Next: Computing primitive n mathend000#-th roots
Up: Fast Polynomial Multiplication based on
Previous: The Fast Fourier Transform
Algorithm 2
mathend000#
Theorem 2
Let n
mathend000# be a power of 2
mathend000# and
R
mathend000# a primitive n
mathend000#-th
root of unity.
Then convolution in
R[x]/
xn-1
mathend000# and
multiplication in R[x]
mathend000# of polynomials whose product has degree less than n
mathend000#
can be performed using
-
3 n log(n)
mathend000# additions in R
mathend000#,
-
3/2 n log(n) + n - 2
mathend000# multiplications by a power of
mathend000#,
- n
mathend000# multiplications in R
mathend000#,
- n
mathend000# divisions by n
mathend000# (as an element of R
mathend000#),
leading to
9/2 n log(n) +
(n)
mathend000# operations in R
mathend000#.
Figure:
FFT-based univariate polynomial multiplication over
/p
mathend000#
![\begin{figure}\htmlimage[no_transparent]
\centering
\includegraphics[scale=0.5]{fftflowchart.eps}
\end{figure}](img93.png) |
Next: Computing primitive n mathend000#-th roots
Up: Fast Polynomial Multiplication based on
Previous: The Fast Fourier Transform
Marc Moreno Maza
2007-01-10