Re: Convolution
- To: mathgroup@smc.vnet.net
- Subject: [mg10938] Re: Convolution
- From: calvitti@lychee.ces.cwru.edu
- Date: Sat, 14 Feb 1998 00:53:13 -0500
- In-Reply-To: Larry Gottlob's message of 12 Feb 1998 06:37:02 -0500
- Organization: Case Western Reserve University
- References: <6bumsu$l93@smc.vnet.net>
>>>>> "L" == Larry Gottlob <lrg@geri.duke.edu> writes:
L> I'm not experienced at Mathematica. I define two functions:
L> Func1[x_] := tau*Exp(-x) Func2[x_] := mu*x^2
L> for instance, where tau and mu are constants.
L> How do I define a function of x that is a convolution of Func1
L> and Func2, and plot that?
you're probably going to have trouble taking the convolution of those
two functions since their integrals diverge. convolution of f and g is
defined as:
convolution[f_,g_,x_] := Integrate[f(y)*g(x-y),y]
you can plot convolution[Func1,Func2,x] like any other function.
alternatively,
1) convolution is also the inverse fourier transform of the product of
the fourier transforms of the f and g
2) use the discretized approximation Sum[f[x]*g[x],{x,-M,M}] where M is
a large number. if you use this method, it's probably much faster to
build up a list (with Table) to represent the convolution over a
specific interval at a specific gridsize...then use ListPlot to plot
it:
discreteConvolution[f_,g_,x_,x0_,x1_,dx_,M_] :=
Table[Sum[f[x]*g[y-x],{x,-M,M}],{y,x0,x1,dx}];
+---------------------------------+
| Alan Calvitti |
| Systems and Control Engineering |
| Autonomous Robotics Group |
| Case Western Reserve University |
+---------------------------------+