Error 1000 in Datafit Function

[Dylan Flint]

Hi. Firstly, I am quite new to Scilab. I used to and still do use VBA
for most of my programming, but I've learnt that it just doesn't cut
it any more.

My problem is that I get an 'Error 1000' coming up when I use the
datafit function. The message that comes with it says "The weighting
matrix must be square" and the error is on line 82 of 'datafit'. If I
could get into the datafit *.bin file then I could see how it works,
but I'm not sure how to open *.bin files. That is the one way
somebody could help me. Otherwise, if you know about Error 1000 and
the weighting matrix, then some info on that directly would be of
great help.

regards
Dylan

[Enrico Segre]

On May 20, 4:44 pm, Dylan Flint <dylan.fl...@gmail.com> wrote:
> If I
> could get into the datafit *.bin file then I could see how it works,
> but I'm not sure how to open *.bin files.

you don't need (except for hacking purposes, such as to hunt bugs of
the scilab compiler) to look in the compiled *.bin file but rather in
its source *.sci. For quick access: try entering datafit under File/
Open source of... in the Scipad editor, or equivalently, issue "edit
datafit scipad" at the shell prompt.

> Otherwise, if you know about Error 1000 and
> the weighting matrix,

nope, but OTOH you didn't provide details on how the error arose
(better so -- I personally can't stand submissions of hundreds of
lines of irrelevant code, and like way better your self help
attitude...;).

hth, Enrico

[ycollet]

Hello,

Normally, when you perform an installation of scilab, scilab installs
datafit.sci and datafit.bin files (try find /usr/lib/scilab-* -name
datafit.sci under linux).

Can you send an exemple which shows how you use datafit ?

YC

[Dylan]

Thanks YC and Enrico.

I had another problem, but I thought I better not post it because I'm
sure I'll figure it out. The datafit problem I was sure was genuine.
It turns out that the initial guess vector should have been a column
vector and not a row vector... hence the error, "The weighting matrix
is not square".

Anyway, now that it is working, I'm not getting the same results as
somebody that used the Nelder mead algoritm. But before I ask for
help again, I'll spend a bit of time trying to sort it out myself. I
got hold of a Scilab textbook so I'll have a look in there.

Hopefully I'll sort it out, but I may post again here. Actually I'll
post in either case, just to let you know if I do get it to work.

regards
Dylan

[Dylan]

Finally! It works. For the most part programming is tricky and
irritating, but when it does all work out, it really does feel good.