Almost everyone initializes their engine by a single integer from random generator. Mersenne twister has huge state and almost everybody initializes it with single integer, that gives you only 232 different sequences. Correct usage is to gather enough "true" random bits into vector or array and pass this data through seed_seq like interface into mersenne twister constructor.
Walter Brown has great presentation skills. Among my favorites. There are many good presenters, but I think that him and Andrei Alexandrescu stand apart for different reasons. They are really good at exposing. Walter Brown's metaprogramming Compendium is impressively well explained too.
I'm guilty of "rolling my own" and getting it wrong - with RNG as well as other things. When nobody is relying on your bad code for anything important, it's OK ; you learn from your mistakes and move on. It's pretty nice to do less work and get a much better result. For RNG I've been using Melissa O'Neil's "randutils" header, which builds on the excellent interfaces & facilities provided in to give an interface that's more convenient for most uses. One highlight is good seeding by default using a variety of local sources of entropy. std::random_device seems like it can help with seeding but it can't be relied on for that (not portably at least). I've heard that Melissa's work (or something much like it) may make it into the language spec.
Dr. Brown mentioned Professor O'Neil's work in his talk, and that her PCG random generator might make it into STL (not into the core language specification). She also co-authored a proposal titled "Simplifying simple uses of " (www.open-std.org/jtc1/sc22/wg21/docs/papers/2016/p0347r0.html)
Yes, it is indeed one big fat document, e.g. the working draft for C++17 here: open-std.org/JTC1/SC22/WG21/docs/papers/2016/n4606.pdf In general, it seems to be common to distinct between "Core Language Specification" and "Library Specification" (sections 17-30 in said document). I figured you'd refer to the former.
It would be interesting to see how the interfaces of random could be adapted to make proper use of quasi random sampling (low discrepancy sequences). I'll like an interface where I could use either.
![[Full Episode] The Restaurant War Thailand ศึกพ่อค้าซ่าแม่ค้าแซ่บ Episode 2 | 29 ก.ย. 67](http://i.ytimg.com/vi/r3lfLNH6Ke4/mqdefault.jpg)
I really appreciate that Walter Brown takes his craft seriously. He seems like a pleasantly professional guy.
That's a really nice tie
22:52 a clock, watching this because I don't need sleep, i need answers.
Almost everyone initializes their engine by a single integer from random generator. Mersenne twister has huge state and almost everybody initializes it with single integer, that gives you only 232 different sequences. Correct usage is to gather enough "true" random bits into vector or array and pass this data through seed_seq like interface into mersenne twister constructor.
Perhaps he meant 1 char?
The random-iterator-stuff is especially interesting!
CppCon video without subtitles?
Wow
Truly random
Walter Brown has great presentation skills. Among my favorites. There are many good presenters, but I think that him and Andrei Alexandrescu stand apart for different reasons. They are really good at exposing. Walter Brown's metaprogramming Compendium is impressively well explained too.
I'm guilty of "rolling my own" and getting it wrong - with RNG as well as other things. When nobody is relying on your bad code for anything important, it's OK ; you learn from your mistakes and move on. It's pretty nice to do less work and get a much better result.
For RNG I've been using Melissa O'Neil's "randutils" header, which builds on the excellent interfaces & facilities provided in to give an interface that's more convenient for most uses. One highlight is good seeding by default using a variety of local sources of entropy. std::random_device seems like it can help with seeding but it can't be relied on for that (not portably at least). I've heard that Melissa's work (or something much like it) may make it into the language spec.
Dr. Brown mentioned Professor O'Neil's work in his talk, and that her PCG random generator might make it into STL (not into the core language specification). She also co-authored a proposal titled "Simplifying simple uses of " (www.open-std.org/jtc1/sc22/wg21/docs/papers/2016/p0347r0.html)
Doesn't the C++ language spec include a specification of what's in the STL? I thought it was one big fat document.
Yes, it is indeed one big fat document, e.g. the working draft for C++17 here: open-std.org/JTC1/SC22/WG21/docs/papers/2016/n4606.pdf
In general, it seems to be common to distinct between "Core Language Specification" and "Library Specification" (sections 17-30 in said document). I figured you'd refer to the former.
It would be interesting to see how the interfaces of random could be adapted to make proper use of quasi random sampling (low discrepancy sequences). I'll like an interface where I could use either.
This is perfect. Exactly what I was searching for.
very clever man. One from not so many talks which I enjoyed.
excellent talk.
At 1:00:00. Lambda functions need ;'s and [] replaced by [=]