Calculating measurement uncertainty for complex-valued quantities

A software technique is described that provides support for uncertainty calculations for complex-valued measurements. The technique is based on classical methods of multivariate statistics and propagation of variance as well as automatic differentiation, which is an established computational method. It facilitates propagation of covariance data by automating the calculation of sensitivity coefficients. The technique introduces a simple abstraction to represent measurement data and defines arithmetic operations and standard functions that manipulate this abstraction. The method is not difficult to implement and is easy to use. An example source code in C++ is included.