[daip] Re: POSSM with BPARM normalisation (fwd)

[Eric Greisen]

R. Niruj Mohan writes:
 > 
 > Hello Amy,
 > 
 > Many thanks for that -- got the updated version now.
 > I have another question (sorry!) ... 
 > when we apply smoothing in POSSM or BPASS, say regular
 > hanning for offline corrections ofr Gibbs ringing
 > (SMOOTH=1,0), what exactly does AIPS do to the end
 > channels ?  I am not getting very sensible results for
 > the end channels after smoothing, compared to before.

It does the convolution using the allowed channels (data are copied to
TEMP then smoothed back into VIS).  From $APLNOT/SMOSP.FOR

               DO 30 IFRQ = BCHAN,ECHAN
                  J1 = MAX (IFRQ - SUPRAD, BCHANS)
                  J2 = MIN (IFRQ + SUPRAD, ECHANS)
                  S = 0.0
                  W = 0.0
                  DO 20 J = J1,J2
                     IF (TEMP(J).NE.FBLANK) THEN
                        L = ABS(IFRQ-J) + 1
                        S = TEMP(J) * SMTAB(L) + S
                        W = SMTAB(L) + W
                        END IF
 20                  CONTINUE
                  IF (W.GT.0.0) THEN
                     VIS(INDX+I-1) = S / W
                  ELSE
                     VIS(INDX+I-1) = 0.0
                     VIS(INDX+2) = 0.0
                     END IF
                  INDX = INDX + INXINC
 30               CONTINUE

Usually the edge channels have both small signal and rather varying
phase so this may actually make the signal drop more

Eric Greisen