** **

**87. Implementation of an FFT with multiplicative
complexity below Heideman-Burrus bound.**

[Implementação de uma FFT com complexidade multiplicativa abaixo do limitante de Heideman-Burrus]

The computational implementation of a new fast
algorithm for computing DFT based on a matrix Laurent series [1] is presented
via SimulinkTM.

The arithmetic complexity,
expressed by the number of nontrivial real floating-point multiplications,
achieves values below the standard Heideman-Burrus
bound [2].

The case N=16 is presented
in details, requiring merely 12 real multiplications and 101 additions.