The use of the non-negativity criterion and histogram matching as figures of merit in direct methods of phase determination

Two figures of merit for the selection of correct phase sets from a number of possible or trial phase sets are defined: (1) based on the criterion that significantly negative points should be sparse in electron-density maps; and (2) based on comparison of electron-density histograms with the theoretically expected histogram [Lunin (1993). Acta Cryst. D49, 90-99]. It is shown that both figures of merit are useful for judging random phase sets and useless for phase sets that originate from direct-methods procedures such as symbolic addition or tangent refinement.