JPS60215279A - Molecule movement simulator - Google Patents

Abstract

PURPOSE:To increase the processing speed by carrying out the multiplication and addition/subtraction to decide whether a molecule exists at a place near a boundary as well as to perform the correction processing and therefore making use of the so-called pipeline processing of a vector processor. CONSTITUTION:A control part 10 sets the constants on a memory 11 to prescribe the length of a side of a cube of an area to be simulated, the total number of molecules within an area and other simulation conditions and also displays the positions of molecules. The initial value is prepared for an array of three-dimensional coordinate data, etc. and a command is given to a vector processor 12 for arithmetic. The vector arithmetic speed of the processor 12 is increased mainly with a constitution where the desired data can be supplied continuously to an arithmetic part 14 with high-speed transfer of blocks between the memory 11 and a vector register 13 and a constitution where the multiplication, division and addition/subtraction mechanisms which can be actuated in parallel to each other are provided independently within the part 14 and these mechanisms can be started simultaneously in a so-called pipeline system.

Description

DETAILED DESCRIPTION OF THE INVENTION [a) Technical Field of the Invention The present invention relates to the application of electronic computers, and more particularly to a molecular dynamics simulator for determining the dynamics of material molecules through calculations by an electronic computer.

(b) Background of the technology There are various techniques for simulating molecular dynamics. For example, the Monte Carlo method is a method for thermodynamically calculating the dynamics of molecules, and is used to explore the dynamics of molecules.

Molecular dynamics method is a method to study the dynamics of molecules based on the calculation of the forces that act between the molecules that make up a substance, and it elucidates the thermodynamic properties of liquids, transport coefficients, etc. at the molecular level. It is said to be effective as a simulation technique for biotechnology, and recently it is being used to elucidate the structure of biological macromolecules in the so-called biotechnology field.

(C) Prior Art and Problems Since these methods determine the positions of all molecules by calculation, the amount of calculation required increases rapidly as the number of target molecules increases.

For example, in the case of the molecular dynamics method, even if the number of molecules is limited to about 500, the calculation will take several hours on current general-purpose high-speed computers, but it is at least suitable for simulations that can obtain practical results. It is said that it is necessary to compile the results for several dozen molecules.

In recent years, computers that can perform high-speed vector calculations have been developed as dedicated computers that process scientific and technical calculations at high speed.
(vector processors) have come into practical use, and by using a suitable vector processor, this simulation can be executed at several times the processing speed of a general-purpose computer. but,
In order to perform simulations of systems containing a sufficient number of molecules, even higher speeds have been desired.

(d) Object of the Invention An object of the present invention is to provide a molecular dynamics simulator that can process molecular dynamics simulations faster than conventional ones.

(e) Structure of the Invention The purpose of this invention is to provide means for holding a plurality of variable values, a means for holding boundary values, a means for holding correction values, and a product of the reciprocal of the boundary value and the variable value when calculating the dynamics of material molecules. a first calculation means for outputting the integer part of the control value as a control value, and a second calculation means for subtracting the product of the control value and the correction value from the variable value, According to the present invention, the calculation of the second calculation means for the variable using the control value obtained by the calculation means and the calculation of the first calculation means for other variables are executed in parallel. Achieved by a molecular dynamics simulator.

In other words, a molecular dynamics simulation involves setting a certain region, calculating the positions of molecules existing in that region based on certain laws, and tracking their dynamics. It is necessary to make corrections for nearby molecules that take into account the influence of molecules outside the area.

For this purpose, it is determined whether the molecule is near the boundary, and the calculations regarding the molecule determined to be near the boundary are corrected, but this judgment and correction processing is performed by multiplication, addition, subtraction, etc. using the configuration of the present invention described above. Consists of. This makes it possible to utilize the so-called pipeline processing of the vector processor, so it is possible to speed up the processing that occupies a large portion of the simulation calculations, and it is possible to achieve the purpose of speeding up the simulation.

(f) Examples of the invention FIG. 1 is a system configuration diagram of an embodiment of the present invention.

For example, the control unit 10 configured with a general-purpose computer has a storage device 11
Above, constants that define the length of one side of the MMi cube to be simulated, the total number of molecules in the region, and other simulation conditions are set, and initial settings such as the array of three-dimensional coordinate data that displays the positions of molecules are set. Prepare the values and instruct the vector processor 12 to perform calculations.

The vector processor 12 consists of a vector register 13, an arithmetic unit 14, etc., and transfers a required portion of data blocks on the storage device 11 to the vector register 13 at high speed, performs a predetermined operation on the data, and stores the result in the vector register 13. It has a configuration in which vector calculations can be processed at high speed by continuously executing the process of storing the data in the storage device 11 through the process.

The high speed of vector calculations in the vector processor 12 is mainly due to the configuration that allows necessary data to be continuously supplied to the calculation unit 14 through high-speed block transfer between the storage device 11 and the vector register 13, and the calculation unit 14.
A multiplication mechanism, a division mechanism that can operate independently and in parallel,
The configuration is such that addition/subtraction mechanisms and the like are provided to allow each arithmetic mechanism to operate simultaneously in a so-called pipeline system.

In general, many vector operations are composed of multiplication and cumulative addition of the results, so the speed can be increased by utilizing this mechanism.

Next, using a simulation based on the molecular dynamics method as an example, we will describe correction for molecules near the region boundary taking into account molecules outside the region. This correction takes a method of indirectly correcting the force by correcting the intermolecular distance for calculating the force between two molecules and using it in subsequent calculations.

FIG. 2 is a diagram explaining the concept of a method for correcting the distance 3 between molecules 1 and 2 for the above purpose. In the figure, 4 indicates a cube with one side having a length of AHx2, which is the area to be simulated. AH is used as a boundary value as described below.

Assume that molecule 1 is the molecule of interest, and in order to calculate the force from molecule 2, the distance between the molecules is determined. Here, when molecule 1 is located close to the boundary of region 4, the distance is corrected.

The correction method is as follows. For example, if the X component A(1) of the distance between the two parts is greater than AI, molecule 1 is considered to be near the boundary, and in this case, the correction value AL is calculated from A(1).
A'' (1) obtained by subtracting (=AH

The figure shows an example where only the X component exceeds AI, and as a result of correction, instead of molecule 2, a molecule assumed to be at the position indicated by 5 in the figure is used to calculate the force acting on molecule 1 in this case. used for.

FIG. 3 shows the flow of processing when the above processing is performed by conventional programming. In step 20, necessary variables are initialized. The part sandwiched between steps 21 and 32 consists of repeating the process to sequentially process one molecule at a time for all molecules in the region (data of the molecule being processed is specified by the subscript I), and by this process, each molecule For each
After the sum of the forces acting on it is obtained! The process proceeds to a process (not shown) for determining the fold position coordinates of the molecule.

The part between steps 22 and 30 constitutes an iterative process to increase the distance between one molecule and all other molecules in the region (corresponding to subscript J), and the part between steps 23 and 29 calculates the distance between a pair of molecules using the coordinate axis X,
This is a repetition of processing separately for the three components Y and Z.

Step 31 is a processing portion that consolidates and accumulates data based on the distance between a pair of molecules.

Step 24 is to transfer data from the storage device 11 to the vector register 13.
In this step, each coordinate axis component of the distance between two molecules is calculated based on the difference in the positional coordinates of the molecules transferred to the subscript variable.
2.3 corresponds to the X, Y, and Z axes, A (K) is each component of the distance, and QN (1, K) and IIN (J,) [) are the position coordinates of the molecule.

The part consisting of steps 25 to 28 is the second
This corresponds to the distance correction process explained with the figures.

Here, steps 25 and 27 are processes for checking whether the absolute value of the distance determined as the difference between the position coordinates of two molecules exceeds AH, and according to the determination, A (K ) value as is (route 33
), a correction value AL corresponding to the side length AHx2 of the area is added to or subtracted from A(K) (step 26.28).

Since the operations in steps 24 to 28 above consist of addition and subtraction and determining whether the results are positive or negative, the vector processor with the above configuration has no choice but to process them serially, and in this part, the configuration of the vector processor as described above is utilized. I can't.

FIG. 4 shows the flow of processing in the configuration of the present invention. The figure corresponds to the portion from step 23 to step 29 in FIG.

In FIG. 4, steps 23, 24, and 29 are the same as in FIG. The step knob 40 following step 24 is a process of multiplying A (K) and RAH by real number calculation and setting the integer part as an integer control value IK. Here, RAH
It is assumed that AH is the reciprocal of AH and is determined in advance.

As is clear from the explanation with reference to FIG. 2, step 4
The control value IK as a result of the calculation of 0 is 1.0 or -1 depending on whether A (K) is larger than AH, between AH and -AH, or smaller than -AH. Therefore, in step 41, the product of the correction value AL and the control value IK is
(K), the same correction as in Fig. 3 is obtained by subtracting it from A (
K).

According to the processing flow in FIG. 4, the flow from step 24 to step 41 includes multiplication, addition and subtraction, and does not require branching due to constant judgment, so the pipeline system of the vector processor configured as described above is effective. It becomes possible to use it.

Although the above explanation has been made using a simulation based on the molecular dynamics method as an example, the present invention is not limited to this.

For example, those skilled in the art can easily implement the present invention based on the above description even in simulations using the Monte Carlo method.

Further, although the above description has been made using a three-dimensional simulation as an example, it is clear that the present invention can be similarly applied to a two-dimensional simulation.

(Effects of the Invention As is clear from the above explanation, the present invention has a remarkable industrial effect of increasing the practical application field of the simulation by speeding up the simulation of molecular dynamics.

[Brief explanation of drawings]

Fig. 1 is a configuration diagram of an embodiment of the present invention, Fig. 2 is an explanatory diagram of correction regarding molecules near the boundary, Fig. 3 is a flowchart of processing in a conventional example, and Fig. 4 is a flowchart of processing in an embodiment of the present invention. It is. In the figure, 1.2.5 is a molecule, 4 is a simulation area, 10 is a control unit, 11 is a storage device, 12 is a vector processor, 13 is a vector register, 14 is an arithmetic unit, 20 to 4 are each processing step Or show the route. Kaya 1 Mecha 2 Prisoner? No. 3 Domo 4 Figure

Claims (1)

[Claims] When calculating the dynamics of material molecules, there is a means for holding a plurality of variable values, a means for holding a boundary value, a means for holding a correction value, a means for holding a correction value, and a third means for outputting the integer part of the product of the reciprocal of the boundary value and the variable value as a control value. 1 calculation means, and a second calculation means for subtracting the product of the control value and the correction value from the variable value, and using the control value obtained by the first calculation means for the variable 1. 1. A molecular dynamics simulator characterized in that the second calculation means performs calculations on the variable and the first calculation means performs calculations on other variables in parallel.