In numerical linear algebra, the effects of an inexact arithmetic on the stability of linear algebra operations are studied. As soon as computations are performed with finite precision, e.g., based on floating point numbers, some fundamental properties of basic mathematical operations are lost, for example, associativity. To guarantee that algorithms for signal processing problems behave as expected, these effects must be considered.

Numerical linear algebra is needed in the following fields of research at our institute:

• Multi-User MIMO Communications
• Sparse Signal Processing

To learn more about numerical linear algebra we kindly refer to the following lectures offered at our institute:

• Mathematische Methoden der Signalverarbeitung
• Numerical Linear Algebra for Signal Processing