An overdetermined system is a system of equations containing more equations than unknowns.
In the linear case it has the form , where
is an
matrix with
.
An overdetermined system need not be inconsistent. For example, and
have the common solution
, whereas
and
have none. In general, a linear overdetermined system has
an exact solution iff
belongs to the column
space of
.
When an exact solution does not exist, the least
squares fitting problem minimizes the squared Euclidean
norm
If the columns of are linearly independent,
the solution obtained by least squares fitting
is unique and satisfies the normal equations.
A QR decomposition provides a method for finding
this solution without explicitly forming
in the real case.