rotor_bank()

MAYAFLUX_API double MayaFlux::Kinesis::Discrete::rotor_bank ( std::span< const double >  window, std::span< double >  coefs  )

noexcept

Sum of N sinusoids advanced by complex rotation instead of a sine call.

Each call rotates every unit phasor by its increment and returns the sum of scale * imaginary part, which is sin(phase) after the advance. A first order correction keeps the phasors on the unit circle, so the amplitude does not drift over long runs. Cheaper than phasor_bank per sinusoid, with no slow amplitude swell: scale the array's scales column for that.

Definition at line 219 of file Kernels.cpp.

{
    if (coefs.empty())
        return 0.0;
 
    const auto n = static_cast<size_t>(std::max(coefs.front(), 0.0));
    if (n == 0 || coefs.size() < 1 + 5 * n)
        return 0.0;
 
    const auto cos_step = coefs.subspan(1, n);
    const auto sin_step = coefs.subspan(1 + n, n);
    const auto scales = coefs.subspan(1 + 2 * n, n);
    const auto re = coefs.subspan(1 + 3 * n, n);
    const auto im = coefs.subspan(1 + 4 * n, n);
 
    rotate_rotors(cos_step.data(), sin_step.data(), re.data(), im.data(), n);
    return weighted_sum(scales, im);
}

References weighted_sum().

Here is the call graph for this function: