THE VOIGT AND COMPLEX ERROR FUNCTION - A COMPARISON OF COMPUTATIONAL METHODS

Several computational procedures for the Voigt function and complex error function are discussed and compared with respect to accuracy and running time. Vectorization of the codes is applied where possible. Computational speed varied over two orders of magnitude. Even without vectorization, restructuring of the source code can yield a significant acceleration. The computational effort for Fourier transform methods is estimated and compared with other methods. For applications involving least-squares-fitting, the evaluation of the complex error function provides an efficient way to calculate both the Voigt function and its partial derivatives.