Mo logo [home] [lexicon] [problems] [tests] [courses] [auxiliaries] [notes] [staff] german flag
Mathematics-Online course: Linear Algebra - Linear Systems of Equations - Approximation Problems

Normal Equation

[previous page] [next page] [table of contents][page overview]

For an $ m \times n$ matrix $ A$ , any solution $ x$ of the least squares problem $ \Vert Ax-b\Vert _2$ satisfies the normal equations

$\displaystyle A^{\operatorname t}Ax = A^{\operatorname t}b. $

\includegraphics[width=.4\moimagesize]{a_normalengleichungen}

Geometrically, this means that the residuum $ Ax-b$ is orthogonal to the columns of $ A$ , i.e. to the subspace $ \operatorname{im} A$ of $ \mathbb{R}^m$ . The solution is unique if rank $ A = n \leq m$ .

(temporary unavailable)

(temporary unavailable)
  automatically generated 4/21/2005