JPH11296348A - Natural random number and hybrid random number generating device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To freely use three kinds of random numbers such as a natural random number, hybrid random number, and pseudo random number by attaining the reproducibility test of a calculation formula by using an arithmetic random number having periodicity. SOLUTION: A natural random number generating device generates a natural random number at a high speed in a system for generating a random pulse signal by detecting particles emitted at random by naturally collapsing atomic nuclei, and measuring a time between the signals of a random pulse generator 1. A hybrid random number generating device generates a highly precise hybrid random number by repeatedly fetching the natural random number as the initial value of an arithmetic formula more quickly than the appearance of the periodicity of an arithmetic random number.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a random number generating device, and more particularly to a natural random number generating device which detects alpha rays emitted by the decay of atomic nuclei and generates a true random number by utilizing the fact that the decay time is random. Combining the device and a hybrid random number generator that substitutes this natural random number for the initial value of the operation to generate a random number by the operation, a natural random number that does not require a random number test and a highly accurate hybrid random number that is close to the limit of the natural random number Related to a device that generates at high speed.

[0002]

2. Description of the Related Art Several proposals have already been made on the generation of physical random numbers utilizing the fact that the decay phenomenon of a nuclear nucleus is a random phenomenon. For example, JP-A-60-316
No. 40 and JP-A-6-154411 by the present inventors.
And JP-A-9-50369. However, although the conventional physical random number generator can generate a true random number, the speed required today (several hundred nanoseconds) is required.
c) In order to generate a random number, a high-concentration radiation source of several curies (Ci) is required, which causes radiation damage to the detector, which is not practical or has a problem in safety management.

The arithmetic random numbers generated by the arithmetic processing have been theoretically elucidated for a long time, and various methods are widely used. One example of the invention is disclosed in
No. 23669 and the like. However, in a random number generation method using a computer, even though it is possible to generate random numbers at a high speed, random numbers are artificially generated, so they are pseudo-random numbers and cannot be true random numbers. There is a drawback of manifestation. On the other hand, the defect of periodicity (reproducibility) of operation random numbers is used for reproducibility testing of simulation formulas.

[0004]

[Problems to be Solved by the Invention]
Using about one-hundredth of a radiation source,
Generates true natural random numbers that need not be verified at 10,000 times the speed. Also, a high-precision random number that is close to the true random number,
Generate with the speed of the latest high-speed computer. On the other hand, it is also possible to generate the random numbers required for the reproducibility test of the simulation formula. In the present application, a random number generation device that satisfies all the conditions currently required in random number generation is completed.

[0005]

According to the first aspect of the present invention, a decay time interval at which radioactive nuclei randomly emit α rays is measured by a PIN diode detector, which is one time smaller than the conventional method.
Generates natural numbers of 32 digits (32 bits) at a speed of 10,000 times. According to the invention of claim 2, the measurement of the time interval is performed by:
By performing the processing with an accuracy of 3000 to 1/5000 of the average value of the time intervals, a natural random number having an extremely uniform appearance probability of the random number is generated.

According to a third aspect of the present invention, the generated natural random number is
Within the time when the periodicity of the random number calculator appears, a random number calculation is performed by repeatedly substituting as an initial value of the calculation, and a high-precision hybrid random number close to the limit of natural random numbers is generated at the latest computer speed.

According to the fourth aspect of the invention, there are three types of natural random numbers that do not require verification, hybrid random numbers that have been verified to be extremely close to natural random numbers, and pseudo-random numbers having periodicity generated by an arithmetic unit. Make random numbers freely available to users as needed.

[0008]

BEST MODE FOR CARRYING OUT THE INVENTION A natural or artificial radioactive nucleus emits α, β, γ-rays and the like one after another and naturally decays.
Since a group of nuclei is a collection of nuclei that are independent of each other, the individual decay phenomena are independent and random phenomena. The decay phenomena of these nuclei and the probability distribution of the decay have been studied for a long time and are known facts.

[0009] Assuming that the time interval between the emission of one nucleus and the next emission of one particle is t, the time interval t
The frequency of the emitted particles is expressed by the ratio (t / T0) between the time interval t and the average value T0 of the time intervals as shown in the following equation (1).
Can be expressed by an exponential function of dN / dt = Aexp (-t / T0) ----------------- (1) Here, the average time T0 is the average number of particles emitted per second < N0 is the reciprocal of N> 1 / <N> seconds. Since individual nuclei decay independently and randomly, each decay time t has a random value. Therefore, one random number can be generated from one measured value of the random time interval t.

In the conventional method, the number of particles radiated in a certain period of time is measured, and one natural random number is generated from the number of particles. Therefore, in order to generate one random number, several thousand count values are required. Met. In the present invention, one random number is generated from one measured value of the time interval t, so that compared to the conventional method,
Radiation intensity can be reduced from 1/1000 to 1 / 10,000. When a source having a legally permissible dose of 100 micro-Curie (μCi) is used as a source intensity, one natural random number of 32 digits (32 bits) can be generated in 500 nanoseconds.

Increasing the intensity of the radiation dose to increase the generation speed of natural random numbers has a safety limit. Therefore, a computer is used to generate random numbers at a higher speed. However, the appearance of periodicity is inevitable in pseudorandom numbers generated by computer operation. Accordingly, the random number calculation is performed by repeatedly assigning the generated natural random number as the initial value of the calculation expression earlier than the time when the cycle appears in the random number calculation.
As a result, high-precision hybrid random numbers close to the limit of natural random numbers are generated at the speed of the computer.

The configuration of an embodiment based on the above principle will be described below. FIG. 1 shows the overall configuration of a natural random number and hybrid random number generation device. The random number generation device includes four random pulse generators 1, four time measuring devices 2, four clock pulse generators 3, a random number storage 4, a random number calculator 5, and a random number type selector 6. Is done.

The details of the random pulse generator 1 are shown in FIG. The shutter 11 is opened and the PIN diode detector 12 is irradiated with α-rays from the α-ray source 10. The small current from the PIN diode detector 12 becomes a weak voltage signal by the preamplifier 13 and is amplified about 100 times by the main amplifier 14 to become a voltage signal of several volts which is easy to detect. The voltage signal from the main amplifier 14 is α
Only the voltage signal by the line is discriminated, and the waveform shaper 16 shapes the signal into a rectangular waveform.

The time interval between the signal from the random pulse generator 1 and the next signal is measured by the time measuring device 2. The time interval is measured in units of the time width of the pulse generated by the clock pulse generator 3 and the number of pulses is measured by an 8-bit (256) counter.

The time exceeding 8 bits is repeatedly calculated with 8 bits and is represented by the remaining measurement number. FIG. 3 shows the concept of a method of generating a natural random number by measuring a time interval.
As shown in FIG. 3, the time interval between signals due to α rays is 8 (00001000), 42
(00101010), 3 (00000011), 27 (00011011), 28+
11 (00001011),... And 1 to 256 (100000000).

In order to make the appearance probabilities of the random numbers uniform, measurement is made with a clock pulse having a pulse width (frequency) that is 1/3000 to 1/5000 of the average value T0 of the time intervals. FIG. 4 is a frequency distribution of time intervals t of decay particles when using a source of <N> = 200 cps. FIG. 4 shows an exponential distribution having an average time interval T0 = 5 ms. When this <N> = 200 cps source collapse phenomenon is read out with an 8-bit counter using a clock pulse (frequency 1 MHz) with a time width of 1 microsecond, which is 1/5000 of T0 = 5 ms. Is shown in FIG.
As shown in FIG. 5, the frequency distribution is uniform.

The random pulse generator 1, the time measuring device 2, and the clock pulse generator 3 each have four (channels), each of which is measured by an 8-bit counter.
Since there are 4 channels of 8-bit counters, 3
It is a 2-bit (32-digit binary number) random number. These random numbers are sent to the random number storage 4.

Taking the latest scientific calculation (simulation) as an example, every three to several tens of microseconds, 3
One 2-bit random number (4 bytes) is used. 10
A 24-hour calculation using one random number every 0 microseconds results in a storage capacity of about 3.5 GB.

When the capacity of the random number storage device 4 becomes available using the natural random number of the random number storage device 4, the shutter 11 automatically opens, and natural random numbers are generated and stored until the capacity is full. . In other words, after using the random number generation device, random numbers are automatically accumulated during a pause period until the next use. α
The dose of the radiation source 10 and the capacity of the random number storage 4 can be freely determined in accordance with the use situation such as the calculation time and the use frequency of the user.

When a high-speed and high-precision random number close to the limit of natural random numbers is required, a hybrid random number 8 is used. The hybrid random number 8 is generated by repeatedly taking in the natural random number stored in the random number storage 4 as an initial value of an arithmetic expression within a period in which the periodicity of the random number calculator 5 appears, and generating the same.

As the random number calculation expression, any existing calculation random number package may be used as long as the calculation speed is high in principle. Theoretical elucidation has been performed for many years, and the linear congruential method and the M-sequence random number generation method are widely used. However, there is an inconsistent and conflicting relationship between what is theoretically possible and true random numbers. Therefore, in this application, results are prioritized over theoretical elucidation. The linear congruential method generates a random number sequence using a recurrence formula of degree 1. However, as long as the order is 1, it is inevitable that the generated random numbers have a multidimensional sparse crystal structure. Therefore, an M-sequence random number method having a high degree and a large number of terms will be described as an example.

1 of the M-sequence random number arithmetic expression (Toesworth method)
FIG. 6 shows a comparison of a hybrid random number generated by repeatedly substituting natural random numbers into the 00 term, a natural random number, and an M-sequence pseudo random number. Various random number test methods have been proposed, but no difference appears in a textbook test method. Here, FIG. 6 shows a result of a test performed by a method called a spacing test by Marsalia (G. Marsalia). This compares the frequency at which the same random number appears repeatedly with the Poisson distribution. As shown in FIG. 6, the hybrid random number can be said to be a high-precision random number close to the limit of a true random number, so that there is no problem in actual use.

On the other hand, when a pseudo-random number having periodicity is required for the reproducibility test of the simulation calculation formula, an arithmetic random number is generated using only the random number calculator 5.

By selecting the type of random number by the random number type selector 6, natural random number 7, hybrid random number 8,
And three kinds of random numbers such as the operation random number 9 and the like, are taken out and used as required by the user. FIG. 7 summarizes the characteristics of these three types of random numbers.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an overall configuration of a natural random number and hybrid generator according to the present invention.

FIG. 2 shows a random pulse generator which forms part of a natural random number and hybrid generator.

FIG. 3 is a conceptual diagram showing a method of generating natural random numbers by measuring time intervals.

FIG. 4 is an experimental diagram showing that the frequency distribution of particles having a time interval t is an exponential function.

FIG. 5 is an experimental view showing an exponential distribution when a time interval is read by an 8-bit counter.

FIG. 6 is a test chart showing a comparison of the accuracy of natural random numbers, hybrid random numbers, and pseudo random numbers.

FIG. 7 is a diagram summarizing features and purposes of three types of random numbers.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Random pulse generator 2 Time measuring device 3 Clock pulse generator 4 Random number storage 5 Random number calculator 6 Random number type selector 7 Natural random number 8 Hybrid random number 9 Arithmetic random number 10 Alpha ray source 11 Shutter 12 PIN diode detector 13 Previous Amplifier 14 Main amplifier 15 Wave height discriminator 16 Waveform shaper

Claims (4)

[Claims] Based on the fact that the time at which an atomic nucleus spontaneously decays and emits α-rays is random, a time interval at which α-rays are emitted is detected by a PIN diode detector, and natural time which does not require an assay is detected. A natural random number generation device that generates random numbers at a speed 10,000 times faster than a conventional physical random number generation method. 2. The first time interval is set to 300 of the average value of the time intervals.
The natural random number generation device according to claim 1, wherein uniform natural random numbers can be generated by measuring with an accuracy of 0 to 1/5000. 3. A high-precision hybrid close to the limit of natural random numbers by performing a random number operation by repeatedly assigning the natural random number as an initial value of the operation earlier than the time when the periodicity of the random number calculator appears. A hybrid random number generator, wherein random numbers are generated at the latest computer speed. 4. The natural random number generation device according to claim 2 and the hybrid random number generation device according to claim 3 are integrated, and three types of the natural random number, the hybrid random number, and the pseudo random number by the random number calculator are required. A high-speed natural random number and hybrid random number generation device characterized in that the device can be used freely according to the conditions.