public void TestNextPowerOf2()
{
Assert.AreEqual(1024, SignalExtension.NextPowerOf2(1000));
Assert.AreEqual(2048, SignalExtension.NextPowerOf2(1024));
Assert.AreEqual(1, SignalExtension.NextPowerOf2(0));
Assert.AreEqual(0, SignalExtension.NextPowerOf2(-100));
}
/// <summary>
/// Compute the forward or inverse Fourier Transform of data, with data
/// containing complex valued data as alternating real and imaginary
/// parts. The length must be a power of 2. This method caches values
/// and should be slightly faster on repeated uses than then FFT method.
/// It is also slightly more accurate.
/// </summary>
/// <param name="data">The complex data stored as alternating real
/// and imaginary parts</param>
/// <param name="forward">true for a forward transform, false for
/// inverse transform</param>
public void TableFFT(ref double[] data, bool forward)
{
var n = data.Length;
// checks n is a power of 2 in 2's complement format
if ((n & (n - 1)) != 0)
{
data = data.SubArray(SignalExtension.NextPowerOf2(n));
n = data.Length;
}
n /= 2; // n is the number of samples
Reverse(ref data, n); // bit index data reversal
// make table if needed
if (_cosTable.Count != n)
{
Initialize(n);
}
// do transform: so single point transforms, then doubles, etc.
double sign = forward ? 1 : -1;
var mmax = 1;
var tptr = 0;
while (n > mmax)
{
var istep = 2 * mmax;
for (var m = 0; m < istep; m += 2)
{
var wr = _cosTable[tptr];
var wi = sign * _sinTable[tptr++];
for (var k = m; k < 2 * n; k += 2 * istep)
{
var j = k + istep;
var tempr = wr * data[j] - wi * data[j + 1];
var tempi = wi * data[j] + wr * data[j + 1];
data[j] = data[k] - tempr;
data[j + 1] = data[k + 1] - tempi;
data[k] = data[k] + tempr;
data[k + 1] = data[k + 1] + tempi;
}
}
mmax = istep;
}
// copy out with optional scaling
if (forward)
{
return;
}
var scale = 1.0 / n;
for (var i = 0; i < 2 * n; ++i)
{
data[i] *= scale;
}
}
/// <summary>
/// Compute the forward or inverse Fourier Transform of data, with
/// data containing complex valued data as alternating real and
/// imaginary parts. The length must be a power of 2.
/// </summary>
/// <param name="data">The complex data stored as alternating real
/// and imaginary parts</param>
/// <param name="forward">true for a forward transform, false for
/// inverse transform</param>
public void DynamicFFT(ref double[] data, bool forward)
{
var n = data.Length;
// checks n is a power of 2 in 2's complement format
if ((n & (n - 1)) != 0)
{
data = data.SubArray(SignalExtension.NextPowerOf2(n));
n = data.Length;
}
n /= 2; // n is the number of samples
Reverse(ref data, n); // bit index data reversal
// do transform: so single point transforms, then doubles, etc.
double sign = forward ? 1 : -1;
var mmax = 1;
while (n > mmax)
{
var istep = 2 * mmax;
var theta = sign * Math.PI / mmax;
double wr = 1, wi = 0;
var wpr = Math.Cos(theta);
var wpi = Math.Sin(theta);
for (var m = 0; m < istep; m += 2)
{
for (var k = m; k < 2 * n; k += 2 * istep)
{
var j = k + istep;
var tempr = wr * data[j] - wi * data[j + 1];
var tempi = wi * data[j] + wr * data[j + 1];
data[j] = data[k] - tempr;
data[j + 1] = data[k + 1] - tempi;
data[k] = data[k] + tempr;
data[k + 1] = data[k + 1] + tempi;
}
var t = wr; // trig recurrence
wr = wr * wpr - wi * wpi;
wi = wi * wpr + t * wpi;
}
mmax = istep;
}
// inverse scaling in the backward case
if (forward)
{
return;
}
var scale = 1.0 / n;
for (var i = 0; i < 2 * n; ++i)
{
data[i] *= scale;
}
}
static bool Test(FFTAction fftFunction, double[] test, IList <double> answer)
{
// Test the given function on the given data and see if the result is the given answer.
var returnValue = true;
var copy = test.ToArray(); // make a copy
if ((test.Length & (test.Length - 1)) != 0)
{
test = test.SubArray(SignalExtension.NextPowerOf2(test.Length));
}
fftFunction(ref copy, true); // forward transform
returnValue &= Compare(copy, answer); // check it
fftFunction(ref copy, false); // backward transform
returnValue &= Compare(copy, test); // check it
return(returnValue);
}
/// <summary>
/// Executes the block
/// </summary>
public override void Execute()
{
var connectingNode = InputNodes[0].ConnectingNode as BlockOutputNode;
if (connectingNode == null || connectingNode.Object == null)
{
return;
}
int beforeSize = 0, afterSize = 0;
if (ExtensionSize > 0)
{
beforeSize = ExtensionSize;
afterSize = ExtensionSize;
}
OutputNodes[0].Object.Clear();
foreach (var signal in connectingNode.Object)
{
if (ExtensionSize <= 0)
{
var size = SignalExtension.NextPowerOf2(signal.Samples.Length);
beforeSize = afterSize = (size - signal.Samples.Length) / 2;
while (beforeSize + afterSize + signal.Samples.Length < size)
{
afterSize++;
}
}
var output = signal.Clone();
SignalExtension.Extend(ref output, ExtensionMode, beforeSize, afterSize);
OutputNodes[0].Object.Add(output);
}
if (Cascade && OutputNodes[0].ConnectingNode != null)
{
OutputNodes[0].ConnectingNode.Root.Execute();
}
}
/// <summary>
/// Compute the forward or inverse Fourier Transform of data, with
/// data containing real valued data only. The output is complex
/// valued after the first two entries, stored in alternating real
/// and imaginary parts. The first two returned entries are the real
/// parts of the first and last value from the conjugate symmetric
/// output, which are necessarily real. The length must be a power
/// of 2.
/// </summary>
/// <param name="data">The complex data stored as alternating real
/// and imaginary parts</param>
/// <param name="forward">true for a forward transform, false for
/// inverse transform</param>
public void RealFFT(ref double[] data, bool forward)
{
var n = data.Length; // # of real inputs, 1/2 the complex length
// checks n is a power of 2 in 2's complement format
if ((n & (n - 1)) != 0)
{
data = data.SubArray(SignalExtension.NextPowerOf2(n));
n = data.Length;
}
double sign = -1;
if (forward)
{
// do packed FFT. This can be changed to FFT to save memory
TableFFT(ref data, true);
sign = 1;
}
var theta = sign * 2 * Math.PI / n;
var wpr = Math.Cos(theta);
var wpi = Math.Sin(theta);
var wjr = wpr;
var wji = wpi;
for (var j = 1; j <= n / 4; ++j)
{
var k = n / 2 - j;
var tnr = data[2 * k];
var tni = data[2 * k + 1];
var tjr = data[2 * j];
var tji = data[2 * j + 1];
var e = (tjr + tnr);
var f = (tji - tni);
var a = (tjr - tnr) * wji;
var d = (tji + tni) * wji;
var b = (tji + tni) * wjr;
var c = (tjr - tnr) * wjr;
// compute entry y[j]
data[2 * j] = 0.5 * (e + sign * (a + b));
data[2 * j + 1] = 0.5 * (f - sign * (c - d));
// compute entry y[k]
data[2 * k] = 0.5 * (e - sign * (a + b));
data[2 * k + 1] = 0.5 * (sign * (-c + d) - f);
var temp = wjr;
// todo - allow more accurate version here? make option?
wjr = wjr * wpr - wji * wpi;
wji = temp * wpi + wji * wpr;
}
if (forward)
{
// compute final y0 and y_{N/2}, store data[0], data[1]
var temp = data[0];
data[0] += data[1];
data[1] = temp - data[1];
}
else
{
var temp = data[0]; // unpack the y[j], then invert FFT
data[0] = 0.5 * (temp + data[1]);
data[1] = 0.5 * (temp - data[1]);
// do packed FFT. This can be changed to FFT to save memory
TableFFT(ref data, false);
}
}
/// <summary>
/// Convolves vectors input and filter using a managed FFT algorithm.
/// </summary>
/// <param name="input">The input signal</param>
/// <param name="filter">The filter</param>
/// <param name="returnOnlyValid">True to return only the middle of the array</param>
/// <param name="margin">Margin to be used if returnOnlyValid is set to true</param>
/// <param name="mode">Mode</param>
/// <returns></returns>
public static double[] ConvolveManagedFFT(double[] input, double[] filter, bool returnOnlyValid = true, int margin = 0, ManagedFFTModeEnum mode = ManagedFFTModeEnum.UseLookupTable)
{
if (input == null || filter == null)
{
return(null);
}
if (input.Length < filter.Length)
{
var auxSignal = input;
input = filter;
filter = auxSignal;
}
var realSize = input.Length + filter.Length - 1;
var size = ((realSize > 0) && ((realSize & (realSize - 1)) == 0) ? realSize : SignalExtension.NextPowerOf2(realSize));
var inputFFT = MemoryPool.Pool.New <double>(size * 2);
var filterFFT = MemoryPool.Pool.New <double>(size * 2);
var ifft = MemoryPool.Pool.New <double>(size * 2);
for (var i = 0; i < input.Length; i++)
{
inputFFT[i * 2] = input[i];
}
for (var i = 0; i < filter.Length; i++)
{
filterFFT[i * 2] = filter[i];
}
ManagedFFT.Instance.FFT(ref inputFFT, true, mode);
ManagedFFT.Instance.FFT(ref filterFFT, true, mode);
for (var i = 0; i < ifft.Length; i = i + 2)
{
ifft[i] = inputFFT[i] * filterFFT[i] - inputFFT[i + 1] * filterFFT[i + 1];
ifft[i + 1] = (inputFFT[i] * filterFFT[i + 1] + inputFFT[i + 1] * filterFFT[i]) * -1;
}
ManagedFFT.Instance.FFT(ref ifft, false, mode);
var ifft2 = MemoryPool.Pool.New <double>(size);
ifft2[0] = ifft[0];
for (var i = 0; i < ifft2.Length - 2; i = i + 1)
{
ifft2[i + 1] = ifft[ifft.Length - 2 - i * 2];
}
int start;
if (returnOnlyValid)
{
size = input.Length - filter.Length + 1;
var padding = (realSize - size) / 2;
start = padding + margin;
size = input.Length - filter.Length - margin * 2 + 1;
}
else
{
start = 0;
size = realSize;
}
var result = MemoryPool.Pool.New <double>(size);
Array.Copy(ifft2, start, result, 0, size);
return(result);
}
