Absolute Zero...

[NickPaul]

Hey all.

I'm doing some work on my linear algebra package. I'm actually writing a function to factorise a square matrix 'A' into LU components. Anyway, that isn't important. Here is the U component I'm looking at:

U =
765 2938 7
-1.37668e-014 -1014.87 -2.58954
-1.95463e-016 1.45717e-016 0.927951

Now those values which are incredibly small are meant to be zero. I'm working with doubles and they are all initialised to zero.

Some relevant code:

lead = U(i,j);
L(i,j) = lead/U(rowCount,j);
...
U(i,k) = U(i,k) - L(i,j)*U(rowCount,k);

Do you think this is due to the division on line 2? I'm not really sure what is going on. MATLAB returns zeros.

Any thoughts?

Nick.

[Bazzy]

Doubles use floating point representation, which approximates their values. A double will rarely have an exact value.

[NickPaul]

Thanks for that Bazzy!