MathGroup Archive 1999

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

Search the Archive

Integration of Oscillating Function

  • To: mathgroup at smc.vnet.net
  • Subject: [mg18445] Integration of Oscillating Function
  • From: Christian Honeker <xian at mpip-mainz.mpg.de>
  • Date: Wed, 7 Jul 1999 00:11:31 -0400
  • Sender: owner-wri-mathgroup at wolfram.com

Mathematica Users!

I am attempting to integrate the following function:

function[x_] :=
Exp[-(u^0.82*Cos[Pi*0.82]+0.00525*u/x)]*Sin[u^0.82*Sin[Pi*0.82]]

in the follow manner:

integratedfunction[x_] := (0.00525/(Pi*x^2)) * NIntegrate[function[x_],
{u,0,Infinity}, WorkingPrecision -> 50, MaxRecursion -> 18, MinRecursion
-> 3, AccuracyGoal -> 10]

where u is the integrand and x varies between 0.0001 and 1.

Mathematica responds with a complaint about the WorkPrecision criteria
not being met,
but still finds a solution.
The solution, however, depends strongly on the integration limits (i.e.
it is unstable). For example,
integration of function[0.0001] should result in zero. A plot of
function[0.0001]
vs. u makes this clear. Depending on the integration range, however,
Mathematica
finds values of integratedfunction[0.0001] between 0 and 65. The
solution for values
of x > 0.006 are correct, however. This is because the oscillation of
function[x_]
decreases rapidly as x increases.

I am hoping that by making Mathematica aware that the function
oscillates strongly,
it can then perform the integration correctly. Setting the option
Methods -> Oscillatory
does not seem to work.

Can anyone help me?

Background:

I am performing photon correlation spectroscopy (PCS) on complex fluids.
Function[x_]
(above) has its origin in the inverse Laplace transformation of the
Kaulrausch-Watts-Williams
(KWW) function. The integration needs to be performed in order to
determine the
distribution of relaxation times which correlate with the size of my
micelles.
I find that the size distribution (integratedfunction[x_]) has the
correct form
for x > 0.006 (see above). I would like to determine the other half of
the
size distribution (x < 0.006), however.

Thank you very much for any tips that you can provide!

Michael Bockstaller
Max Planck Institute for Polymer Research



  • Prev by Date: Re: Re: O.D.E in Power Series
  • Next by Date: RE: Plot vs FindMinimum
  • Previous by thread: Re: Re: O.D.E in Power Series
  • Next by thread: Re: [Q] 2D graphic routines and data sets in 2D lists