JPH11249872A - Method for generating random number and its system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To generate random numbers without periodicity and regularity at a high speed without requiring a physical trial by receiving a data piece to be a base for generating the random numbers from a data source consisting of multiple computers, executing a prescribed data processing based on the data piece and generating the random numbers. SOLUTION: A data source 1 being a system constituted of a computer 11 or multiple computers connected by a communication network 12 supplies a data piece to be a base for generating random numbers. A system 2 for generating the random numbers is provided with a receiving device 21 for fetching the data piece from the data source 1, an arithmetic processor 22 for generating the random numbers to be a target by executing a fixed data processing based on the fetched data piece and a storage device 23 for storing the obtained random numbers. Then, proper data is taken-out from the data source 1 being a huge information system and fetched data is worked and processed in the arithmetic processor 22 so that the random numbers without the occurrence of periodicity is generated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a random number generating method, and more particularly to a random number generating method and system for generating a random number sequence having no periodicity and regularity.

[0002]

2. Description of the Related Art Conventionally, as a numerical calculation method for generating a random number, many methods for obtaining a pseudo random number using a recurrence formula such as a linear congruential method have been proposed. Generation of random numbers based on these methods can be performed at high speed by using a computer.

[0003]

However, the disadvantages of the pseudo-random numbers obtained by these methods are that they have periodicity, and in the case of the linear congruential method, the regularity of a multidimensional sparse crystal structure appears. Properties.

On the other hand, as a method of generating a pseudo-random number having a finite number of bits having no periodicity, a method that has been used for a long time is to use a coin or dice as a means to generate a 2-bit or 6-bit random number. To be generated. These methods apply the principle of mechanical motion.

That is, an object given kinetic energy gradually loses kinetic energy due to the action of frictional force and aerodynamic force in the process of performing physical movement, and eventually comes to rest. It uses the property of being accidental without reproducibility.

In these methods, it is assumed that the following two properties hold.

[0007] 1. The result of each trial is defined by a finite number of degrees of freedom (number of events). For example, as a stationary state, in the case of a coin, either the front or the back side always faces upward, and in the case of a dice, any one of six sides in which the numbers from 1 to 6 are written The face always faces up,
That's what it means.

[0008] 2. The above 1. The occurrence of each event allegedly occurring in is also expected with certainty.

The method of generating random numbers using such a dynamic motion has the disadvantage that it takes time for each trial, and is not suitable for generating a large number of random numbers at high speed.

Accordingly, the present invention has been made in view of the above problems, and has as its object to provide a random number generation method for generating random numbers having no periodicity and regularity at high speed without the need for physical trials. It is to provide a system.

As a publication related to the present invention described below, for example, Japanese Patent Application Laid-Open No. 62-406148 discloses a random number generating circuit for generating a random number from unspecified data communicated on a bus line. Japanese Patent Laid-Open No. 7-36672 discloses a configuration of a random number generator that generates a random number using a plurality of shift registers.

[0012]

According to the present invention, there is provided a data fragment for generating a random number in a data source comprising a computer or a system comprising a number of computers interconnected by a communication network. And performs predetermined data processing based on the fetched data fragments to generate a desired random number.

[0013]

DESCRIPTION OF THE PREFERRED EMBODIMENTS A preferred embodiment of the present invention will be described. The present invention is to generate random numbers having no periodicity at high speed using a communication network, that is, a large-scale system in which computers and terminals are connected by lines. More specifically, the random number generation method of the present invention, in a preferred embodiment thereof, converts various data fragments processed in a system (aggregate) composed of computers and a large number of computers connected by a communication network. By utilizing the inherent uncertainty and obtaining numerical data based on these data fragments, pseudorandom numbers with no periodicity are generated at high speed.

In a preferred embodiment of the random number generating method of the present invention, referring to FIG. 1, a data source 1 which is a computer or a system composed of a large number of computers connected by a communication network, Supply the data fragment from which to create. The system 2 for generating random numbers includes a receiving device 21 for capturing a data fragment from the data source 1 and an arithmetic processing device 22 for generating a desired random number by executing a predetermined data process based on the captured data fragment.
And a storage device 23 for storing the obtained random numbers.

For example, the data source 1 is composed of computers 11 connected by a communication network 12, and includes a large number of computers operating as mail servers. It is assumed that transmission is possible.

The total data amount of mail transmitted from each mail server is monitored at regular time intervals in units of information such as bytes. If the amount is even, 0 is corresponded, and if the amount is odd, 1 is corresponded. , To the receiving device 21. Thus, one 0 or 1 value data is transmitted from each mail server to the receiving device 21 at regular time intervals,
The receiving device 21 receives 0-1 value data in the order of reception, and generates a data string.

The data sequence transmitted from one mail server itself has disorder, and the order in which the data transmitted from each mail server arrives at the receiving device 21 becomes random due to the fluctuation of the communication load. .

Further, the ratio of the number of appearances of 0 and 1 in the 0-1 value data string generated in the receiving apparatus 21 and the bias seen in the variation are sufficiently large as the data source 1 and are numerous. When considering a system including a number of mail servers, it can be regarded as non-artificial.
This makes it possible to form a uniform 2-bit random number sequence that does not show periodicity.

As in this example, by extracting appropriate data from the data source 1 as a huge information system, and as will be described below as an embodiment, the extracted data is further processed as needed by an arithmetic processing unit. By performing the processing at 22, a random number can be generated.

[0020]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of the present invention;

[Embodiment 1] In this embodiment, a large-scale aggregate (system) composed of a computer and terminals connected via a communication line is considered. As a system, as shown in FIG. 2A, a data source 1 including a plurality of computers P1 to Pn
And a configuration including a random number generation system 2 connected to each of the computers P1 to Pn of the data source 1, or a plurality of terminals Pi (i = 1, 2,...) As shown in FIG. Each of the computers Cj (j = 1, 2,...) Can be considered as a system connected to each other via a communication line. As a large-scale system having the form shown in FIG. 2B, the Internet is a typical example. In this case, since the existing system is used as it is, and the number of computers connected to the Internet is enormous throughout the world, it is suitable as a data source for generating random numbers.

The operation of one embodiment of the present invention will be described. On each computer, a state variable function relating to (part of) the operating state of the system, which is considered to be not based on the operation, is constructed in an appropriate manner.

In one embodiment of the present invention, as an example of the state variable function, a total value obtained by accumulating the sum of the character lengths of the messages transferred from each terminal to one computer per unit time is used.

These values obtained from each computer are sequentially transferred to the receiving device 21 of the random number generation system 2 (see FIG. 1).

The data transmitted in this manner becomes a periodic sequence when the computers which are the transmission sources are arranged in the data transmission order in terms of the transmission time. However, regarding the reception time, such a regularity is not generally exhibited even if the computers as transmission sources are arranged in the order of data reception due to variation in transfer time.

However, the number of data sent to the arithmetic processing unit is smaller than the unit time for counting the number of characters of a message because the scale of the system is small, and the time required to transfer the total data from each computer to the arithmetic processing unit is small. If the bias between individual computers and the fluctuations of each transfer are small enough to be ignored, there is a possibility that some degree of regularity will appear in the order in which the aggregated data arrives. For example, many servers on the WWW (World Wide Web) Considering a sufficiently large system, such as receiving data from them, the order in which data arrives can be considered to be random.

For each element of the numerical data sequence collected in random order in this way, if the value is odd, 0 is set,
If it is an even number, processing for returning 1 is performed in the arithmetic processing unit 22. With the above method, a 2-bit random number sequence can be generated.

Further, as shown in FIG.
In FIG. 3, the data sent to the receiving device 21 (see FIG. 1) is represented by a binary integer, and a sequence formed by connecting the data in the receiving order is divided into L (L> 2) columns (in FIG. 3, L = 2).
7) By performing a process of obtaining a binary decimal, 0.0 to
It is possible to generate a real random number of 1.0 section as an array.

However, when the data is expressed as a binary integer, the most significant digit is always 1, and the expected appearance frequencies of 0 and 1 are not equal (the appearance frequency of 1 is higher). In order to avoid this, the number of digits is unified by supplementing 0 as necessary so as to be the smallest representable even or odd number of digits. In FIG. 3, the numbers are unified to even-numbered digits.

Embodiment 2 As another embodiment of the present invention,
Another example of the state variable function of the system used to generate random numbers will be described.

In this embodiment, a multi-user environment in which UNIX is an OS (operating system) is assumed. Consider a function defined for each time, which takes the value 0 when the number of logged-in users displayed as the execution result of the “who” command is odd and takes the value 1 when it is even. Further, a sequence of appropriately sampled time values, t0, t1, t2,. . . . Is selected, and a sequence of values of the state variable function obtained at these times is used. As a sequence of time values to be sampled, for example, a method of using pseudorandom numbers or taking a sequence of times having a certain time interval is used.

However, when the above-mentioned state variable function is adopted, if it is known that the number of logged-in users in a specific time zone (of a day) is normally determined, If the time value is sampled without distinction between the time zone and the other time zones, the data value collected in the time zone will appear stochastically. In order to avoid this, it is better to sample the time value except for the time zone. That is, it is preferable to take measures to minimize the possibility that the sequence generated from the state variable function on one computer has an empirically clear correlation.

In this way, each computer transmits the sequence data generated by itself to the totaling processing device (corresponding to the random number generation system 2 in FIG. 1) prepared in advance by an arbitrary length.

The tallying processor reads the transmitted data in an appropriate order and arranges it as a one-dimensional data string. Thus, a pseudo-random number sequence can be generated. However, the following restrictions are set when reading data.

An upper limit is set for the amount of data or connection time that can be continuously received from one transmission source. When the upper limit is reached, data reception from the current data transmission source is interrupted and another transmission is performed. Start reading the original data again. This process is repeated.

The order of determining the receiving destination of the data to be read is not intentionally managed, but follows the order that is accidentally determined from the timing of transmission and reception on the communication network.
However, in this case, it is assumed that the probability of connection with a specific computer does not particularly increase due to a communication network factor. As a result, the correlation observed in the order of access to data from each computer can be sufficiently ignored.

That is, assuming that the total number of computers connected on the communication network is N, and they are given identification numbers from 1 to N to distinguish them, the corresponding An infinite sequence M (1), M (2),. . . . Can be regarded as being substantially uniformly distributed.

The 0-1 value data sequence generated on the tabulation processing apparatus by such a method is a part of the data sequence sent from a specific computer when viewed alone. However, even if non-uniformity is seen in the data sequence, the effect is negligible as a data string of infinite length after aggregation.

This tendency increases as the size N of the aggregate of computers connected by the communication network increases, and the reliability as a uniform random number increases.

By summarizing the data strings obtained from a plurality of computers in this way, the 0-1 value data string generated by a specific computer by such a method can be used for each computer which cannot be artificially predicted. It utilizes the above disorder of the operating state and the disorder of the communication on the communication network. Therefore, unlike a pseudorandom number generated from a numerical method, it does not assume a recurrence formula, and thus has no regularity or periodicity.

[Embodiment 3] As still another embodiment for generating random numbers, the sequence M
(1), M (2),. . . . Itself may be considered as an N-valued pseudo-random number sequence and used. However, in order to guarantee the reliability of this sequence as a random number, the uniformity of the timing of data communication on the communication network assumed in the condition (2) must be maintained. This condition is considered to be satisfied if the data transfer time to the aggregation processing device varies greatly.

[Embodiment 4] FIG. 4 is a diagram showing a configuration of a fourth embodiment of the present invention. In FIG. 4, first, a uniform random number sequence is generated in the arithmetic processing unit 22 (FIG. 1) by the means described in the embodiment. The storage device 23 (FIG. 1) stores the generated uniform random number sequence.

Next, the arithmetic processing unit 22 (FIG. 1) refers to the uniform random number sequence in the storage device, executes random number generation means based on an inverse function method, a binary selection method, or another known algorithm, and Generate various random number sequences that follow a distribution or a discrete distribution.

This embodiment has an effect that, after a random number according to a certain distribution is generated in the arithmetic processing unit, another random number sequence can be continuously generated.

[Embodiment 5] In the first embodiment, when a 2-bit random number sequence is generated, 0-1
Instead of calculating the value data string, each computer does not accumulate the number of characters in the message, but indicates 0-1 indicating the odd / evenness of the message.
The data, that is, the value of the last place when the number of characters is represented in a binary number may be transmitted. In this case, the processing intensively performed by the arithmetic processing device in the first embodiment is now distributed to the respective computers, and the efficiency is high.

[0046]

As described above, according to the present invention,
There is an effect that a random number sequence having no periodicity and regularity can be generated at high speed without physical trial. As a result, it is possible to provide random numbers that are required for practical use such as a large-scale simulation using the Monte Carlo method and that are less biased in sample variation.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of an embodiment of the present invention.

FIG. 2 is a diagram showing a configuration of a first exemplary embodiment of the present invention.

FIG. 3 is a diagram for explaining a second and a third embodiment of the present invention.

FIG. 4 is a diagram showing a configuration of a fourth embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Data source 2 Random number generation system 11 Computer 12 Communication network 21 Receiving device 22 Operation processing device 23 Storage device

Claims (11)

[Claims] 1. A data source comprising a computer or a system composed of a large number of computers interconnected by a communication network, generates a data fragment from which a random number is generated, and a predetermined data fragment is generated based on the data fragment. A random number generating method, which performs data processing to generate a target random number. 2. A computer comprising a plurality of computers interconnected by a communication network as a data source, including a plurality of computers operated as a mail server, wherein a plurality of computers operated as a mail server are transmitted at predetermined time intervals. The data amount is monitored in a predetermined information amount unit, and values 0 and 1 are made to correspond according to even / odd of the numerical value representing the data amount. Is transmitted, and a system for receiving the data generates a random number by generating a data sequence from 0 or 1 received in the order received from the mail server. 3. In a system comprising a plurality of computers, each of the computers constituting the system transmits numerical data representing a state relating to an operating state of the system which fluctuates with time, and the system receives the data. A random number generation method characterized by generating a 2-bit random number sequence by associating 0 and 1 according to the evenness of the data in the order of reception. 4. The random number generator according to claim 3, wherein the data is a total value obtained by accumulating the total length of message characters transferred from the terminal connected to the computer to the computer per unit time. Method. 5. A system for transmitting numerical data representing a state relating to an operating state of a system which fluctuates with time in a system including a plurality of computers from a computer constituting the system, and receiving the data, A numerical sequence formed by connecting the binary integer representations in the order of reception is divided into a predetermined number of digits L, and an L-bit binary decimal number is obtained from the separated data, so that a section from 0.0 to 1.0 is obtained. And generating a real random number. 6. The random number generator according to claim 5, wherein a leading zero of said numerical value sequence is supplemented as necessary so that said number of digits becomes the smallest representable even or odd number of digits. Method. 7. In a system including a plurality of computers, a system for transmitting sequence data generated by each computer at an arbitrary length to an aggregation processing device and receiving the sequence data is transmitted from each of the computers. A random number generation method, wherein a pseudo-random number sequence is obtained by reading data in a predetermined order and arranging it as a one-dimensional data sequence. 8. An identification number from 1 to N is assigned to the computer, and a sequence in which the identification numbers of the corresponding computers are arranged according to the order of the data read by the processing device for summing up is a pseudorandom number of N values. The random number generation method according to claim 7, wherein 9. A random number generating method, comprising: performing a predetermined operation on a uniform random number generated by the random number generating method according to claim 1 to obtain a desired random number. 10. A data source comprising a computer or a system composed of a number of computers interconnected by a communication network, and a receiving means for receiving a data fragment from which a random number transmitted from the data source is generated. An arithmetic processing means for executing predetermined data processing as needed based on the captured data fragments to generate a desired random number; and a storage means for storing the generated random number. Random number generation system. 11. The random number generation system according to claim 10, wherein a predetermined random number is obtained by further performing a predetermined operation on said random number stored in said storage means.