Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.



Re: Dirac in Convolution
Posted:
Mar 13, 2013 9:19 AM


On 10.03.13 20:58, omid najafi wrote: > in this program >  > clear;clc > dx=0.01; > x=2:dx:2; > f1=heaviside(x+1)heaviside(x1); > f2=dirac(x1); > f3=dx*conv2(f1,f2); > xc=4:dx:4; > subplot(3,1,1);plot(x,f1) > subplot(3,1,2);plot(x,f2) > subplot(3,1,3);plot(xc,f3) >  > we know > int(dirac(xa)*f(x),inf, inf) = convolution(dirac(xa),f(x)) = f(xa)
conv2 doesn't know that. It assumes you mean a discrete convolution (where using dirac doesn't really make sense, unless you want a distribution back). By the time f2 is defined, your dirac has already gone numeric (i.e., it is 0 or Inf), and any such connection is lost. Write the integral form instead, with a symbolic x.
HTH, Christopher



