Re: numerical integration problem

*To*: mathgroup at smc.vnet.net*Subject*: [mg77868] Re: numerical integration problem*From*: dh <dh at metrohm.ch>*Date*: Mon, 18 Jun 2007 07:09:09 -0400 (EDT)*References*: <f532hu$36j$1@smc.vnet.net>

Hi Sean, you could e.g. numerically approximate your function x[w] using FunctionInterpolation or even Interpolation. Afterwards, the integration should be easy. hope this helps, Daniel Sean wrote: > Hi, > I know that to post a technical question here, it is best to give an example. > But my program is huge but straightforward, so I hope to make clear my problem by simply describing how I design the program. > > My problem at hand is that I have a list of linearly coupled equations whose coefficients are normal functions of some parameters. > for a simple example > > x*f[w]+y*g[w]=z[w]; > x*b[w]+y*m[w]=n[w]; > > I need to solve for x and y. Of course, the naive way is to put these equations into a matrix and solve for x and y----remember my problem has a 20 X 20 matrix. After this, x and y are expressed in terms of these functions. I just defind them as > > x[w_]:=...; > y[w_]:=...; > Now I need to do a numerical integration over these functions x[w], and y[w].. But the trouble comes. Because of these huge expressions for the target functions x[w] and y[w], my dram is immediately eaten up and mathematica automatically terminates the program... > I read in a previous post > that suggests if I redefine > x[w_?NumerQ];= ... ; > y[w_?NumerQ]:=...; > > Would this help, or there is a smarter way of fomulating my problem so that it would run faster and use less memory. > > Your help is greatly appreciated, > > Sean >