Frederic Tessier

Looking back through the log history (in rcs and sccs version control logs!), and I can confirm that the lines containing the `20.863` constant in the `fix_brems` subroutine first appeared...

(see discussion also in #292)

Hmm, that's not negligible. I wonder if we make the low-res version the default, with a prominent _notice_ to turn the hi-res option when "needed" (with a clearer definition), or...

Wow, thanks for that perspective; perhaps `pcg64` is the better contender. Will have to do some more reading on this entertaining drama! Then again, if the code is in the...

Darn! Skip all this and implement xoshiro256+ right away? 🤓

Posted in the thread again. Here are my replies regarding the pcg vs xoshiro debate: Interesting and entertaining feud between the xoshiro and pcg generator authors! From there I was...

On pcg vs xoshiro: after additional reading, I side with the `xoshiro` generator, for its dead simple code, its speed, and its ability to jump ahead to generate independent random...

Only thing left as an implementation detail is how to sample doubles uniformly on [0,1). Do we just take the high bits from the xoshiro256+ 64-bit integer result and plug...

Using the 53 most significant bits as the mantissa is sufficient for our purpose in EGSnrc, although strictly speaking it is not uniform: values in (0, 2e-53) are never picked....

I have not looked at the `ranlux` prng recently, because they are (were) painfully slow generators. But an implementation is useful to eventually check other generators. Thanks for the reference,...