to solve an equations system

[Guest molesqualeux]

Hi , I need your help .

I search on the web an algorithm that can solve an n equations system at n unknowns .

 

Thanks a lot!

[matt grime]

Can you do it for two unknowns? Then you can do it for n unknowns. Or look up Gaussian elimination.

[fuhrerkeebs]

Or, if you have a range where you think the solution will be, you could just approximate numerically with a do/until loop. Or, if you don't have a range, you could use determinants (may be a little hard if you want a programmable algorithm...though I've never actually tried to program it). Or, you could just use matrices (the famous AX=B ).

[bloodhound]

gaussian elimination is the way to go. you could use cramer's rule. but as n gets big it becomes pointless

[Dave]

Gaussian elimination is definately the best method.

[Guest molesqualeux]

thanks , I hope I 'll find the algorithm on Internet.

I know how to use it manually but in order to program it , it's very difficult.

 

Above all , it must write the algorithm and then program it with pascal for example

[Perennial]

This one has at least somewhat understandable structure :

 

http://math.fullerton.edu/mathews/n2003/Web/GaussianJordanMod/GaussianJordanMod.html

[Guest molesqualeux]

this is a part of the result of my pascal program

 

procedure pivot(var mat1:matrice);

begin

 

 

 

for k:=1 to DIM_MAX-1 do

 

 

 

 

begin

 

for i:=k to DIM_MAX-1 do

 

begin

 

for j:=k to DIM_MAX-1 do

begin

mat1[i+1,1+j]:= mat1[i+1,j+1] * mat1[k,k] - mat1[i+1,k]* mat1[k,j+1] ;

mat1[i+1,k]:=0;

end;

 

 

 

end;

end;

 

 

end;