[PATCH 2/2] gas model: replace Redlich-Kwong with least-square quintic

[Lubomir I. Ivanov]

On 3 March 2016 at 00:28, Linus Torvalds <torvalds at linux-foundation.org> wrote:
> Lubomir, what tool did you use for the least-square fitting?
>

those equations in particular came from this website:
http://polynomialregression.drque.net/online.php

> Because it would be interesting to see if we can do this for the table at
>
>   http://www.baue.org/library/zfactor_table.php

i can take a look a these a bit later - e.g. in 30 minutes.

>
> and see how reasonable it is to approximate gas mixes by doing the
> O2/N2/He calculations and then just "mixing" the compressibility (looking
> at the O2 and N2 tables and then the air table, I think a stupid and
> straightforward linear mixing is probably _entirely_ unphysical, but not
> necessarily too far off).
>
> For example, looking at 3000 psi, and then just doing
>
>   0.947*0.21 + 1.054*0.79
>
> to do a linear mixing of the O2/N2 values gives you 1.0315, which is quite
> close to the actual air compressibility value.
>
> So I think it might be possible to take those baue.org tables, create
> polynomials for them, and then have a simple linear mix for the gases, and
> get reaonable approximations for all relevant scuba gases..
>

i don't exactly understand the physics here, but in terms of "plotting
a data set" and looking at how the graphs / isotherms variate, the
linear mix might be a tolerable solution.

lubomir
--