Dynamic Complex Arrays about enoki

[wjakob]

Complex<DynamicArray> is the way to go. Dynamic arrays always come at the deepest level of the hierarchy. What is the problem with the map function? (I assume you mean vectorize?)

[gergol]

Thanks for the quick reply! Sorry for being unclear. I mean the the DynamicArray::map function. My goal is to be able to use the output of an FFT with enoki.

[wjakob]

Enoki assumes that the data is arranged in SoA format. In that case, you should be able to map the two components of your complex vector without problems (with two map calls). If the data is interleaved, you will need to gather each component.

[gergol]

So here is how I try to do it:

using namespace enoki;
{
  using FloatP   = Packet<float, 8>;
  using FloatX  = DynamicArray<FloatP>;
  using ComplexX = Complex<FloatX>;

  std::vector<float> interleaved;

  auto cplx = ComplexX::map(interleaved.data(), 10);

}

However, the last line fails to compile with error: no member named 'map' in enoki::Complex<enoki::DynamicArray<enoki::Packet<float, 8, true, enoki::RoundingMode::Default> >, true>'; did you mean 'max' , because Complex inherits from StaticArray.

[wjakob]

As I said -- you need a map per component (FloatX::map) or two FloatX gathers in the interleaved case.

[gergol]

OK, I see. I was thinking there was a way to do something similar to a reinterpret_cast<complex<float>>. Thanks for your help!