Rate This Document
Findability
Accuracy
Completeness
Readability

kml_fft(f)_plan_many_r2r

Create a plan for the n-dimensional R2R transform of howmany data sequences.

The data sequence of a single FFT does not need to be contiguous and can cross steps.

Interface Definition

C interface:

kml_fft_plan kml_fft_plan_many_r2r(int rank, const int *n, int howmany, double *in, const int *inembed, int istride, int idist, double *out, const int *onembed, int ostride, int odist, const kml_fft_r2r_kind *kind, unsigned flags);

kml_fftf_plan kml_fft_plan_many_r2r(int rank, const int *n, int howmany, float *in, const int *inembed, int istride, int idist, float *out, const int *onembed, int ostride, int odist, const kml_fftf_r2r_kind *kind, unsigned flags);

Fortran interface:

RES = KML_FFT_PLAN_MANY_DFT_R2R(RANK, N, HOWMANY, IN, INEMBED, ISTRIDE, IDIST, OUT, ONEMBED, OSTRIDE, ODIST, KIND, FLAGS);

RES = KML_FFTF_PLAN_MANY_DFT_R2R(RANK, N, HOWMANY, IN, INEMBED, ISTRIDE, IDIST, OUT, ONEMBED, OSTRIDE, ODIST, KIND, FLAGS);

KML_FFT_REDFT11 and KML_FFT_ROODFT11 support only sequences whose length is an integer multiple of 4. Other transforms support only sequences whose length is an integer multiple of 2.

Return Value

The function returns a structure pointer of the kml_fftf_plan type. This object is used as a parameter in the kml_fft(f)_execute function to perform FFT on the current input and output. In addition, the object may also be added into the kml_fft(f)_execute_r2r function as a parameter to perform FFT on the new input and output.

If this function returns a non-null pointer, the plan has been successfully executed. Otherwise, the plan failed to be executed.

Parameters

Parameter

Data Type

Description

Input/Output

rank

int

Dimension of FFT. The constraint is 1 ≤ rank ≤ 3.

Input

n

const int*

Indicates an array whose dimension is rank, including the size of each dimension in the FFT sequence. The constraint is n[i] ≥ 1, for i in 0 to rank - 1.

Input

howmany

int

howmany indicates how many multi-dimensional FFTs are needed.

Input

in

  • Double precision: double*
  • Single precision: float*

Inputs the data to be transformed.

Input

inembed

const int*

inembed is an array whose size is rank or NULL. This array indicates the size of each dimension of a larger space for input FFT data storage.

Constraint: inembed[i] ≥ n[i] for i in 0, rank-1. Or if inembed == NULL, inembed is equal to n.

Input

istride

int

Interval between successive elements of the i-th dimensional FFT input sequence.

Input

idist

int

idist indicates the interval between FFT input sequences.

Input

out

  • Double precision: double*
  • Single precision: float*

Outputs the data generated using FFT.

Output

onembed

const int*

onembed is an array whose size is rank or NULL. This array indicates the size of each dimension of a larger space for output FFT data storage.

Constraint: onembed[i] ≥ n[i] for i in 0, rank-1. Or if onembed == NULL, onembed is equal to n.

Input

ostride

int

Interval between successive elements of the i-th dimensional FFT output sequence.

Input

odist

int

odist indicates the interval between FFT output sequences.

Input

kind

  • Double precision: const kml_fft_r2r_kind*
  • Single precision: const kml_fftf_r2r_kind*

kind is an array whose size is rank, including the R2R transform type in each dimension of an FFT sequence. kind[i] (for i in 0 to rank - 1) has the following options:

  • KML_FFT_R2HC
  • KML_FFT_HC2R
  • KML_FFT_DHT
  • KML_FFT_REDFT00
  • KML_FFT_REDFT01
  • KML_FFT_REDFT10
  • KML_FFT_REDFT11
  • KML_FFT_RODFT00
  • KML_FFT_RODFT01
  • KML_FFT_RODFT10
  • KML_FFT_RODFT11

Input

flags

unsigned int

A planning option, not in use.

Input

Dependencies

C: "kfft.h"

Fortran: "kfft.f03"

Examples

C interface:

    int rank = 1; 
    int *n; 
    n = (int*)kml_fft_malloc(sizeof(int) * rank); 
    n[0] = 2; 
    int howmany = 3; 
    int istride = 1; 
    int ostride = 1; 
    int idist = 2; 
    int odist = 2; 
    double init[6] = {120, 1, 8, 8, 120, 1}; 
    double *in; 
    in = (double*)kml_fft_malloc(sizeof(double) * 6); 
    for (int i = 0; i < 6; i++) { 
        in[i] = init[i]; 
    } 
    double *out; 
    out = (double*)kml_fft_malloc(sizeof(double) * 6); 
    kml_fft_r2r_kind *kind; 
    kind = (kml_fft_r2r_kind*)kml_fft_malloc(sizeof(kml_fft_r2r_kind) * rank); 
    kind[0] = KML_FFT_DHT;     
    kml_fft_plan plan; 
    plan = kml_fft_plan_many_r2r(rank, n, howmany, in, NULL, istride, idist, out, NULL, ostride, odist, kind, KML_FFT_ESTIMATE); 
    kml_fft_execute_r2r(plan, in, out); 
 
    kml_fft_destroy_plan(plan); 
    kml_fft_free(n); 
    kml_fft_free(kind); 
    kml_fft_free(in); 
    kml_fft_free(out); 
 
    /* 
     * out = {1.210000e+02, 1.190000e+02, 1.600000e+01, 0.000000e+00, 
     *        1.210000e+02, 1.190000e+02} 
     */

Fortran interface:

    INTEGER(C_INT) :: RANK = 1 
    INTEGER(C_INT) :: N(1) 
    INTEGER(C_INT) :: KIND(1) 
    REAL(C_DOUBLE), DIMENSION(6) :: INIT 
    TYPE(C_DOUBLE), POINTER :: IN(:) 
    TYPE(KML_FFT_COMPLEX), POINTER :: OUT(:) 
    INTEGER, POINTER :: INEMBED(:), ONEMBED(:) 
    TYPE(C_PTR) :: PIN, POUT 
    INTEGER(C_INT) :: HOWMANY = 3 
    INTEGER(C_INT) :: ISTRIDE = 1 
    INTEGER(C_INT) :: OSTRIDE = 1 
    INTEGER(C_INT) :: IDIST = 2 
    INTEGER(C_INT) :: ODIST = 2 
    INTEGER(C_SIZE_T) :: SIZE 
    RSIZE = 6 
    PIN = KML_FFT_MALLOC(8 * SIZE) 
    POUT = KML_FFT_MALLOC(8 * SIZE) 
    CALL C_F_POINTER(PIN, IN, SHAPE=[6]) 
    CALL C_F_POINTER(POUT, OUT, SHAPE=[6]) 
    CALL C_F_POINTER(C_NULL_PTR, INEMBED, SHAPE=[0]) 
    CALL C_F_POINTER(C_NULL_PTR, ONEMBED, SHAPE=[0]) 
    N(1) = 2 
    DATA INIT/120, 1, 8, 8, 120, 1/ 
    INTEGER :: I 
    DO WHILE(I <= 6) 
        IN(I) = INIT(I) 
    END DO 
    KIND(1) = KML_FFT_DHT 
 
    TYPE(C_PTR) :: PLAN 
    PLAN = KML_FFT_PLAN_MANY_DFT_R2R(RANK, N, HOWMANY, IN, INEMBED, ISTRIDE, IDIST, OUT, ONEMBED, OSTRIDE, ODIST, KIND, KML_FFT_ESTIMATE); 
    CALL KML_FFT_EXECUTE_DFT_R2R(PLAN, IN, OUT); 
 
    CALL KML_FFT_DESTROY_PLAN(PLAN) 
    CALL KML_FFT_FREE(PIN) 
    CALL KML_FFT_FREE(POUT) 
    ! 
    ! OUT = /1.210000E+02, 1.190000E+02, 1.600000E+01, 0.000000E+00, 
    !        1.210000E+02, 1.190000E+02/ 
    !