Prev: named common and subprogram names
Next: Why it is forbidden?
From: monir on 22 Dec 2009 11:16
On Dec 22, 5:53 am, "robin" <robi...(a)bigpond.com> wrote:
> e p chandler wrote in message ...
>
> >"monir" <mon...(a)mondenet.com> wrote in message
> >news:4c037cfc-2a89-44a4-93c5-e6a0c57b4974(a)g23g2000vbr.googlegroups.com....
> >On Dec 20, 6:42 pm, "e p chandler" <e...(a)juno.com> wrote:
>
> >---> Yes, you can get away with DIMENSIONING the 1D arrays in your
> >subroutines to (1). Yes you can pass a 1D array that is actually longer than
> >that declared.
>

> Robin wrote:
> That was a mid-1960s hack for adjustable dimensions - long since
> obsolete.
> Monir was not talking about that.
>
> These issues were discussed at length and explained in the post
> "Could this result in array distortion"
> that you inititiated.

Robin;

1) In the earlier post: "Can this result in array distortion ??", the
question was regarding passing multidimensional arrays with diffirent
sizes than those declared in the calling routine.
Here the question is regarding using argument list integers to dim
local multidimensional arrays.
NOT the same!

2) It appears at this point and after extensive numerical
experimentation that the problem might actually be associated with
some kind of "numerical instability" rather than a code issue. The
reason for this tentative conclusion is that my algorithm (F77) works
fine under "certain conditions", which I'm trying to find out.

3) For example, if I pass fictitious data values for the 3D array
"ya" (same dimensions, and everything else remain UNCHANGED) to the
main routine:

! ************************************************************
SUBROUTINE Splint3d(x1a, x2a, x3a,
1 ya, maxPln,maxR,maxTH,
2 k, m, n,
3 x1, x2, x3, Z)
! ************************************************************
! USES: Splie2d , Splin2d , Splined , Splintd

the program works perfectly, producing the correct results (up to
11-12 decimal figures), while using the actual analytical data for
"ya" (with no singularities or jump discontinuities), it produces
rubbish!
Keep in mind that the only thing that's changing here is the values of
the 3D array "ya".
(..the above 4 arrays: x1a(1D),x2a(1D),x3a(1D),ya(3D) are dim exactly
as in the calling routine, as pointed out in my post Dec 20, 4:46 pm))

Earlier, JB wrote:
> After all, programming is supposed to be a painful exercise in frustration.

I entirely agree!!

Regards.
Monir
First  |  Prev  | 
Pages: 1 2 3
Prev: named common and subprogram names
Next: Why it is forbidden?