Recent developments in classical density modification

Classical density-modification techniques (as opposed to statistical approaches) offer a computationally cheap method for improving phase estimates in order to provide a good electron-density map for model building. The rise of statistical methods has lead to a shift in focus away from the classical approaches; as a result, some recent developments have not made their way into classical density-modification software. This paper describes the application of some recent techniques, including most importantly the use of prior phase information in the likelihood estimation of phase errors within a classical density-modification framework. The resulting software gives significantly better results than comparable classical methods, while remaining nearly two orders of magnitude faster than statistical methods.