MathGroup Archive 2006

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

Search the Archive

Re: Performance improvement needed - Help.

  • To: mathgroup at smc.vnet.net
  • Subject: [mg63575] Re: [mg63552] Performance improvement needed - Help.
  • From: Pratik Desai <pdesai1 at umbc.edu>
  • Date: Fri, 6 Jan 2006 05:24:19 -0500 (EST)
  • References: <200601050812.DAA19819@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

Tonybony wrote:

>Hello,
>I am trying to improve the performance of a bottle neck operation. I've
>used functional programming and read about Packed Arrays. My current
>operation is:
>
>Efunc[k_, x_, y_, z_] := Radialf[k, VNmax, x, y,z].Map[(MMatrix[[# -
>Nmin + 1]].SphericalHarmonicY[#, Range[-#, #],Theta[x, y, z], Phi[x,
>y]]) &, VNmax, {1}];
>
>EMatrix= Apply[Efunc[k, #1, #2, 0.0] &, Grid, {2}];
>
>Where I defined,
>GridX = Range[Xini, Xfin, Xinc];
>Grid = Outer[List, GridX, GridX];
>VNmax = Range[Nmin, Nmax];
>GridG = Table[{n, m}, {n, Nmin, Nmax}, {m, -n, n}];
>Radial[k_, n_, x_, y_, z_] := BesselJ[n,k Sqrt[x^2+y^2+z^2]];
>MMatrix= Apply[(Exp[-#2 fio] MAmpl[[#1 - Nmin + 1, Abs[#2] + 1]]
>MI1[[#1 - Nmin + 1, Abs[#2] + 1]] Cosfio) &, GridG, {2}];
>
>I have tried the command,
>GridG = Developer`ToPackedArray[Table[{n, m}, {n, Nmin, Nmax}, {m, -n,
>n}]];
>To see if I could build a packed array so that I could better
>performance, but didn't succeed. Maybe I am creating a wrong array, but
>I don't know of any alternative.
>  
>
I think that might be the problem, based on the documentation I have 
seen PackedArray seems to work only for rectangular tensors (whatever 
that means)
http://documents.wolfram.com/mathematica/Built-inFunctions/AdvancedDocumentation/LinearAlgebra/LinearAlgebraInMathematica/Performance/PackedArrays/AdvancedDocumentationLinearAlgebra5.0.0.html
When I tried the array provided by you in GridG, I get the following
Clear[GridG, data]
Nmin = 0;
Nmax = 10;
data1 = Table[{n, m}, {n, Nmin, Nmax}, {m, -n, n}];
data1 // MatrixQ;
 >>False
Hence PackedArray does not seem to work

Hope  this helps

Praik

>For each element of GridG I have a pair of point (x,y) that I pass
>along a set of function for calculations. Is their an efficient way of
>doing this?
>
>Thanks,
>Antonio
>
>  
>


  • Prev by Date: Re: Best linear Fit to slope data with Fixed starting point/value.
  • Next by Date: Re: Kernel hang during FindFit under 5.1 for Linux
  • Previous by thread: Performance improvement needed - Help.
  • Next by thread: Re: Performance improvement needed - Help.