Features of computational processes based on SM – transformation

Abstract

The features and functionality of processing of the one – dimensional (vector) arrays of numerical data by the use of computational method with the formation of difference slices are considered. At the base of this method SM – transformation principles are used. The allocation of the minimum of non-zero component of the element array in this case is considered as the formation of the internal processing threshold and is the basic procedure of SM - transformation in each processing cycle. As a result, not only the operation of parallel multi-operand summation of the number array is realized, but also there is an opportunity to restore the initial number array, as well as to sort its elements according to the growth of their numerical values and to form their ranks. Two matrices of binary masks are used for this, that are formed during the processing, which are the matrices of zero and positive criterion, respectively, inherent in the elements of the current difference  slices. In addition, in each processing cycle the internal thresholds form a vector of internal thresholds as a result, which takes part in restoring the elements of the initial array. The it is presented in the article the basic relations of difference slice processing, and the examples that confirm their validity using data presented in the form of a table.