org.das2.qds.util

Class FFTUtil

java.lang.Object

org.das2.qds.util.FFTUtil

public class FFTUtil
extends java.lang.Object

Utilities for FFT operations, such as getting the frequencies for each bin and fftPower.

Constructor Summary

Constructors

Constructor and Description
FFTUtil()

Method Summary

Modifier and Type	Method and Description
static ComplexArray.Double	fft(GeneralFFT fft, QDataSet vds) helper method for doing FFT of a QDataSet.
static QDataSet	fft(GeneralFFT fft, QDataSet vds, QDataSet weights) Perform the fft to get real and imaginary components for intervals.
static QDataSet	fftPower(GeneralFFT fft, QDataSet vds)
static QDataSet	fftPower(GeneralFFT fft, QDataSet vds, QDataSet weights) Produces the power spectrum of the dataset.
static QDataSet	fftPower(GeneralFFT fft, QDataSet vds, QDataSet weights, QDataSet powxTags) Produces the power spectrum of the dataset.
static double[]	getFrequencyDomainTags(double fs, int size)
static QDataSet	getFrequencyDomainTags(QDataSet timeDomainTags) return the frequency tags for the given time offset tags, so f[i]= i / n*T where T= time[1] - time[0] and n= len(time).
static QDataSet	getFrequencyDomainTagsForPower(QDataSet dep0) get the frequency tags, for use when calculating the power in each channel.
static QDataSet	getTimeDomainTags(QDataSet frequencyDomainTags) return the time domain tags for inverse fft.
static QDataSet	getWindow10PercentEdgeCosine(int size) Window with ones in the middle, and then the last 10% taper with cos.
static QDataSet	getWindowHanning(int size) return a "Hanning" (Hann) window of the given size.
static QDataSet	getWindowUnity(int size) Window that is all ones, also called a boxcar.
static ComplexArray.Double	ifft(GeneralFFT fft, QDataSet vds) helper method for doing inverse FFT of a QDataSet.
static QDataSet	ifft(GeneralFFT fft, QDataSet vds, QDataSet weights) Perform the inverse fft to get real and imaginary components for intervals.
static boolean	uniformCadence(QDataSet x) return true if the rank 1 timetags are uniform.
static QDataSet	window(QDataSet ds, int size) returns a rank 2 dataset from the rank 1 dataset, where the FFT would be run on each of the datasets.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

FFTUtil

public FFTUtil()

Method Detail

fftPower

public static QDataSet fftPower(GeneralFFT fft,
                                QDataSet vds)

window

public static QDataSet window(QDataSet ds,
                              int size)

returns a rank 2 dataset from the rank 1 dataset, where the FFT would be run on each of the datasets.

Parameters:

ds - rank 1 dataset of length N

size - size of each FFT.

Returns:

rank 2 dataset[N/size,size]

getWindowUnity

public static QDataSet getWindowUnity(int size)

Window that is all ones, also called a boxcar.

Parameters:

size - the window size

Returns:

window

getWindow10PercentEdgeCosine

public static QDataSet getWindow10PercentEdgeCosine(int size)

Window with ones in the middle, and then the last 10% taper with cos.

Parameters:

size - the window size

Returns:

window

getWindowHanning

public static QDataSet getWindowHanning(int size)

return a "Hanning" (Hann) window of the given size.

Parameters:

size -

Returns:

fftPower

public static QDataSet fftPower(GeneralFFT fft,
                                QDataSet vds,
                                QDataSet weights)

Produces the power spectrum of the dataset. This is the length of the Fourier components squared, normalized by the bandwidth. The result dataset has dimensionless yunits. It's assumed that all the data is valid. Note when the input is in mV/m, the result will be in (V/m)^2/Hz.

Parameters:

fft - FFT engine

vds - rank 1 dataset with depend 0 units TimeLocationUnits.

weights - rank 1 datasets that is the window to apply to the data.

Returns:

the rank 2 FFT

fftPower

public static QDataSet fftPower(GeneralFFT fft,
                                QDataSet vds,
                                QDataSet weights,
                                QDataSet powxTags)

Produces the power spectrum of the dataset. This is the length of the Fourier components squared, normalized by the bandwidth. The result dataset has dimensionless yunits. It's assumed that all the data is valid. Note when the input is in mV/m, the result will be in (V/m)^2/Hz.

Parameters:

fft - FFT engine

vds - rank 1 dataset with depend 0 units TimeLocationUnits.

weights - rank 1 datasets that is the window to apply to the data.

powxTags - if non-null, then use these xtags instead of calculating them for each record.

Returns:

the rank 2 FFT

fft

public static QDataSet fft(GeneralFFT fft,
                           QDataSet vds,
                           QDataSet weights)

Perform the fft to get real and imaginary components for intervals.

Parameters:

fft - FFT code to use, such as GeneralFFT.newDoubleFFT(len)

vds - QDataSet rank 1 dataset with depend 0 units TimeLocationUnits.

weights - QDataSet rank 1 dataset containing weights, as in hanning. null indicates no weights.

Returns:

the rank 2 FFT

ifft

public static QDataSet ifft(GeneralFFT fft,
                            QDataSet vds,
                            QDataSet weights)

Perform the inverse fft to get real and imaginary components for intervals.

Parameters:

fft - FFT code to use, such as GeneralFFT.newDoubleFFT(len)

vds - QDataSet rank 2 dataset with depend 0 units TimeLocationUnits and depend_1=['real','imaginary'].

weights - QDataSet rank 1 dataset containing weights, as in hanning. null indicates no weights.

Returns:

the rank 2 FFT

uniformCadence

public static boolean uniformCadence(QDataSet x)

return true if the rank 1 timetags are uniform.

Parameters:

x - rank 1 monotonically increasing timetags.

Returns:

true if the rank 1 timetags are uniform.

getFrequencyDomainTagsForPower

public static QDataSet getFrequencyDomainTagsForPower(QDataSet dep0)

get the frequency tags, for use when calculating the power in each channel. This removes the DC channel, and folds over the negative frequencies. This also keeps a cache for performance.

Parameters:

dep0 - the timetags.

Returns:

the frequency tags.

fft

public static ComplexArray.Double fft(GeneralFFT fft,
                                      QDataSet vds)

helper method for doing FFT of a QDataSet. This forward FFT returns normalized data. GeneralFFT.newDoubleFFT(ds.length());

Parameters:

fft - the FFT engine

vds - rank 1 ds[n] or rank 2 ds[n,2]

Returns:

See Also:

Ops.fft(org.das2.qds.QDataSet)

ifft

public static ComplexArray.Double ifft(GeneralFFT fft,
                                       QDataSet vds)

helper method for doing inverse FFT of a QDataSet.

Parameters:

fft - the FFT engine

vds - rank 2 dataset[n;real,complex]

Returns:

rank 2 dataset[n;real,complex]

See Also:

Ops.ifft(org.das2.qds.QDataSet)

getFrequencyDomainTags

public static double[] getFrequencyDomainTags(double fs,
                                              int size)

Parameters:

fs - the sampling frequency

size - the size of the time domain data.

Returns:

the frequencies of the bins

getTimeDomainTags

public static QDataSet getTimeDomainTags(QDataSet frequencyDomainTags)

return the time domain tags for inverse fft.

Parameters:

frequencyDomainTags -

Returns:

the time Domain Tags

getFrequencyDomainTags

public static QDataSet getFrequencyDomainTags(QDataSet timeDomainTags)

return the frequency tags for the given time offset tags, so f[i]= i / n*T where T= time[1] - time[0] and n= len(time). for the first n/2 points and f[i]= (n21-n+i) / ( n*T ) for the second half. When units have a known inverse, this is returned.

Parameters:

timeDomainTags -

Returns:

the frequency domain tags.