Module planning

source code

Classes

[hide private]

Plan
Object representing a fftw plan used to execute Fourier transforms in fftw

Functions

[hide private]

create_aligned_array(shape, dtype=<type 'numpy.complex64'>, boundary=16)
Create an array which is aligned in memory

source code

execute(plan)
Execute fftw-plan, i.e.

source code

guru_execute_dft(plan, inarray, outarray)
Guru interface: perform Fourier transform on two arrays, outarray=fft(inarray) using the given plan.

source code

destroy_plan(plan)
Delete the given plan

source code

select(inarray, outarray)
From a given input and output np array select the appropriate fftw3 plan to create.

source code

_create_complex2real_plan(inarray, outarray, flags)
Internal function to create complex fft plan given an input and output np array and the direction and flags integers

source code

_create_complex_plan(inarray, outarray, direction, flags)
Internal function to create complex fft plan given an input and output np array and the direction and flags integers

source code

_create_real_plan(inarray, outarray, realtype, flags)
Internal function to create real fft plan given an input and output np array and the realtype and flags integers

source code

_create_plan(inarray, outarray, direction='forward', flags=['estimate'], realtypes=None, nthreads=1)
Internal function to create a complex fft plan given an input and output np array and the direction and flags integers source code

_cal_flag_value(flags)
Calculate the integer flag value from a list of string flags

source code

print_plan(plan)
Print a nerd-readable version of the plan to stdout

source code

fprint_plan(plan, filename)
Print a nerd-readable version of the plan to the given filename

source code

Variables

[hide private]

fftw_flags = {'conserve memory': 4, 'destroy input': 1, 'estim...

realfft_type = {'discrete hartley': 2, 'halfcomplex c2r': 1, '...

fft_direction = {'backward': 1, 'forward': -1}

__package__ = 'fftw3f'

Function Details

[hide private]

create_aligned_array(shape, dtype=<type 'numpy.complex64'>, boundary=16)

source code

Create an array which is aligned in memory

Parameters:
    shape       --  the shape of the array 
    dtype       --  the dtype of the array (default=typeDict['singlecomplex'])
    boundary    --  the byte boundary to align to (default=16)

execute(plan)

source code

Execute fftw-plan, i.e. perform Fourier transform on the arrays given when the plan was created

guru_execute_dft(plan, inarray, outarray)

source code

Guru interface: perform Fourier transform on two arrays, outarray=fft(inarray) using the given plan. Important: This function does not perform any checks on the array shape and alignment for performance reasons. It is therefore crucial to only provide arrays with the same shape, dtype and alignment as the arrays used for planning, failure to do so can lead to unexpected behaviour and even python segfaulting.

Variables Details

[hide private]

fftw_flags

Value:

{'conserve memory': 4,
 'destroy input': 1,
 'estimate': 64,
 'exhaustive': 8,
 'measure': 0,
 'patient': 32,
 'preserve input': 16,
 'unaligned': 2}

realfft_type

Value:

{'discrete hartley': 2,
 'halfcomplex c2r': 1,
 'halfcomplex r2c': 0,
 'realeven 00': 3,
 'realeven 01': 4,
 'realeven 10': 5,
 'realeven 11': 6,
 'realodd 00': 7,
...