Hello!

I've come to Eigen after jumping the CBLAS boat. I'm just getting some practice now. I'm currently trying to use write a function which does a change of basis on a Hermitian matrix. Is the following the most efficient way (runtime wise) to do it using the Eigen library?

Code: Select all

template <int N> using matrix = Eigen::Matrix<std::complex<double>, N, N>;
template <int N> void basis_change(matrix<N> &rho, const matrix<N> &U)
{
  matrix<N> rho_prime;
  rho_prime.template triangularView<Eigen::Upper>() =
          U * rho.template selfadjointView<Eigen::Upper>() * U.adjoint();
  rho = rho_prime;
}


Thanks in advance!

Sounds good. Note that if N is not too small, like greater than 16, then there is no real benefits of specifying the sizes at compile-time. Also, note that Eigen is very slow when compiling without compiler optimization.