cannam@127: cannam@127: cannam@127: cannam@127: cannam@127:
cannam@127:cannam@127: Next: 1d Discrete Hartley Transforms (DHTs), Previous: 1d Real-even DFTs (DCTs), Up: What FFTW Really Computes [Contents][Index]
cannam@127:The Real-odd symmetry DFTs in FFTW are exactly equivalent to the unnormalized cannam@127: forward (and backward) DFTs as defined above, where the input array cannam@127: X of length N is purely real and is also odd symmetry. In cannam@127: this case, the output is odd symmetry and purely imaginary. cannam@127: cannam@127: cannam@127:
cannam@127: cannam@127: cannam@127:For the case of RODFT00
, this odd symmetry means that
cannam@127: Xj = -XN-j,where we take X to be periodic so that
cannam@127: XN = X0.Because of this redundancy, only the first n real numbers
cannam@127: starting at j=1 are actually stored (the j=0 element is
cannam@127: zero), where N = 2(n+1).
cannam@127:
The proper definition of odd symmetry for RODFT10
,
cannam@127: RODFT01
, and RODFT11
transforms is somewhat more intricate
cannam@127: because of the shifts by 1/2 of the input and/or output, although
cannam@127: the corresponding boundary conditions are given in Real even/odd DFTs (cosine/sine transforms). Because of the odd symmetry, however,
cannam@127: the cosine terms in the DFT all cancel and the remaining sine terms are
cannam@127: written explicitly below. This formulation often leads people to call
cannam@127: such a transform a discrete sine transform (DST), although it is
cannam@127: really just a special case of the DFT.
cannam@127:
cannam@127:
cannam@127:
In each of the definitions below, we transform a real array X of cannam@127: length n to a real array Y of length n: cannam@127:
cannam@127: cannam@127:An RODFT00
transform (type-I DST) in FFTW is defined by:
cannam@127:
An RODFT10
transform (type-II DST) in FFTW is defined by:
cannam@127:
An RODFT01
transform (type-III DST) in FFTW is defined by:
cannam@127:
An RODFT11
transform (type-IV DST) in FFTW is defined by:
cannam@127:
These definitions correspond directly to the unnormalized DFTs used
cannam@127: elsewhere in FFTW (hence the factors of 2 in front of the
cannam@127: summations). The unnormalized inverse of RODFT00
is
cannam@127: RODFT00
, of RODFT10
is RODFT01
and vice versa, and
cannam@127: of RODFT11
is RODFT11
. Each unnormalized inverse results
cannam@127: in the original array multiplied by N, where N is the
cannam@127: logical DFT size. For RODFT00
, N=2(n+1);
cannam@127: otherwise, N=2n.
cannam@127:
cannam@127:
In defining the discrete sine transform, some authors also include cannam@127: additional factors of cannam@127: √2(or its inverse) multiplying selected inputs and/or outputs. This is a cannam@127: mostly cosmetic change that makes the transform orthogonal, but cannam@127: sacrifices the direct equivalence to an antisymmetric DFT. cannam@127:
cannam@127:cannam@127: Next: 1d Discrete Hartley Transforms (DHTs), Previous: 1d Real-even DFTs (DCTs), Up: What FFTW Really Computes [Contents][Index]
cannam@127: