How to resolve floating point zero error in the following code

for i=1:3
    for j=1:3
        for k=1:z
            p=Q(i,j,k)*(Z(k+1)-Z(k));
            q=Q(i,j,k)*(Z2(k+1)-Z2(k));
            r=Q(i,j,k)*((Z(k+1))^3-(Z(k))^3);
            A(i,j)= A(i,j)+p;
            B(i,j)= B(i,j)+q;
            D(i,j)= D(i,j)+r;
        end
    end
end

[Walter Roberson]

What floating point zero error?

What inputs should we give when we run the code?

[Manish]

2

1/1000

0

0

0

0

0

[Walter Roberson]

Okay, so what should we be paying attention to?

Your ABD matrix is nearly singular, with rcond about 1e-9 . inv() of it is not going to be all that accurate. You should strive to avoid using inv(), preferring to use \

abd = inv(ABD); R1 = abd*[Nx;Ny;Nxy;Mx;My;Mxy];

would be better as

R1 = ABD \ [Nx;Ny;Nxy;Mx;My;Mxy];

[Manish]

okay i will do that but;

in ABD matrix ABD(1,3) = ABD(2,3) = ABD(3,1) = ABD(3,2) = 0

and i ain't getting that

[Walter Roberson]

https://matlab.wikia.com/wiki/FAQ#Why_is_0.3_-_0.2_-_0.1_.28or_similar.29_not_equal_to_zero.3F

You are suffering from round-off error. It is normal.

[Manish]

There is no way to to resolve it ?

Because i have to display the ABD matrix and

if i use A(A~= 0) = 0;

whole A matrix is resulting in zero matrix although every element is not approximately zero.