Re: Another fsolve question -- Beginner

[Przemek Klosowski]

On 05/26/2011 01:04 PM, MMolloy wrote:
> Hi all,
>
> I have had a look and can't seem to find a post which answers my question -
> apologies if there is one. My question is probably pretty simple but I'm
> having a hard time of it because I'm new to Octave - especially functions in
> Octave.
>
> I'm running;
> Ubuntu Lucid x64
> Octave 3.2.3 for x86_64-pc-linux-gnu.
>
> The problem:
>
> function y = f (x)
> y = zeros(10,1)
> load E1.csv;
> for i=1:10
> y(i) = (1.14.*exp(-x(i)./1.14)) .- (3.79.*exp(-x(i)./3.79)) .- E1(i)
> endfor
> endfunction
> load RG1.csv
> [x, fval, info] = fsolve(@f, RG1);

OK, but consider that the solver executes your function repeatedly, so 
the file E1.csv is being re-read continuously, which has got to hurt the 
execution time. Read it once into a global variable, and use 'global E1' 
within the function f() and in the main function.

Also, end your lines in semicolons, because otherwise you get printouts 
of computed values from those lines.

Also, you're working too hard: your 1:10 loop with element-by-element 
operations should use Octave's vectorized notation,
 y=(0.1.*exp(-x./0.1)) - (0.33.*exp(-x/0.33)) - E;

> My problem is that the matrices for B and E (which I compute elsewhere) have
> a size of [10000,1]. So when I use this;

I think that maybe all this re-reading is leaking some temporary 
storage. Please try rewriting the code and report back if it helped; if 
it didn't, please provide the data so that we can run your code for 
ourselves to see what is happening.