q: Smith book code segment 3090 ch 26

Neural Networks,ch 26,, p. 479 online 
( yes, the hardback has already been ordered)

In the following code,
I can't seem to follow how the convolution is calculated.

I'm pretty sure I follow the basic technique in ch6-7, 
but perhaps the use of the variables and the complex format
make it difficult.  I certainly see an empty REX and IMX, except 
for the "impulse" inserted at IMX[12].. I'm sure part of my confusion 
is that I'm not following how/why the vars are used.

some additional explanation would be appreciated...



3080 IMX[12] = 1
3090 '                                      'CALCULATE IMPULSE RESPONSE
3100 FOR I% = 12 TO N%-1
3110 FOR J% = 0 TO NP%
3120 REX[I%] = REX[I%] + A[J%] * IMX[I%-J%] + B[J%] * REX[I%-J%]
3130 NEXT J%
3140 NEXT I%


thx,