Services & Resources / Wolfram Forums
-----
 /
MathGroup Archive
2006
*January
*February
*March
*April
*May
*June
*July
*August
*September
*October
*November
*December
*Archive Index
*Ask about this page
*Print this page
*Give us feedback
*Sign up for the Wolfram Insider

MathGroup Archive 2006

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Precision of arguments to FunctionInterpolation

  • To: mathgroup at smc.vnet.net
  • Subject: [mg68514] Re: Precision of arguments to FunctionInterpolation
  • From: "Andrew Moylan" <andrew.j.moylan at gmail.com>
  • Date: Wed, 9 Aug 2006 04:17:57 -0400 (EDT)
  • References: <eakeg1$r1j$1@smc.vnet.net><easjk5$fti$1@smc.vnet.net> <eauv4s$15k$1@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

Peter Pein wrote:
> Andrew Moylan schrieb:
> > Here is some further information that makes the strangeness more
> > obvious:
> >
> > The following three different calls to FunctionInterpolation all fail:
> >
> >  FunctionInterpolation[lowPrecisionSin[x], {x, 0, 1.`14}]
> >  FunctionInterpolation[lowPrecisionSin[x], {x, 0, 1.`20}]
> >  FunctionInterpolation[lowPrecisionSin[x], {x, 0, 1}]
>
> the latter works in Mathematica 5.1 for Windows

You're right Peter, this last one works fine in Mathematica 5.2 for
Windows too. I don't know now why I thought it didn't.


> > Secondly, does anyone know the meaning of the InterpolationPrecision
> > option to FunctionInterpolation? In particular, with the default
> > setting of InterpolationPrecision->Automatic, what value is
> > automatically chosen?
> >
> You can test this using
>
> Reap[FunctionInterpolation[((Sow[Precision[#1]]; #1) & )[lowPrecisionSin[x]], {x, 0., 1}]][[2, 1]]
>
> which gives:
>
> {5.1924023244417254, Infinity, Infinity, 4.999999999999999, 5.000000000000001, 5., 5., 5.,
>    5.000000000000001, 4.999999999999999, 5., 5.000000000000001, 5., 5., 4.999999999999999, 5.,
>    5., 4.999999999999999, 5.000000000000001, 5.000000000000001, 5.000000000000001, 5., 5.}
>
> If you know the maximum precision of the function you want to interpolate (5 in this case), you can use
> InterpolationPrecision->5:
>
> intfuns = {f14, f20, finf} = (FunctionInterpolation[lowPrecisionSin[x],
>         {x, 0, SetPrecision[1., #1]}, InterpolationPrecision -> 5] & ) /@ {14, 20, Infinity};
>
> SameQ @@ intfuns
> --> True
>
> Through[{f14, lowPrecisionSin, Sin}[0.6`14.000000000000002]]
> --> {0.5646420701356566528`4.127871245474387, 0.5646424733950353572`5.000000000000002, 0.5646424733950353572009454457`14.056991706843485}

Thanks Peter, this sheds some light on the meaning of the
InterpolationPrecision option and how to choose it.


I think that the default value for InterpolationPrecision is
MachinePrecision, because the following succeeds with the same result
as when using "InterpolationPrecision -> Automatic":

FunctionInterpolation[lowPrecisionSin[x], {x, 0, 1},
InterpolationPrecision -> MachinePrecision]

whereas any explicit numerical setting above about 13 for
InterpolationPrecision fails. But again, when the upper bound of the
region of interpolation is changed from 1 to the machine-precision
number "1.", the interpolation suceeds for _any_ numerical value of the
InterpolationPrecision option less than about 23, which is quite
strange.


  • Prev by Date: Re: Newbie question about column sums of arrays
  • Next by Date: Re: Can we prove that it is never a square?
  • Previous by thread: Re: Precision of arguments to FunctionInterpolation
  • Next by thread: Re: Trouble with Module in a package