Date:         Tue, 15 May 2007 23:38:56 -0400
Reply-To:     Richard Ristow <wrristow@mindspring.com>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Richard Ristow <wrristow@mindspring.com>
Subject:      Re: Paired compute looping help
Comments: To: rwwalker@optonline.net
In-Reply-To:  <e31ecc921248d.464a70d1@optonline.net>
Content-Type: text/plain; charset="us-ascii"; format=flowed

At 10:47 PM 5/15/2007, Bob Walker wrote:

[test data omitted; see cited posting]

>Here's one approach:
>
>DEFINE !BUBBA()
>!LET !MIN=1 !LET !MAX=5
>VECTOR V=!CONCAT("V",!MIN) TO !CONCAT("V",!MAX).
>
>!DO !A=!MIN !TO !MAX !DO !B=!A !TO !MAX
>!IF (!A<>!B) !THEN
>COMPUTE !CONCAT("V_",!A,!B)=V(!A)+V(!B).
>!IFEND !DOEND !DOEND
>EXECUTE.
>DATASET NAME BUBBA WINDOW=FRONT.
>
>!ENDDEFINE.

Now, that is a SWEET little piece of macro writing.

Curious: since you're unrolling the COMPUTEs in a macro loop anyway,
why not generate the arguments to the COMPUTEs by catenation, rather
than vector indexing? Like this (WRR-not saved separately):

[test data]

DEFINE !BUBBA()
!LET !MIN=1 !LET !MAX=5

!DO !A=!MIN !TO !MAX !DO !B=!A !TO !MAX
!IF (!A<>!B) !THEN
COMPUTE !CONCAT("V_",!A,!B)=!CONCAT("V",!A) + !CONCAT("V",!B).
!IFEND !DOEND !DOEND
!ENDDEFINE.

PRESERVE.
SET MPRINT ON.
!BUBBA.
  144 M>
  145 M>  .
  146 M>  COMPUTE V_12 = V1 + V2.
  147 M>  COMPUTE V_13 = V1 + V3.
  148 M>  COMPUTE V_14 = V1 + V4.
  149 M>  COMPUTE V_15 = V1 + V5.
  150 M>  COMPUTE V_23 = V2 + V3.
  151 M>  COMPUTE V_24 = V2 + V4.
  152 M>  COMPUTE V_25 = V2 + V5.
  153 M>  COMPUTE V_34 = V3 + V4.
  154 M>  COMPUTE V_35 = V3 + V5.
  155 M>  COMPUTE V_45 = V4 + V5.
  156 M>  .
RESTORE.
  157 M>  RESTORE.

DATASET NAME BUBBA WINDOW=FRONT.

LIST.
[data listing]