Swapped typedefs for pcg32_k64_fast & pcg32_k64_oneseq?

[BJTor]

Are these typedefs really correct? (lines 1905 & 1906 of pct_random.hpp):

typedef pcg_engines::ext_mcg_xsh_rs_64_32<6,32,true>        pcg32_k64_oneseq;
typedef pcg_engines::ext_oneseq_xsh_rs_64_32<6,32,true>     pcg32_k64_fast;

(note that pcg_32_k64_onseq is an ext_mcg engine, but pcg32_k64_fast is ext_oneseq. This seems at odds with other definitions where the 'fast' version always means an mcg engine)

[bhaller]

IIRC there is a previous issue pointing out that these are incorrect. @BJTor are you working from the latest GitHub release, or from the current code? A number of problems have been fixed in the current code, beyond the last release. This might be one of them?

[BJTor]

I was using the released version. I'll try again with the latest update. However I just looked in the GitHub repo and these still looked to be swapped: https://github.com/imneme/pcg-cpp/blob/master/include/pcg_random.hpp

However this isn't a big deal - if they are definitely wrong I'll just use my own typedefs.

[bhaller]

However this isn't a big deal - if they are definitely wrong I'll just use my own typedefs.

I can't speak to whether they are definitely wrong; I'm just a user here, not a developer. But see, e.g., #33. I think it was mentioned in another issue too, but I'm not finding it right now.

[BJTor]

OK, thank for the pointer. Looks like this is just a repeat of issue 33 - I'll close this issue. Your help is greatly appreciated. Even just knowing that others are still monitoring and care about this project is reassuring!

Once I've gained some more confidence with this project I might even propose a fix for this - it seems a fairly clear typo and so long as I can work out how to test it, seems an easy enough change.

[BJTor]

Seems like exactly the same issue as #33.

[bhaller]

...Your help is greatly appreciated. Even just knowing that others are still monitoring and care about this project is reassuring!

It seems like a great PRNG overall – wicked fast, statistically sound. I don't know much about this stuff, so I'm out of my depth on it; but I'm sold on it enough that I switched my simulator over to using it. It does seem to have been somewhat abandoned by its author, unfortunately; but it seems to be in good enough shape that it's usable despite that. There might be a fork of it that is better supported; I haven't looked into that thoroughly. It might also be worthwhile to merge some of the open PRs here into it; some people have made PRs that look pretty solid, but again I'm out of my depth so I can't really judge. Let me know if you find out more. :->

I emailed the author maybe two weeks ago to encourage her to do a pass over this and make a new release incorporating bug fixes and stuff. No reply. :-< Life gets busy, I get it; I've got some abandonware on my own. But this seems like a particularly worthwhile project, so it's a shame.