WOLFRAM

Services & Resources / Wolfram Forums / MathGroup Archive

MathGroup Archive 2007

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

Search the Archive

Re: Quick integral.

  • To: mathgroup at smc.vnet.net
  • Subject: [mg73491] Re: Quick integral.
  • From: dontdont at gmail.com
  • Date: Tue, 20 Feb 2007 06:15:15 -0500 (EST)
  • References: <erc23r$mv4$1@smc.vnet.net>
On Feb 19, 3:36=C2=A0am, "Jouvenot, Fabrice" <F.Jouve... at liverpool.ac.uk>
wrote:
> Thanks for your answers.
>
> As some of you asked for the notebook here is it :http://www.onyrium.net/=
Mathematica/CoordScat.nb
>
> At the end there is an integral named toto[], we launch it and have an
> evaluation time of how this implementation in a large notebook will
> affect time.
>
> We want to minimize a really lot this time. We try to make more physics
> approximations, but there is certainly mathematica things we can do to
> quicker everythings.
>
> One more things to know : we do not need a large accuracy on the
> results, large errors are ok.
>
> Thanks for your help.
>
> Fabrice.

The integrand to NIntegrate seems far more complicated than the amount of
accuracy that you are asking for.  This is probably why it is so slow to
integrate.

There seem to be problems with the notebook you posted that make it
unreliable to try to evaluate.
<< Graphics`Graphics`
seems to need a semicolon at the end and there seem to be other
things that make it evaluate and produce no output, for example if it
is evaluated, changed in any tiny way, saved, and then tried to
evauate again.

toto[l_] := Sum[dNd=CF=95[l, angle=CF=95[=CE=A8, thc, =CE=B8source], thc]*
   AcceptancePMT[angleLightOM[vOM,
   vlight[thc, =CE=A8, =CE=B8source, l]]], {=CE=A8, =CF=80/2,2=CF=80,=CF=80=
}];

is only slightly faster, because of the very complicated expression
to be evaluated, but produces approximately the same result as
NIntegrate evaluating across the entire interval.

dontdont at gmail.com
  • Prev by Date: GUIKit tool to select frames
  • Next by Date: Limit of Error function Erf
  • Previous by thread: Re: Quick integral.
  • Next by thread: RE: Re: Quick integral.