public static void DebugILGPU()
{
var frequencies = FitsIO.ReadFrequencies(@"C:\dev\GitHub\p9-data\small\fits\simulation_point\freq.fits");
var uvw = FitsIO.ReadUVW(@"C:\dev\GitHub\p9-data\small\fits\simulation_point\uvw.fits");
var flags = new bool[uvw.GetLength(0), uvw.GetLength(1), frequencies.Length]; //completely unflagged dataset
double norm = 2.0;
var visibilities = FitsIO.ReadVisibilities(@"C:\dev\GitHub\p9-data\small\fits\simulation_point\vis.fits", uvw.GetLength(0), uvw.GetLength(1), frequencies.Length, norm);
var visibilitiesCount = visibilities.Length;
int gridSize = 256;
int subgridsize = 8;
int kernelSize = 4;
int max_nr_timesteps = 1024;
double cellSize = 1.0 / 3600.0 * PI / 180.0;
var c = new GriddingConstants(visibilitiesCount, gridSize, subgridsize, kernelSize, max_nr_timesteps, (float)cellSize, 1, 0.0f);
var watchTotal = new Stopwatch();
var watchForward = new Stopwatch();
var watchBackwards = new Stopwatch();
var watchDeconv = new Stopwatch();
watchTotal.Start();
var metadata = Partitioner.CreatePartition(c, uvw, frequencies);
var psfGrid = IDG.GridPSF(c, metadata, uvw, flags, frequencies);
var psf = FFT.Backward(psfGrid, c.VisibilitiesCount);
FFT.Shift(psf);
var psfCutDouble = CutImg(psf);
var psfCut = ToFloatImage(psfCutDouble);
FitsIO.Write(psfCut, "psfCut.fits");
var totalSize = new Rectangle(0, 0, gridSize, gridSize);
var imageSection = new Rectangle(0, 128, gridSize, gridSize);
var bMapCalculator = new PaddedConvolver(PSF.CalcPaddedFourierCorrelation(psfCut, totalSize), new Rectangle(0, 0, psfCut.GetLength(0), psfCut.GetLength(1)));
var fastCD = new FastSerialCD(totalSize, psfCut);
fastCD.ResetLipschitzMap(ToFloatImage(psf));
var gpuCD = new GPUSerialCD(totalSize, psfCut, 100);
var lambda = 0.5f * fastCD.MaxLipschitz;
var alpha = 0.8f;
var xImage = new float[gridSize, gridSize];
var residualVis = visibilities;
/*var truth = new double[gridSize, gridSize];
* truth[30, 30] = 1.0;
* truth[35, 36] = 1.5;
* var truthVis = IDG.ToVisibilities(c, metadata, truth, uvw, frequencies);
* visibilities = truthVis;
* var residualVis = truthVis;*/
for (int cycle = 0; cycle < 4; cycle++)
{
//FORWARD
watchForward.Start();
var dirtyGrid = IDG.Grid(c, metadata, residualVis, uvw, frequencies);
var dirtyImage = FFT.BackwardFloat(dirtyGrid, c.VisibilitiesCount);
FFT.Shift(dirtyImage);
FitsIO.Write(dirtyImage, "dirty_" + cycle + ".fits");
watchForward.Stop();
//DECONVOLVE
watchDeconv.Start();
bMapCalculator.ConvolveInPlace(dirtyImage);
FitsIO.Write(dirtyImage, "bMap_" + cycle + ".fits");
//var result = fastCD.Deconvolve(xImage, dirtyImage, lambda, alpha, 1000, 1e-4f);
var result = gpuCD.Deconvolve(xImage, dirtyImage, lambda, alpha, 1000, 1e-4f);
if (result.Converged)
{
Console.WriteLine("-----------------------------CONVERGED!!!!------------------------");
}
else
{
Console.WriteLine("-------------------------------not converged----------------------");
}
FitsIO.Write(xImage, "xImageGreedy" + cycle + ".fits");
FitsIO.Write(dirtyImage, "residualDebug_" + cycle + ".fits");
watchDeconv.Stop();
//BACKWARDS
watchBackwards.Start();
FFT.Shift(xImage);
var xGrid = FFT.Forward(xImage);
FFT.Shift(xImage);
var modelVis = IDG.DeGrid(c, metadata, xGrid, uvw, frequencies);
residualVis = Visibilities.Substract(visibilities, modelVis, flags);
watchBackwards.Stop();
var hello = FFT.Forward(xImage, 1.0);
hello = Common.Fourier2D.Multiply(hello, psfGrid);
var hImg = FFT.Backward(hello, (double)(128 * 128));
//FFT.Shift(hImg);
FitsIO.Write(hImg, "modelDirty_FFT.fits");
var imgRec = IDG.ToImage(c, metadata, modelVis, uvw, frequencies);
FitsIO.Write(imgRec, "modelDirty" + cycle + ".fits");
}
}
public static void DebugdWStack()
{
var frequencies = FitsIO.ReadFrequencies(@"C:\dev\GitHub\p9-data\large\fits\meerkat_tiny\freq.fits");
var uvw = FitsIO.ReadUVW(@"C:\dev\GitHub\p9-data\large\fits\meerkat_tiny\uvw0.fits");
var flags = FitsIO.ReadFlags(@"C:\dev\GitHub\p9-data\large\fits\meerkat_tiny\flags0.fits", uvw.GetLength(0), uvw.GetLength(1), frequencies.Length);
double norm = 2.0;
var visibilities = FitsIO.ReadVisibilities(@"C:\dev\GitHub\p9-data\large\fits\meerkat_tiny\vis0.fits", uvw.GetLength(0), uvw.GetLength(1), frequencies.Length, norm);
for (int i = 1; i < 8; i++)
{
var uvw0 = FitsIO.ReadUVW(@"C:\dev\GitHub\p9-data\large\fits\meerkat_tiny\uvw" + i + ".fits");
var flags0 = FitsIO.ReadFlags(@"C:\dev\GitHub\p9-data\large\fits\meerkat_tiny\flags" + i + ".fits", uvw0.GetLength(0), uvw0.GetLength(1), frequencies.Length);
var visibilities0 = FitsIO.ReadVisibilities(@"C:\dev\GitHub\p9-data\large\fits\meerkat_tiny\vis" + i + ".fits", uvw0.GetLength(0), uvw0.GetLength(1), frequencies.Length, norm);
uvw = FitsIO.Stitch(uvw, uvw0);
flags = FitsIO.Stitch(flags, flags0);
visibilities = FitsIO.Stitch(visibilities, visibilities0);
}
var maxW = 0.0;
for (int i = 0; i < uvw.GetLength(0); i++)
{
for (int j = 0; j < uvw.GetLength(1); j++)
{
maxW = Math.Max(maxW, Math.Abs(uvw[i, j, 2]));
}
}
maxW = Partitioner.MetersToLambda(maxW, frequencies[frequencies.Length - 1]);
var visCount2 = 0;
for (int i = 0; i < flags.GetLength(0); i++)
{
for (int j = 0; j < flags.GetLength(1); j++)
{
for (int k = 0; k < flags.GetLength(2); k++)
{
if (!flags[i, j, k])
{
visCount2++;
}
}
}
}
var visibilitiesCount = visCount2;
int gridSize = 4096;
int subgridsize = 16;
int kernelSize = 8;
int max_nr_timesteps = 1024;
double cellSize = 1.6 / 3600.0 * PI / 180.0;
int wLayerCount = 32;
double wStep = maxW / (wLayerCount);
var c = new GriddingConstants(visibilitiesCount, gridSize, subgridsize, kernelSize, max_nr_timesteps, (float)cellSize, wLayerCount, wStep);
var c2 = new GriddingConstants(visibilitiesCount, gridSize, subgridsize, kernelSize, max_nr_timesteps, (float)cellSize, 1, 0.0);
var metadata = Partitioner.CreatePartition(c, uvw, frequencies);
var psfVis = new Complex[uvw.GetLength(0), uvw.GetLength(1), frequencies.Length];
for (int i = 0; i < visibilities.GetLength(0); i++)
{
for (int j = 0; j < visibilities.GetLength(1); j++)
{
for (int k = 0; k < visibilities.GetLength(2); k++)
{
if (!flags[i, j, k])
{
psfVis[i, j, k] = new Complex(1.0, 0);
}
else
{
psfVis[i, j, k] = new Complex(0, 0);
}
}
}
}
var psfGrid = IDG.GridW(c, metadata, psfVis, uvw, frequencies);
var psf = FFT.WStackIFFTFloat(psfGrid, c.VisibilitiesCount);
FFT.Shift(psf);
FitsIO.Write(psf, "psfWStack.fits");
var totalSize = new Rectangle(0, 0, gridSize, gridSize);
var bMapCalculator = new PaddedConvolver(PSF.CalcPaddedFourierCorrelation(psf, totalSize), new Rectangle(0, 0, psf.GetLength(0), psf.GetLength(1)));
var fastCD = new FastSerialCD(totalSize, psf);
var lambda = 0.4f * fastCD.MaxLipschitz;
var alpha = 0.1f;
var xImage = new float[gridSize, gridSize];
var residualVis = visibilities;
for (int cycle = 0; cycle < 8; cycle++)
{
var dirtyGrid = IDG.GridW(c, metadata, residualVis, uvw, frequencies);
var dirty = FFT.WStackIFFTFloat(dirtyGrid, c.VisibilitiesCount);
FFT.Shift(dirty);
FitsIO.Write(dirty, "dirty_" + cycle + ".fits");
bMapCalculator.ConvolveInPlace(dirty);
FitsIO.Write(dirty, "bMap_" + cycle + ".fits");
var result = fastCD.Deconvolve(xImage, dirty, lambda, alpha, 10000, 1e-4f);
FitsIO.Write(xImage, "xImageGreedy" + cycle + ".fits");
FFT.Shift(xImage);
var xGrid = FFT.Forward(xImage);
FFT.Shift(xImage);
var modelVis = IDG.DeGridW(c, metadata, xGrid, uvw, frequencies);
var modelGrid = IDG.GridW(c, metadata, modelVis, uvw, frequencies);
var model = FFT.WStackIFFTFloat(modelGrid, c.VisibilitiesCount);
FFT.Shift(model);
FitsIO.Write(model, "model_" + cycle + ".fits");
residualVis = Visibilities.Substract(visibilities, modelVis, flags);
}
}
public static void DebugSimulatedApprox()
{
var frequencies = FitsIO.ReadFrequencies(@"C:\dev\GitHub\p9-data\small\fits\simulation_point\freq.fits");
var uvw = FitsIO.ReadUVW(@"C:\dev\GitHub\p9-data\small\fits\simulation_point\uvw.fits");
var flags = new bool[uvw.GetLength(0), uvw.GetLength(1), frequencies.Length]; //completely unflagged dataset
double norm = 2.0;
var visibilities = FitsIO.ReadVisibilities(@"C:\dev\GitHub\p9-data\small\fits\simulation_point\vis.fits", uvw.GetLength(0), uvw.GetLength(1), frequencies.Length, norm);
var visibilitiesCount = visibilities.Length;
int gridSize = 256;
int subgridsize = 8;
int kernelSize = 4;
int max_nr_timesteps = 1024;
double cellSize = 1.0 / 3600.0 * PI / 180.0;
var c = new GriddingConstants(visibilitiesCount, gridSize, subgridsize, kernelSize, max_nr_timesteps, (float)cellSize, 1, 0.0f);
var watchTotal = new Stopwatch();
var watchForward = new Stopwatch();
var watchBackwards = new Stopwatch();
var watchDeconv = new Stopwatch();
watchTotal.Start();
var metadata = Partitioner.CreatePartition(c, uvw, frequencies);
var psfGrid = IDG.GridPSF(c, metadata, uvw, flags, frequencies);
var psf = FFT.BackwardFloat(psfGrid, c.VisibilitiesCount);
FFT.Shift(psf);
var psfCut = PSF.Cut(psf);
FitsIO.Write(psfCut, "psfCut.fits");
var random = new Random(123);
var totalSize = new Rectangle(0, 0, gridSize, gridSize);
var bMapCalculator = new PaddedConvolver(PSF.CalcPaddedFourierCorrelation(psfCut, totalSize), new Rectangle(0, 0, psfCut.GetLength(0), psfCut.GetLength(1)));
var fastCD = new FastSerialCD(totalSize, psfCut);
//fastCD.ResetAMap(psf);
var lambda = 0.5f * fastCD.MaxLipschitz;
var alpha = 0.8f;
var approx = new ApproxParallel();
var approx2 = new ApproxFast(totalSize, psfCut, 4, 8, 0f, 0.25f, false, true);
var xImage = new float[gridSize, gridSize];
var residualVis = visibilities;
/*var truth = new double[gridSize, gridSize];
* truth[30, 30] = 1.0;
* truth[35, 36] = 1.5;
* var truthVis = IDG.ToVisibilities(c, metadata, truth, uvw, frequencies);
* visibilities = truthVis;
* var residualVis = truthVis;*/
var data = new ApproxFast.TestingData(new StreamWriter("approxConvergence.txt"));
for (int cycle = 0; cycle < 4; cycle++)
{
//FORWARD
watchForward.Start();
var dirtyGrid = IDG.Grid(c, metadata, residualVis, uvw, frequencies);
var dirtyImage = FFT.BackwardFloat(dirtyGrid, c.VisibilitiesCount);
FFT.Shift(dirtyImage);
FitsIO.Write(dirtyImage, "dirty_" + cycle + ".fits");
watchForward.Stop();
//DECONVOLVE
watchDeconv.Start();
//approx.ISTAStep(xImage, dirtyImage, psf, lambda, alpha);
//FitsIO.Write(xImage, "xIsta.fits");
//FitsIO.Write(dirtyImage, "dirtyFista.fits");
//bMapCalculator.ConvolveInPlace(dirtyImage);
//FitsIO.Write(dirtyImage, "bMap_" + cycle + ".fits");
//var result = fastCD.Deconvolve(xImage, dirtyImage, 0.5f * fastCD.MaxLipschitz, 0.8f, 1000, 1e-4f);
//var converged = approx.DeconvolveActiveSet(xImage, dirtyImage, psfCut, lambda, alpha, random, 8, 1, 1);
//var converged = approx.DeconvolveGreedy(xImage, dirtyImage, psfCut, lambda, alpha, random, 4, 4, 500);
//var converged = approx.DeconvolveApprox(xImage, dirtyImage, psfCut, lambda, alpha, random, 1, threads, 500, 1e-4f, cycle == 0);
approx2.DeconvolveTest(data, cycle, 0, xImage, dirtyImage, psfCut, psf, lambda, alpha, random, 10, 1e-4f);
if (data.converged)
{
Console.WriteLine("-----------------------------CONVERGED!!!!------------------------");
}
else
{
Console.WriteLine("-------------------------------not converged----------------------");
}
FitsIO.Write(xImage, "xImageApprox_" + cycle + ".fits");
watchDeconv.Stop();
//BACKWARDS
watchBackwards.Start();
FFT.Shift(xImage);
var xGrid = FFT.Forward(xImage);
FFT.Shift(xImage);
var modelVis = IDG.DeGrid(c, metadata, xGrid, uvw, frequencies);
residualVis = Visibilities.Substract(visibilities, modelVis, flags);
watchBackwards.Stop();
}
var dirtyGridCheck = IDG.Grid(c, metadata, residualVis, uvw, frequencies);
var dirtyCheck = FFT.Backward(dirtyGridCheck, c.VisibilitiesCount);
FFT.Shift(dirtyCheck);
var l2Penalty = Residuals.CalcPenalty(ToFloatImage(dirtyCheck));
var elasticPenalty = ElasticNet.CalcPenalty(xImage, (float)lambda, (float)alpha);
var sum = l2Penalty + elasticPenalty;
data.writer.Close();
}
