SU397923A1 - STOCHASTIC FUNCTIONAL CONVERTER - Google Patents

Description

one

The proposed device relates to the field of digital computing and is intended for learning random numbers with given distribution laws.

A stochastic functional converter is known for producing random numbers with a piecewise linear distribution density. It contains a generator of uniformly distributed random numbers, connected via an addition multiplier, the corresponding AND circuit and OR to the first input of the output register, and through a selection block to the input of a memory block, one output of which is connected to the corresponding input of the selector , the outputs of a probabilistic binary element are connected to the corresponding inputs of the “I.

However, in such a converter, to obtain each output number, the generator should be uniformly distributed in the interval of 0.1 random numbers to produce four numbers that undergo various transformations.

The purpose of the invention is to simplify the operation of the converter.

In the proposed stochastic functional converter, this goal is achieved by introducing an increment adder and using a probabilistic binary 0.5 element, one output of which is connected to the input of a single-digit increment adder, the second input of which is connected to the memory unit and the output to the output register. The drawing is a diagram of the proposed

converter

It contains a generator of random numbers 1 uniformly distributed in the interval of 0.1, interval selection block 2, memory block 3, one-bit increment accumulator 4, binary 0.5 probability — element 5, multiplication-computation block; o, schemes “Ii 7, 8, scheme“ OR 9: i output register W.

The operation of the stochastic functionary of the smallest converter is based on the operation of stochastic slithering over random variables with triangular Simpson distributions, which corresponds to the stochastic realization of the piecewise linear approximation of the distribution density with uniformly:. splitting argument when

X consi.

In the case of a stochastic functional converter, the algorithm is implemented by random sampling with probabilistic PJ triangular distributions, which are realized by 0.5 superpositioning and triangular distributions obtained by performing multiplication and addition of distribution functions. In order to carry out a random sampling of triangular distributions in the stochastic functional transducer, an interval selection block 2 is provided, with one input and output connected to memory block 3, and another input with a generator uniformly distributed in the interval of 0.1 random numbers /. The memory block 3 stores the values of the distribution function F (Xj), calculated from the probabilities PJ, and the boundary points Xj of the intervals (Xj, Xjl. I). To perform multiplication operations, add distribution functions, use the multiplication block of distribution functions 6, which is based on algorithms for dividing the input numbers into large ones (multiplication operations of distribution functions) and smaller ones (operations of adding distribution functions). The stochastic functional converter under consideration uses a uniform partition of the argument when, where m is an integer. In this case, the simplest technical implementation is obtained, and the 0.5-superposition of triangular distributions is carried out by introducing a probabilistic binary O. element 5 and adder of one-bit increments 4 into the stochastic functional converter. The stochastic functional converter works as follows. The generator 7 outputs three numbers, uniformly distributed in the range of 0.1 numerical sequence. One of them comes to interval selection block 2, where it is compared with the values of the distribution function F (Xj). As a result, the interval Xj, Xj + 2 is selected and the value X-, from the memory unit 3 enters the one-digit increment adder 4, where it is added to the single-digit increment of the unit output of the probabilistic binary 0.5-element 5. The result of this sum Xj or Xj +. Aj.L 1 is written to the high bits of the output register 10. The other two numbers from the generator / go to block 6, where they are divided into larger and smaller ones. Through controlled zero and unit outputs of the probabilistic binary 0.5-element of the circuit "And 7 or 8, respectively, the greater or lesser number through the circuit" OR 9 is written into the low bits of the output register 10. The number sequence, density the distribution of which, within the limits of the accuracy of approximation, corresponds to a given one. The considered stochastic functional converter is simple, since there is no need to store in the memory block the coefficients that specify the required angle of inclination of the approximate right in each section. This makes it possible to reduce memory by a factor of three. In addition, for each output number, the converter uses three (instead of four) numbers from the generator of uniformly distributed numbers, which makes it possible to reduce the amount of equipment or increase its speed. The invention is a stochastic functional converter containing a generator of uniformly distributed random numbers, one output of which is connected to the first input of the selection unit, two other outputs are connected to the multiplication-complex block, the outputs of which are connected to the first input of the output register , the output of the selector unit is connected to the input of the memory unit, the first output of which is connected to the second input of the selector unit, the second I inputs of the And circuit connected to the corresponding outputs are likely ostnogo binary element, characterized in that, in order to simplify the operation of the apparatus, it comprises an adder increments, the output of which is connected to the second input of the output register, the first input - to the second output of the memory unit, the second - to one output probability tnostnogo binary element.