CNS calculation headaches

Linus Torvalds torvalds at linux-foundation.org
Sat Aug 10 11:46:57 PDT 2019


On Sat, Aug 10, 2019 at 11:30 AM Linus Torvalds
<torvalds at linux-foundation.org> wrote:
>
> I'm attaching my R script so that people can try it themselves,

Side note: unlike Dirk, I'm not actually a math major. Yes, I did math
as a minor, but most of that was discrete math and set theory. And my
R skills are "I can google and cut-and-paste things". I don't use R in
any normal situation.

This is just the long version of "caveat emptor".  The script may be
garbage. The whole approach is dubious at best and has no physicality
to explain using a polynomial for something like this. Using a power
function makes a whole lot more sense from a physics standpoint, I
suspect, but from a data fit standpoint I just think it's horrendous.

And a *sane* function would obviously give a "this is infinitely safe"
for a pO2 of 0.21. The polynomial does not do that, although it does
give a long time (but "many hours" is not the "you can live on that
level indefinitely" answer). Again, the fourth-order one gives better
results even at the pO2 of 0.21 level. But again, a fourth-order
function makes _zero_ physical sense.

It just looks like a fairly good curve in that "0.2 < pO2 <2.0" range.

But there might be other much saner models.

                   Linus


More information about the subsurface mailing list