Workletize the logistic map.

This commit is contained in:
Nick Thompson 2021-04-09 12:42:40 -04:00
parent 57e8750b33
commit eac6c21c84

@ -13,51 +13,90 @@
#include <vector>
#include <vtkm/cont/DataSetBuilderUniform.h>
#include <vtkm/io/ImageWriterPNG.h>
#include <vtkm/worklet/WorkletMapField.h>
// The logistic map is xᵢ₊₁ = rxᵢ(1-xᵢ).
// If we start this iteration out at (say) x₀ = 0.5, the map has "transients",
// which we must iterate away to produce the final image.
struct LogisticBurnIn : public vtkm::worklet::WorkletMapField
{
using ControlSignature = void(WholeArrayOut);
using ExecutionSignature = void(_1, WorkIndex);
template <typename OutputArrayPortalType>
VTKM_EXEC void operator()(OutputArrayPortalType& outputArrayPortal, vtkm::Id workIndex) const
{
vtkm::Id width = outputArrayPortal.GetNumberOfValues();
double rmin = 2.9;
double r = rmin + (4.0 - rmin) * workIndex / (width - 1);
double x = 0.5;
// 2048 should be enough iterations to get rid of the transients:
int n = 0;
while (n++ < 2048)
{
x = r * x * (1 - x);
}
outputArrayPortal.Set(workIndex, x);
}
};
// After burn-in, the iteration is periodic but in general not convergent,
// i.e., for large enough i, there exists an integer p > 0 such that
// xᵢ₊ₚ = xᵢ for all i.
// So color the pixels corresponding to xᵢ, xᵢ₊₁, .. xᵢ₊ₚ.
struct LogisticLimitPoints : public vtkm::worklet::WorkletMapField
{
using ControlSignature = void(WholeArrayIn, WholeArrayOut);
using ExecutionSignature = void(_1, _2, WorkIndex);
template <typename InputArrayPortalType, typename OutputArrayPortalType>
VTKM_EXEC void operator()(const InputArrayPortalType& inputArrayPortal,
OutputArrayPortalType& outputArrayPortal,
vtkm::Id workIndex) const
{
vtkm::Id width = inputArrayPortal.GetNumberOfValues();
double x = inputArrayPortal.Get(workIndex);
double rmin = 2.9;
double r = rmin + (4.0 - rmin) * workIndex / (width - 1);
vtkm::Vec4f orange(1.0, 0.5, 0.0, 0.0);
// We can't display need more limit points than pixels of height:
vtkm::Id limit_points = 0;
vtkm::Id height = 1800;
while (limit_points++ < height)
{
vtkm::Id j = vtkm::Round(x * (height - 1));
outputArrayPortal.Set(j * width + workIndex, orange);
x = r * x * (1 - x);
}
}
};
int main()
{
size_t height = 1800;
size_t width = height * 1.618;
vtkm::Id height = 1800;
vtkm::Id width = height * 1.618;
vtkm::cont::DataSetBuilderUniform dsb;
vtkm::cont::DataSet ds = dsb.Create(vtkm::Id2(width, height));
std::vector<double> x(width, 0.5);
vtkm::cont::ArrayHandle<double> x;
x.Allocate(width);
vtkm::cont::Invoker invoke;
invoke(LogisticBurnIn{}, x);
double rmin = 2.9;
for (size_t i = 0; i < width; ++i)
vtkm::cont::ArrayHandle<vtkm::Vec4f> pixels;
pixels.Allocate(width * height);
auto wp = pixels.WritePortal();
for (vtkm::Id i = 0; i < pixels.GetNumberOfValues(); ++i)
{
double r = rmin + (4.0 - rmin) * i / (width - 1);
int n = 0;
// 2048 should be enough iterations to be "converged";
// though of course the iterations actually don't all converge but cycle or are chaotic.
while (n++ < 2048)
{
x[i] = r * x[i] * (1 - x[i]);
}
}
vtkm::Vec4f v(1.0, 0.5, 0.0, 0.0);
std::vector<vtkm::Vec4f> pixelValues(width * height, vtkm::Vec4f(0, 0, 0, 0));
size_t iterates = 0;
// We don't need more iterates than pixels of height,
// by the pigeonhole principle.
while (iterates++ < height)
{
for (size_t i = 0; i < width; ++i)
{
double r = rmin + (4.0 - rmin) * i / (width - 1);
double y = x[i];
assert(y >= 0 && y <= 1);
size_t j = std::round(y * (height - 1));
pixelValues[j * width + i] = v;
x[i] = r * x[i] * (1 - x[i]);
}
wp.Set(i, vtkm::Vec4f(0, 0, 0, 0));
}
invoke(LogisticLimitPoints{}, x, pixels);
std::string colorFieldName = "pixels";
ds.AddPointField(colorFieldName, pixelValues);
ds.AddPointField(colorFieldName, pixels);
std::string filename = "logistic.png";
vtkm::io::ImageWriterPNG writer(filename);
writer.WriteDataSet(ds, colorFieldName);