This module is to be used when one has a set of experimentally determined phases and it is has been decided that the model should be restrained to this information. The function minimized is
Where and is the figure of merit of the phase.
It has been found that the gradient of the residual for this term can be calculated using an algorithm similar to Agarwal's method for calculating the gradient of the RFACTOR term. The same convolution is performed but the coefficients used to calculate the map ( ) are described below.
As you can see, it is not at all clear from the coefficients what the meaning of the map is. However, because the gradient of the phase residual is calculated from this map using Agarwal's convolution, we know that this map should behave like a normal difference map. There will be positive or negative density centered on an atom if its occupancy or temperature factor is in error and there will be an asymmetric, plus-minus, peak if the position of the atom is incorrect. Because the sources of noise in this kind of map are quite different from those in the usual difference maps, it could very well be profitable to directly examine a map of this form to aid in the interpretation of a model in the early stages of structure determination.
Because of the similarity between the calculations required for this module and those required for the RFACTOR module there are no independent programs for this module. The program Rfactor will calculate all the required values.