On 30 nov, 01:23, wrote:
On Nov 28, 1:28 pm, Wimpie wrote:

Hello,

I made an excel sheet to simulate Fresnel diffraction patterns. I used
a 2 way data table to generate the "infinitesimal" contributions from
the EM field. After summation of the contributions in the 2D tables,
a have one result cell (that holds |E|) and three input cells
(variables) for x-, y-, and z- coordinate for the observation point.
When I change the coordinates manually, the results seem OK.

Does somebody know a solution for this (I do not prefer a VB one)?

Excel is a notoriously bad thing to use for statistics or numerical
analysis. Aside from the pain the rear nature once you get past
trivial manipulations, there are some serious calculation issues to
worry about (since Excel isn't rigorously tested for its
implementation, these also change as versions change)

http://www.burns-stat.com/pages/Tuto...addiction.html
covers a lot of the problems.

You're better off biting the bullet and using something like Matlab
(academic license $100) or Octave (free, but not quite as fancy as
Matlab, but mostly compatible) or similar products.

Hello,

Thanks for replying Richard and Jim. I hoped somebody had a similar
problem and knew a work-around. I know Excel is not the perfect tool
to do this, but I did similar things with reasonable results. I hoped
using data tables would speed up the spreadsheet design (until now it
didn't).

I do not have OO.org installed on this machine. A brother in law has
it on his PC, so when I visit him I will try to load the worksheet
into OO calc (or set it up in OO calc).

When this (OO calc) doesn't give the desired result, I have to change
to something that supports programming. I am not a student so Matlab
might not be option. I looked at "octave", "python" and "R". These
require some study also. Other option is to use an old general purpose
SW development package (C or Pascal based).

I think I have to spend some time to decide what to use...

Best regards,

Wim
PA3DJS
www.tetech.nl
removing abc results in valid PM

On Sun, 30 Nov 2008 14:24:28 -0800 (PST), Wimpie
wrote:

I know Excel is not the perfect tool
to do this, but I did similar things with reasonable results.

Hi Wim,

After following Jim's link, and reading it to considerable depth, I
wouldn't trust Excel to do anything more than 3rd grade arithmetic.
Even then, there are too many examples to suggest I still could get
burned. Take heed of their cautions that "reasonable results" are not
always accurate results.

Could the financial meltdown be attributable to rounding errors in
spreadsheets?

Make the investment and go with R.

73's
Richard Clark, KB7QHC

Wimpie writes:

Hello,

Thanks for replying Richard and Jim. I hoped somebody had a similar
problem and knew a work-around. I know Excel is not the perfect tool
to do this, but I did similar things with reasonable results. I hoped
using data tables would speed up the spreadsheet design (until now it
didn't).

I do not have OO.org installed on this machine. A brother in law has
it on his PC, so when I visit him I will try to load the worksheet
into OO calc (or set it up in OO calc).

When this (OO calc) doesn't give the desired result, I have to change
to something that supports programming. I am not a student so Matlab
might not be option. I looked at "octave", "python" and "R". These
require some study also. Other option is to use an old general purpose
SW development package (C or Pascal based).

I think I have to spend some time to decide what to use...

You can write plug-ins for Gnumeric in Python. However, it's best to
run Gnumeric on Linux (or another Unix of your choice). There is
Windows version, but unfortunately, it is somewhat unstable.

Jon LA4RT