next up previous contents
Next: The algorithm Up: NFFT - nonequispaced fast Previous: The second approximation -   Contents

The case $ d>1$

Starting with the original problem of evaluating the multivariate trigonometric polynomial in (2.1) one has to do a few generalisations. The window function is given by

$\displaystyle \varphi\left(\ensuremath{\boldsymbol{x}}\right):=\varphi_0\left(x_0\right) \varphi_1\left(x_1\right) \hdots \varphi_{d-1}\left(x_{d-1}\right)$    

where $ \varphi_t$ is an univariate window function. Thus, a simple consequence is

$\displaystyle c_{\ensuremath{\boldsymbol{k}}}\left(\tilde \varphi\right) = c_{k...
...ft(\tilde \varphi_1\right) \hdots c_{k_{d-1}}\left(\tilde \varphi_{d-1}\right).$    

The ansatz is generalised to

$\displaystyle s_1\left(\ensuremath{\boldsymbol{x}}\right) := \sum_{\ensuremath{...
...{x}} - \ensuremath{\boldsymbol{n}}^{-1}\odot\ensuremath{\boldsymbol{l}}\right),$    

where the FFT size is given by $ \ensuremath{\boldsymbol{n}}:=\ensuremath{\boldsymbol{\sigma}} \odot \ensuremath{\boldsymbol{N}}$ and the oversampling factors by $ \ensuremath{\boldsymbol{\sigma}}=\left(\sigma_0,\hdots,\sigma_{d-1}\right)^{\top}$ . Along the lines of (2.7) one defines

$\displaystyle \hat g_{\ensuremath{\boldsymbol{k}}} := \left\{ \begin{array}{ll}...
...symbol{n}}} \backslash I_{\ensuremath{\boldsymbol{N}}} . \\ \end{array} \right.$    

The values $ g_{\ensuremath{\boldsymbol{l}}}$ can be obtained by a (multivariate) FFT of size $ n_0
\times n_1 \times \hdots \times n_{d-1}$ as

$\displaystyle g_{\ensuremath{\boldsymbol{l}}} = \frac{1}{\left\vert I_{\ensurem...
...right)}, \quad \ensuremath{\boldsymbol{l}} \in I_{\ensuremath{\boldsymbol{n}}}.$    

Using the compactly supported function $ \psi\left(\ensuremath{\boldsymbol{x}}\right)=\varphi\left(\ensuremath{\boldsym...
...ght)\chi_{[-{m \over n},{m \over
n}]^d}\left(\ensuremath{\boldsymbol{x}}\right)$ , one obtains

$\displaystyle s\left(\ensuremath{\boldsymbol{x}}_j\right) := \sum_{\ensuremath{...
... - \ensuremath{\boldsymbol{n}}^{-1}\odot\ensuremath{\boldsymbol{l}}\right) \, ,$    

where $ \tilde \psi$ again denotes the one periodic version of $ \psi$ and the multi-index set is given by

$\displaystyle I_{\ensuremath{\boldsymbol{n}},m} \left(\ensuremath{\boldsymbol{x...
... \odot \ensuremath{\boldsymbol{x}}_j + m \ensuremath{\boldsymbol{1}}\right\}\,.$    


next up previous contents
Next: The algorithm Up: NFFT - nonequispaced fast Previous: The second approximation -   Contents
Jens Keiner 2006-11-20