getting NaN as result

Dear MATLAB helpers,

I'm trying to run a project for 2 month but I think the problem is bad scaling.

this is the code :

K=zeros(15);
for i=1:4
    Ke=bar3e(Ex(i,:),Ey(i,:),Ez(i,:),[E Ep(i,1)]);
    K=assem(Edof(i,:),K,Ke);
end

Also I attached bar3e function, assem and variables value such as E , it may help.

I will be thankful for any suggestion.

All the best

[per isakson]

Which are the steps needed to reproduce your problem?

    >> [Ke]=bar3e(ex,ey,ez,ep)
    Undefined function or variable 'ex'. 
    >> [K,f]=assem(edof,K,Ke,f,fe)
    Undefined function or variable 'edof'.

[Hessam]

The above code is a part of my project, all the functions or variables are defined.

my whole project is attached.

I get warning Matrix is singular, close to singular or badly scaled every time and this is because of getting NaN as result for K, Ke.

thank you

[Fabio Freschi]

Unfortunately I can't do some tests myself. Anyway, I have an idea. My guess is that you are trying to solve a FEM problem or similar and you are not applying constraints (or boundary conditions) to the stiffness matrix. To check if I am right, remove one row and one column from the stiffness matrix and rhs and solve the problem. This is equivalent to set an unknown to 0. For example:

% this is the unknown we want to fix to 0
ifix = 1;
% tis is not very efficient but simple!
K(ifix,:) = [];  % remove row
K(:,ifix) = [];  % remove col
f(ifix) = [];    % rhs

% solve. The solution might be meaningless, but the system is now
% constrained and you should not have singularities anymore
x = K\f;

If this works you can try to apply meaningful boundary conditions to your system.

Let us know!

Fabio

[Hessam]

Thank you Fabio, I like your idea.

I did what you said but it leads to this error :

Index exceeds matrix dimensions.

Error in extract
        ed(i,1:(n-1))=a(t(i,:))';

Error in FEM2 
Ed=extract(Edof,a);

What should I do now??

I'm so confused, I couldn't solve this warning for a long time.

Thanks.