Open up any linear algebra textbook and have a look at the matrix
entries. Are there all integers? Are they all written as decimals? There's
probably at least some that are fractions. Matrices in computer programs are
almost always in decimal form. The exception is symbolic mathematics programs
like Maple and Mathematica. That's because computers store non-whole numbers as
floating-point values instead.
Floating-points are fine most of the time, but they're often
not exact. What happens if we work the fractions directly?