind have their issues.

Just to be clear: do you really want to derail a thread dedicated to a rule you seem to care about, with an offtopic discussion?

No. Where do I put it? Anyway, the results will be left here.

EDIT: Dfind, just as expected, left me the block. QFind may work.

[Haycat2009]

QFind’s longest partial is unhealthy and only runs forward for 3 gens. Dfind returned a really large still life, and each of Kfind’s partials requires at least one B5 condition turned off. So, no results.

[dvgrn]

QFind’s longest partial is unhealthy and only runs forward for 3 gens. Dfind returned a really large still life, and each of Kfind’s partials requires at least one B5 condition turned off. So, no results.

What are "Dfind" and "Kfind", exactly? If there's existing code that's being run as you describe, then please post a copy of or a link to that existing code, before posting any more reports on the results of running that code.

Long story short, your posts so far don't seem to be contributing to collective knowledge about B35/S23; what you've said so far is vague and confusing enough that the posts are starting to get reported.

[Haycat2009]

I just cannot get other results other than meaningless partials - essentially, there is nothing to contribute.

[Para]

I just cannot get other results other than meaningless partials - essentially, there is nothing to contribute.

Do you have source code, say, in C - for dfind, kfind, and jfind? I would very much like to know their methodology and approach, as well as what types of patterns they search for.

[Haycat2009]

I just cannot get other results other than meaningless partials - essentially, there is nothing to contribute.

Do you have source code, say, in C - for dfind, kfind, and jfind? I would very much like to know their methodology and approach, as well as what types of patterns they search for.

Not yet - hard to find the source code of programs that they do not want you to see. Anyways, I suspect that it is obfuscated - apparently you can do that.

I officially give up spaceship-searching until 2035 - this has progressed to be ridiculous.

[Para]

I just cannot get other results other than meaningless partials - essentially, there is nothing to contribute.

Do you have source code, say, in C - for dfind, kfind, and jfind? I would very much like to know their methodology and approach, as well as what types of patterns they search for.

Not yet - hard to find the source code of programs that they do not want you to see. Anyways, I suspect that it is obfuscated - apparently you can do that.

I officially give up spaceship-searching until 2035 - this has progressed to be ridiculous.

I'm skeptical as to if these programs actually exist - why do "they" not want you to see them? Where is the code of those programs hosted?

[Haycat2009]

I just cannot get other results other than meaningless partials - essentially, there is nothing to contribute.

Do you have source code, say, in C - for dfind, kfind, and jfind? I would very much like to know their methodology and approach, as well as what types of patterns they search for.

Do you know the chrome source code?

[Para]

I just cannot get other results other than meaningless partials - essentially, there is nothing to contribute.

Do you have source code, say, in C - for dfind, kfind, and jfind? I would very much like to know their methodology and approach, as well as what types of patterns they search for.

Do you know the chrome source code?

Actually, yes - Chromium is completely open source and its github repo can be found here: https://github.com/chromium/chromium
So you are implying that you can use these search programs, but cannot view their code?

[Haycat2009]

Chrome code is public

Ok, fine. The point is that the code is obsfuscated, and I do not know how do deal with that. Also, my friend already told me not to send anyone the code as “those nasty people will publish the code under their name and give us researchers no credit”.

I do not want to make this a personal problem, but pressure from you is not working. I see your intent - nice cover story.

[dvgrn]

I officially give up spaceship-searching until 2035 - this has progressed to be ridiculous.

That does seem like a good summary. @Haycat2009, another of these posts got reported. Please have a look at the tangle of vague and implausible statements that you got yourself into in this case, and then please be careful not to post anything along these lines again.

If you have verifiable results to report, then report them. Reports about results that you can't actually make public are, quite simply, not welcome here -- not even in the Sandbox.

I'll go ahead and lock this thread now.