Robert, I totally agree with you on the use of a two functions interpolation. One to be used up to 1.5 the other one (that seems to me to be well approximated by a linear interpolation) for higher values.
That would lead to less "underestimation" and to a behavior that seems to be more realistic. Cheers, R. Inviato da iPhone di Eng. Riccardo Brama, Ph.D. Chief of Engineering @Dive Industries > Il giorno 10 ago 2019, alle ore 17:04, Robert Helling <hell...@atdotde.de> ha > scritto: > > Willem, > >> On 10. Aug 2019, at 16:10, Willem Ferguson <willemfergu...@zoology.up.ac.za> >> wrote: >> >> An interesting alternative, Robert. I am not happy with the deviation at 1.5 >> and 1.6. One would have to check what the effect of these two points are on >> the power curve. What is the effect on the overall fit of the power curve if >> one omits those two points? What of a 3rd order polynomial that could in >> principle accommodate the inflection at 1.4? I am not averse to a >> mathematical solution because the linear interpolation also causes some >> inaccuracy. >> >> > > here is the same on a log scale: > > <PastedGraphic-2.pdf> > > I would not be happy to fit this with a line for all points including the > last two. Rather, I would use a new line for the last three points (and > extrapolate that) for values above pO2=1.5bar. > > > Robert > _______________________________________________ > subsurface mailing list > subsurface@subsurface-divelog.org > http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface
_______________________________________________ subsurface mailing list subsurface@subsurface-divelog.org http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface