WO1997046949A1  Molecular modeling system and molecular modeling method  Google Patents
Molecular modeling system and molecular modeling method Download PDFInfo
 Publication number
 WO1997046949A1 WO1997046949A1 PCT/JP1997/001951 JP9701951W WO9746949A1 WO 1997046949 A1 WO1997046949 A1 WO 1997046949A1 JP 9701951 W JP9701951 W JP 9701951W WO 9746949 A1 WO9746949 A1 WO 9746949A1
 Authority
 WO
 WIPO (PCT)
 Prior art keywords
 molecular
 potential
 interaction
 parameters
 device
 Prior art date
Links
Classifications

 C—CHEMISTRY; METALLURGY
 C07—ORGANIC CHEMISTRY
 C07K—PEPTIDES
 C07K1/00—General methods for the preparation of peptides, i.e. processes for the organic chemical preparation of peptides or proteins of any length

 G—PHYSICS
 G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
 G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEINRELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
 G16B15/00—ICT specially adapted for analysing twodimensional or threedimensional molecular structures, e.g. structural or functional relations or structure alignment

 G—PHYSICS
 G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
 G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
 G16C10/00—Computational theoretical chemistry, i.e. ICT specially adapted for theoretical aspects of quantum chemistry, molecular mechanics, molecular dynamics or the like

 G—PHYSICS
 G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
 G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
 G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
 G16C20/90—Programming languages; Computing architectures; Database systems; Data warehousing
Abstract
Description
Akira fine manual molecular modeling system and molecular modeling method TECHNICAL FIELD
The present invention, molecular modeling system (Molecular Modeling Systems) partial child Moderui spoon (Molecular Modeling) molecular dynamics (Molecular Dynamics) molecular mechanics (Molecular Mechanics), material design (Materials design), molecular design (molecular design) Pharmaceutical Design (pharmaceutical drug design) conductor processing modeling and design, chemistry and chemical treatment, protein modeling (protein modeling), and the art of each of the modeling Hadouwea, how the space soft toe air Tools and Tools It belongs to.
The present invention further, ¾ optimization (optimization), Detafui Tsuti ring (data fitting), the process optimization (process optimization), and storm to the technical field of methods and tools area of the realtime process control (real time process control) to. BACKGROUND
To develop new technologies and products in the technical field of the present invention, after 济的 analysis Tekihara Ryo  used during interactions 沉関 number (analytical atomic interaction functionals) force component child model systems and molecular modeling methods there. The analytical ': problems are arising from intermolecular interactions functional, calculations or is Ru efficiently der in terms of cost is a very inaccurate when describing an interaction between atoms, or calculations somewhat accurate but it is that the relatively high cost. Is shown in the molecular modeling system and molecular modeling method, speed and accuracy for calculating the interaction potential (interaction Potentials) are very i constant. Generation of molecular potential (molecular Potentials) and known or unknown potentiometer Nsharu potential Roh instrument lame Isseki for (Potential parameters), the molecular modeling system and molecular modeling methods that have been carried out separately. Problems are associated with this, for the task of generating new accurate warp 济的 of potential use in performing molecular modeling, it is necessary to use a suitable system experience, and user input data output fin evening to  there is a need to create a Fuwesu.
Potensha Le used in the molecular modeling system and molecular modeling methods the computation cost can not obtain reliable results or can not be calculated so high, or so very inaccurate. If that describes the interaction of atoms, Lena one dough commonly used di yaw lens type functional (LennardJones type functional) (T.Halicioglu, GM Pound, Phi s. Status Solidi, (a) Vol. 30, page 319, 1975) is r ^{beta} (r but is Junji KaradaHiroshi function that provides a physically significant attractive interactions proportional between atoms distance) (pure 2 body functional), the child since the function form of the repulsive force is not physically equivalent, particularly significantly excessive ^ valence atoms question attraction for atoms dissimilar type (UNLIKE type), the result , prediction of reacting with atomic structure not i! happens that be accurate. Indeed, Lena one Dojiyonzu type functional is ΪΕ probability only when describing for fl! Interaction uniform noble gases. Illustrating the interaction between atoms, ... The layer elaborate Buckingham type functionals (Buckingham type functional) (J Schulte, RR Lucchese, WH Marlow, J. Chem Phys, 99 (2) winding, 1 178 [pi 1993 years ), there is a meaning between the atoms of higher than r ^{e} following number of attractive terms and the physical, original? The question repulsive force is Junjikarada functional containing indices to function form. The Buckingham type functional, when describing the interaction of two separate atoms, even if these atoms are not the original of the same type, can provide a ^ have accuracy, this function is pure twobody resistance (pure 2bodv nature) since it is, in the presence of a third original 7that can not be able to accurately describe the change in the atomic interactions. Constitution Li Nga eleven Weber type functionals commonly used (Stillinger Weber type functional) (F. H. Stillinger, TA Weber, Phy s. Rev. 31 (8), pp. 5262, 1985) are twobody cross It contains functional on the effects (2body interaction) Hiroshiseki number and Sankyu interact about (3body interaction). Twobody interaction terms of its is a term of Lena one Dojiyonzu type, terms of threebody interaction is exponential functional (exponential functional) was weighted in angledependent terms (angular dependent term). Since twobody interaction terms are prime interactions Institute Li Nga  the accuracy of the © We ICHIBA first mold potential is essentially limited, its reliability homogeneous such gay element (homogeneous systems not only obtained). Triple functionals (3body functional), when forming one molecule two atoms in the reaction approaches (dimer (dimer)) is not described, threebody section (3body the contribution of the term) is reduced. In this case, the size of the threebody section stearyl I Li Nga eleven Weber functionals cause severe problems with increased accuracy. Constitution Ringer one Weber functional because of the exponential function and the angular dependence of threebody interactions, cost calculations than Junji body functions high. Other known triple functional of data one soft functional (Tersoff functional) (J. Tersoff, P ys. Rev., B 39 (8), pp. 5566, 1989), even if the uniform interaction huge number of parameters Isseki calculation cost is not very high because the widely used because it is necessary. For a threebody atomic interactions one gallium atoms and two gay atom, functional form and each parameter is not known in the literature. Multiparameter functional, generally there are two range 畴 the given dataset for example, a normal optimization methods and tools to be adapted to the intermolecular interaction potential. That is based on the simplex method (The Simplex methods) and the descent method (The Gradient Descent methods) to those Zui ¾ in, stains Jure one tee Tsu door Neil Monte Carlo 'method (The simulated annealing Monte Carlo methods). The simplex method and the steep descent method, its reliability and there is a fire listening dependent to that and have you problems in the initial value of the optimization. Because each algorithm is optimized to give glance the minimum point of the parameter set Bok, in this case, the minimum point is, if a force ,, or the highest minimum point or a global minimum point which can be achieved others because never each found the minimum point to check if one der Runoka of a number of local minimum points. Therefore putting Yee Glo one Bal minimum point or maximum parameter set for functional to be optimized is very redundant calculation cost becomes high. Some descent method is to indicate that an first derivative and second derivative of the functional to be optimized are required, computational expense and inconvenience of these methods is increased. Ani Le optimization method is nonpolynomial method overcomes most of the problems caused by local minimum points in the parameter Isseki space, the calculation cost, § two one Le procedures and order, the user one experience, and the way of user intervention to Aniru process relies heavily. Aniruparame costs Isseki optimal Kasho management includes a user skill level, depends on the sensitivity of functional for small deviations of the parameter values. Of Pena, Aniru method convergence is optimized (TJP Penna, Phys. Rev., E5 R 1, 1995 year) is somewhat overcome high computational costs associated with Aniru S optimization process, Pena method, parameters to be optimized  evening does not work effectively only if the functional functioning a few cases and simple enough. Thus even Pena optimization method, is not suitable for complex functionals you 冇 multiple dependent variables to optimize fast, it requires skilled users one careful monitoring and intervention, and that ^賴性 is highly dependent on the user one choice of the initial test para Metase' Bok initial value. Pena method is another method of optimizing the large ¾ problem can be overcome some, not simple pan閱数 be ½ optimization, i.e. if it contains two or more different types functions, and its pan If the function is dependent on a number of adjustable parameters, the efficiency is deteriorated as well as other methods. These problems are, parameters one evening flying distribution of (visiting distribution) that all is the same, that the flying distribution of Aniru temperature (the annealing temperature) is the same and have all of the parameters Isseki Nitsu, and of these parameter Isseki, even if the para main Isseki unacceptable, Ru Oko by all following the same Aniru order. Pena is co one sheet one  Lorenz (CauchyLorentz) of parameters Isseki distribution quenching temperature (quenching Temperature) is generally turtle Toroporisu temperature (the generalized Metropolis temperature) and different One in which the pointed out Chikaraku, this improved the above problems can not be overcome even with.
Molecular modeling system and molecular modeling methods, when para Metase' bets can thus be provided to the input of each of the molecular modeling tools or user against potential molecular dynamics (Molecular Dynamics) method (Molecular dynamics constituting the molecule was with the atoms considered as new Bokun dynamics of particles, make them was motion through the potential energy force field equations, the Re their intended to calculate over time, detection of fluctuations in the short time of the molecule 討and it is used in the analysis of changes in Λ body structure that can be taken. since it is found dynamic state do not know a static measurement and meter », 冇力 hand method.) and molecular mechanics for the analysis of functions and behavior of the molecule (molecular mechanics) method (the potential of the molecule is approximated by the force field of classical mechanics, the energy calculated on the basis of the potential  original so is minimized By moving the position of the child, how to determine the stable structure of the molecule. Molecules can be calculated at an earlier time by one digit or more as compared with the orbital method, can also be calculated in a very large molecule may tooth.) Intermolecular phase used in the calculation ¾: selection of work for Hotensharu (molecular interaction potentials) is limited. Creating precise molecular potential functional and molecular potential parameters Isseki is usually carried out away from the molecular modeling system to 閱 generation potential, optimization of the parameters Isseki, the reduction knowledge intermolecular interaction potential a redundant tasks requiring great experience user, further ^ and have calculations costs, and need to Intafuwe one single for input to a separatory f modeling system Isseki parameters and generated potential externally Become. Molecular dynamic mechanics apparatus and method, molecular mechanics apparatus and method, and the molecular potential di energy laser evening device and molecular modeling system and molecular modeling method of the present invention to have a method to overcome these problems.
fit "calculation is through 济的 analysis heterogeneity molecule Q phase S acts functional (analytical heterogeneous molecular interaction functionals) force, ', used in atomic and minute shea Mi Interview record one to down method and software Tuwea. Analysis atomic intermetallic phases ¾ effect problems arising from functional, these functionals are in Thus cost but Ru efficient der, although in describing the interaction between the atoms is very inaccurate, and yes rather sentence accuracy is that is one ¾ cost of good, calculated according to the. present invention, a problem in the art of the present invention that accurately describes the interaction between heterogeneous atoms a and at a low computational expense precision the molecular heterogeneity interaction potential apparatus and method of the present invention, a general functional form that does not depend on the type of atoms can be overcome by take advantage.
One gallium and about threebody atomic Q phase ¾ action of ¾ί terminal of two gay element (the 3 body atomic interaction), the function form and each parameter Isseki not known in the literature. HaraRyo function form and one gallium and two gay element of the present invention  of the parameter set Bok of threebody interactions, and that no known the interaction functional, each parameter Ichitahan Iffl since the unknown para Metase' I, problems occurring in the technical field of the present invention, can be overcome by the potential apparatus and square of the present invention ^. Complex multiparameter functional, such as molecular potential is given  it is very difficult to dataset optimized Bok was gills. If because the normal optimization methods is the order for calculating costs according to optimization instead be when you'll overcome function minimum point can be not and locally to 冋避 fleeing function minimum point of the local . The routine optimization methods require a lot of skill and intervention of the user one. These problems can be overcome by the high speed optimized apparatus with ¾ optimization method of the present invention.
Initially known and unknown functional dependencies of the process at high speed when optimizing the control process (functional dependencies) prediction, the prediction and control of the process, at the same time accurate and reliable predictions at a short time 內¾ Masui be carried out. Process control tools normally used lacks an or both desirable properties of the. This problem can be overcome fast optimization apparatus and optimization method of the present invention.
The molecular modeling process requires expertise physical and chemical properties of the constituent atoms of the model molecules. The present invention, a problem that high degree of expert knowledge on the molecular modeling processing is required, using a molecular model modified Kisupa Bok database device, the best combination of automatic model KaSo a and modeling methods it can be overcome by choosing. Disclosure of the Invention
And a limited number of molecular potential, and since the potential is limited the interest w text Ru accuracy, to solve the problem to to put the Ru molecular modeling systems and molecular modeling methods, and the precise molecular potential and Potential to generate parameters, the present invention uses an optimization tool and square ^ and heterogeneous intermolecular interaction potential instrumentation E and method of the present invention of the present invention, it can be accurately and economically produce a potential parameters present utilizing molecular potential generator of the present invention. Nonuniform interaction potentiometer tangential apparatus and method of the present invention is used to resolve the low accuracy and ^ dependability of problems with the interaction of different types of atoms.
To solve the individual selection is low accuracy resulting from that limited the reliability problems of molecular dynamics coordinate development equipment (Molecular Dynamics coordinate propagator) and molecular potential device, and molecular modeling system of the present invention molecular modeling methods can be arbitrarily combined individual selection of coordinate development device and molecular Potensha Le device and method.
Since the selection of optimization and potential apparatus and method of the molecular mechanics to solve the problems that arise from the fact that is limited, the molecular modeling system and partial modeling method of the present invention, the optimization apparatus and molecular potential apparatus and method it can be combined with the selected arbitrarily.
The new intermolecular interaction potential by generating, order to solve problems arising when fin evening face the potential to molecular modeling tools, molecular potential di We Nere Ichita of the present invention, the present invention It is configured as a molecular modeling system and integral part of molecular modeling method (an integral part).
To solve the normal threebody interaction potential accuracy of the used and unreliable L, and problems with particular not divide the potential from the Hare problems and uneven threebody interactions, nonuniformity of the present invention intermolecular interactions generic Seki 5X (the generic heterogeneous molecular interaction potentials)! nonuniform molecular question interaction functional of ie threebody levels, migraine as base one scan for the new interaction Po Tensharu apparatus and method Irare, the functional is composed of different kinds of a small number of functions with a small number of adjustable parameters Isseki, and two rest and physical properties and chemical properties of different atomic interactions triple is the pan functions of disproportionation   a twobody factor (2 body p art) and represented in a special form of a threerest factor.
Geimoto  gallium  for solving the gay ^ connexion caused problems due to not known economic intermolecular interaction potential interaction to you appropriately and accurately modeling, disproportionation of the present invention  molecular potential, the threebody interaction energy surface by the first principle calculation (the exact first principle 3body interaction energy surface) is jk over 丫匕.
Complex multiparameter Isseki functional, in order to solve the problems associated with accurate optimization fast molecular potential and control process, individual test distribution of the parameter (trial distribution), the parameters of 侗 's Aniru temperature ( Annealing temperature)> and individual ha ^{0} lame Isseki receiving probability of the parameter Isseki (parameter acceptance probability) and receiving Aninore has a temperature (acceptance annealing temperature), highspeed Aniru method of the present invention, highspeed selfregulation of the present invention type optimization system (fast self regulating optimizer) and optimization (optimization method) to the ¾ filled are as Ichisu.
To solve the problems associated with making accurate predictions in the control process and the highspeed real time interrogation process control, optimization of the present invention and prediction, as well as the individual test distribution of the parameters, parameter Isseki individual Aniru temperature, and highspeed Aniru method of the present invention having the parameters taken of individual parameters receptor probability and receiving Aniru temperature, and highspeed selfregulated optimal KaSo s of the present invention to predict the parameters in the control process and the realtime process control It wants to use as a basis for the optimization method.
Regardless to require a high degree of expertise for its user to resolve Toi题 time was otherwise know the best combination of molecular modeling Soinui and methods for molecular data given no expertise, molecular modeling expert data base Ichisu apparatus and method of the present invention, the characteristic HaraRyo , minute Kode Isseki based on the base, select ¾ optimal combination of the modeling device and method self ί manner to be used the best combination of equipment and methods in molecular modeling system and molecular modeling methods as a basis for automatic determination. [A brief description of the ¾ surface
J 1 is a diagram showing the structure of molecular modeling system of the present invention, Department 2, minute molecular modeling system of the present invention? "Is a diagram showing the structure of a modeling system processor 105,
Figure 3 is a diagram showing a configuration of a molecular dynamics (MD) device 107 of the present invention, FIG. 4 is a diagram showing a configuration of a molecular mechanics (MM) device 108 of the present invention, FIG. 5, the is a diagram showing the structure of a molecule potential di Nere Ichita (MPG) instrumentation S 109 of the invention,
6, if the molecular modeling system has only molecular dynamics system, if the molecular modeling system has only molecular dynamics device and molecular mechanics device or molecular modeling system of the present invention, is the used case is a diagram showing a comparison of steps necessary to resolve the problems in example 1,
Figure 7 is a diagram showing a configuration of a potential instrumentation E 1073 molecular dynamics device 107 of the present invention,
Figure 8 is a Department showing the configuration of a tracking device 1072 of the molecular dynamics device 107 of the present invention,
9, as a measure of computational efficiency problems molecular modeling of Example 1, relative CPU more hours related calculations are ίΠ Les Oxnard over Johns potential instrumentation H, as a measure of accuracy, from the calculated value of ab Imtio method relative deviation, as a measure of performance, by employ performance = (1 + relative deviation) Z (relative CPU), molecular modeling system of the present invention is utilized to select properly the potential device, Takashi, is a small to] to achieve the performance,
10 is a diagram showing a configuration of a potential device 1083 of minute mechanical instrumentation I 108 of the present invention, 11 is a diagram showing the structure of the optimizing device 1082 of molecular mechanics device 108 of the present invention,
Figure 12 shows a ^ 施例 molecular modeling of 1 issues, by proper selection of the molecular modeling system was the combining potential device of the present invention, to achieve the best performance for a precision and computational costs It is a view,
Figure 13 is a diagram for comparing when the ability to solve the problem of molecular mechanics of Example 1, which were selected each optimization apparatus ¾ optimization device 1082 of molecular mechanics device 108,
1¾1 14 is a diagram showing a structure of a potentiometer catcher Le device 1093 molecular Potential di We Ne laser evening device 109 of the present invention,
Figure 15 is a diagram showing a structure of optimum KaSo 1092 molecules potential generator apparatus 109 of the present invention,
Figure 16 is a memory device 114, shows distributed processing instrumentation ft 117, and the configuration of the input device 101 of the molecular modeling system of the present invention (MMS),
17, in the case of selecting the device in each memory device 114 and distributed processing instrumentation fi 117 most efficiently, the CPU time required to calculate the interaction between atoms of Example 1 Molecular Dynamics is a diagram showing the performance of academic modeling problems,
Figure 18 is had contact to the molecular mechanics problems of Example 1 containing argon noble gas atoms, to describe the interaction of atoms, when using the Rena one dough Ji's potentiometer Shah Le device 10821, descent optimization apparatus 10823 the is a diagram indicating the most efficient use with a single instruction Mar Chideta processor 1171 and distributed shared memory device 1142,
Figure 19 is a single instruction multiple data (SIMD), multi Inochigo multiple data (MIMO), distributed shared memory (DSM) and distributed CPU time required in the case of iS also efficient selection of instrumentation ffi unshared memory relative CPU ^ Te cowpea in question, a Department showing the performance of the molecular potential generation problem of example 1 against,
1¾ 20 is a diagram showing a configuration of a data output device 1074 of the molecular dynamics ManabuSo K 107 of the present invention,
Figure 21 is a diagram showing a configuration of a data output device 1084 of the molecular mechanics instrumentation S 108 of the present invention,
Figure 22 is a diagram showing a configuration of a data output device 1094 of the molecular potential di We Nere Ichita device 109 of the present invention,
Figure 23 is a function section (1) consisting of (9), be a drawing two rest factor f 2 nonuniform interaction functional of molecular modeling method and shows the functional form of the triple factor f 3 of the present invention , 1 24 is S flowchart of his own regulatory multiparameter optimization method of the present invention, 25 is a diagram showing the performance of the optimization method of the present invention compared to high speed Pena algorithm,
Figure 26 is a optimization method of the present invention, a Gaussian functional yW ! A exp ^ x ^ ^ B,) the two parameters), the original functional y (x) = 210 exp ( x 2 / (2 + 1.45)) (a view showing compared to the starting test parameters Isseki is 卨速 algorithm according Pena optimized convergence and accuracy for 100 function values of all equal to 1), 1 27 , of the present invention: simplex method accuracy and computational efficiency optimization method, in ¾ 施例 optimizing with respect to the ab initio data set Bok having parameters and 600 data points functionals described in example 6, descent method is a diagram comparing the Uranaiten manner Monte Carlo method and Pena method,
Figure 28, for example missile trajectory prediction and respective control processes as seen Oite, some dependent of the initial unknown functional process dependency variable process predicted for controlling the optimization apparatus 400 is a diagram showing the configuration,
Figure 29 is performed, for example of the power grid β load completion, the the measuring process and prediction of the initial known functional process dependency of the dependent variable of some processes such as Yee are in each control flop port processes control is a diagram showing the optimization apparatus 500 of the configuration for,
Figure 30 is a view to showing the configuration of a Ekisupa one bets database processing apparatus 111 of the present invention. BEST MODE FOR CARRYING OUT THE INVENTION
[Example 1]
Example 1 of the present invention with reference to FIGS. 1 to 15 described below.
Molecular modeling system input device 101 in FIG. 1, molecular modeling system processing unit 105, molecular modeling system output device 106, a memory device 114, data mass storage device 115, network Bok work file system unit 116, the distributed processing equipment 117 and comprises a CRT device 118, it shows a functional block diagram of the molecular modeling system connected by these devices all common data / Adoresubasu 113 (MMS). Molecular dynamics in FIG 2 (MD) unit 107, molecular mechanics (MM) device 108, the molecular potential di We Nereta (MPG) 109 and molecular models Kako Kisupatode Isseki molecular modeling systems having a base instrumentation fi 111 It shows a functional block diagram of the processing device 105. By using the molecular modeling system processor, molecular dynamics and molecular dynamics molecular modeling potential Ru using the task min 7 model Kanogu lame Isseki problems arise from not such Ikoto found; said problems associated with the generation of the unknown parameters; to produce the unknown parameters in the outer, in particular associated connecting parameters Ichita of Suechi that generated in the outside via the Intafu I scan the molecular modeling system problems; calculated for selected combinations of potential devices and optimization apparatus molecular modeling systems are limited say that the inefficiency problem, and combinations best molecular model for any molecular modeling device problem you do not know how to use the system is overcome. ^ SL problems are of molecular dynamics (MD) device 107 (FIG. 3) is a portion potential device 1073 and the tracking device 1072; potentiometer Shall device is a part of the molecular mechanics (MM) instrumentation [108 (FIG. 4) 1083 optimum KaSo S 1082; is overcome by and molecular potential di We ne regulator (MPG) instrumentation S 109 potential device 1093 best KaSo 1092 is a portion of (Oka 5). More Before I ¾ problem, the molecular modeling system processing unit 105, molecular dynamics device 107, molecular mechanics device 108, the potential di Nere YuSo S 109 and Expert Bok database instrumentation IS 111 (all of which instrumentation each other data / comprising a Adoresubasu device 110 are connected (Figure 2)), means for selecting a potential device devices 1073,1083 and 1093 內 the device 107, 108 and 109, means for selecting a tracking device in the apparatus 1072, optimum and comprising means for selecting the optimal KaSo [of instrumentation fi 1082 and in 1092, as well as one of the selection of potential devices to uneven potential device of the present invention, and the present invention the single selection of the optimizing device and apparatus, and by selecting the best combination of features by Ri automatically modeling device to expert database apparatus 111, overcomes It is. By the molecular modeling system and molecular modeling system process ¾ il 105, it is possible to achieve high modeling efficiency, as well as the modeling accuracy, a ^ 賴 1 and a high speed efficiency.
In molecular dynamics device 107 (FIG. 3), the data manually apparatus 1071 performs the manual processing of data molecules ί Li J mechanics device, tracking devices and instrumentation E 1073 內 potential device in the device 1072, the molecular accordance with user input the modeling task 'row. De Isseki output instrumentation E 1074 of device 107 processes the model data of the instrumentation E 1072, 1073, processes the output data Isseki by defined modeled task by the user input. 4, the data input device of FIG. 5 1081,1091, and data output instrumentation ffi 1084, 1094 also has the same function as described above. Ekisupa In one Bok database instrumentation ifr ι ι ι (¾ 30!), The data input device or nil Ekisupa  performs input processing of data collected by the database device, Ekisupa one Bokude Isseki based device 1 112 and Ekisupa one bets database determined apparatus (the Expert Data Base Decision unit) 1 113 performs the selection of the modeling device in accordance with the user input. Instrumentation data output device 1114 of the S 111, the device 11 12,1 113 processes the output data Isseki of, processing the output data by the defined modeled task by a user input, the modeling device of the apparatus 11 13 It proposes a selection. High modeling efficiency molecular modeling system of the present invention, the embodiment of FIG. 6 binary 7dynamics i.e., indicated by a semiconductor process issues the interaction of argon rare gas atom and Gaycontaining semiconductor surface. Results of Efficiency Test (Efficiency results) are compared with model systems of different configurations. In this embodiment, the interaction between the argon rare gas atoms are known, are known in the interaction of the serial semiconductor atoms question, and the interaction between the semiconductor and the noble gas atoms are not initially known assume that. Independent of molecular dynamics system or molecular mechanics system molecular dynamics system having, or in order to solve the above problems by using the molecular modeling system of the present invention, implementing the steps of the several system users must, in some of its step, the user of Kas evening 厶 program Cody ring includes a difficult task requiring knowledge of the user one of Ri written many times. This opens the interface shown in FIG. 6, in order to solve the above problems, molecular dynamics system having molecular dynamics systems, and molecular mechanics system must implement the total 10 steps, 4 of these steps step is required user Kas evening no intervention. Molecular modeling system of the present invention is to implement the problem within only 6 stearylup, custom step of the user one is unnecessary der Ru of the high partial Chimoderu efficiency is obtained. And more selective efficient molecular model KaSo S, with the assistance by expert database system, the user of the standard molecular modeling system, complex industrial materials, design chemicals, also be used for other molecular design problems it can.
髙 I precision andreliability by molecular dynamics system, the individual minute  for explaining the f interaction can be achieved by selecting the most appropriate potential devices within device 1073 (FIG. 7) . Thus, for example, minutes containing Al Gon noble gas atoms and Gaycontaining semiconductor atoms  problems of modeling molecular dynamics is the most suitable and accurate potential instrumentation a [Les Na one de to the noble gas atoms  Job lens select (Lennard Jones) potential device 10731 can be mentioned, et al is, most suitably from 0 to the semiconductor atoms: exact potential instrumentation BL [Institut Li Nga eleven Weber (StillingerWeber) such potential device 10733 of select〗 the triple potential devices include, and are most suitable and accurate potential device [Buckingham (Buckingham) type potential instrumentation S 10732 Ga举Ge to describe the interaction of the noble gas atoms and semiconductor atoms ] it can be described most accurately by selecting. When the potential instrumentation s describing the interaction of different types of ίί SL one ^ only be selected, the prediction of the molecular dynamics and reliability becomes significantly inaccurate loss. If you choose the different ¾ for all interactions also accurate Po Tensharu apparatus or molecular orbital potential device 10737, unnecessary computational expense is very expensive consuming. Thus, by selecting individual most suitable potential instrumentation ifi, it is a child achieve high computational efficiency. The calculation result of the examples of the problem shown in FIG. 9 shows, as a measure of computational efficiency, the calculation functions use 卨速 COMPUTE i.e. Lena one Dojonzu device relative CPU time Q small  are then. As a measure of accuracy, of the exact calculated value determined ab initio method [Molecular orbital method for total ^ the electronic structure of the molecule, using empirical parameters minute calculation of each 祯 Ψ empirical BunRyo  unlike orbital, such empirical parameters without using Isseki, are shown relative difference from method of calculating integrated by Hisoka Imu way. As a measure of sexual ability, performance  (1 + relative deviation) / is adopted (relative CPU), which is the value obtained means that high performance of Zenkyu closer to 1. To provide the molecular modeling system of the present invention, high if utilized in proper choice of potential device precision used to calculate the individual interaction molecular dynamics problem, molecular dynamics in reliability and efficiency best performance is achieved to solve the problems that will be appreciated from FIG. Their most appropriate tracker (propagation units) for each potential device respectively Ru can achieve higher efficiency by selecting (Figure 8). Molecular models Kako Kisupa Bok database apparatus 111, the nonspecialist Interview  to support The, according to the model of the target intended molecular data of interest (molecular data object), the optimal combination of molecular modeling device It makes it possible to determine the selection.
High accuracy and reliability by molecular mechanics instrumentation a can and child achieved by selecting the most appropriate potential device that describes the interaction between the individual molecules. Thus, for example, a semiconductor of the molecular mechanics problem of optimizing the molecular structure containing one gallium atoms and two Kei MotoHara child is most accurately described by selecting the correct potential device most suitable person Although kill, as such potential device (FIG. 10), Steri changers one © We over bar potential device 10833 or such as a potential device 10839 of the present invention, threebody in potential device 1083 (FIG. 1 0) Potensha there is Le devices. The potential device to explain the interaction of the different types  If you do not select or turn, the reliability of the molecular structure where the maximum error is calculated by increases and as a result the molecular modeling system is significantly reduced. If you choose the most suitable molecular modeling optimal KaSo ift for each object of the structure optimization, a very large computational efficiency can be achieved. Some of the optimizing device, it is necessary to assemble the ffl first derivative of each of the potential which need a Sshianmatori box to the first derivative (the Hessian matrix) at potentiometer Shall device, the Conoco, Renner de one is a suitable if used with job lens potential device (the LennardJones potential unit) 10831, the use of the optimization apparatus together with the threebody potential device is most unsuitable. Therefore, by selecting an optimum KaSo location for each potential device utilized, as possible out to achieve a more human hearing computational efficiency. Figure 13 shows the ability to solve the problem of the molecular modeling. As a measure of its performance, relative computational expense in the case of using the SiGaSi molecular structure optimization of the relative accuracy of the optimization apparatus of the present invention and (second column in FIG. 12) 10828 (11) (FIG. 12 a third column) of. Its performance values are calculated by the performance = (1 + relative accuracy) I (relative CPU cost). Performance that is very good value closer to 1. As can be seen from FIG. 12, the best performance due to accuracy and computational cost, it can be achieved by molecular modeling device of the present invention there is proper selection of Opushon of combinatorial port Tensharu device. With proper choice of optimizing device performance is improved to al. Performance, for the case where other optimization device is used for the problems the molecular mechanics, indicated by the relative deviation monument and required CPU time questions from accurate ab initio minimum energy structures. Other optimization device, compared to the best performance potential instrumentation E used in the Po Tensharu optimum KaSo K 10828 i.e. the molecular mechanics problem of the present invention (FIG. 12). Figure 13 shows that performance is low when there is a limit to the selection of the optimization apparatus. That is, when using the usual standard optimization apparatus used, such as Shinpuretsu box device (Simplex unit) 10821 or dive instrumentation fi (Gradient Descent unit) 10823, less accurate and ^ dependability of ¾ optimized structure and calculation time is long. Top ^ performance against issues the molecular mechanics can be achieved by have river optimal KaSo E 10828 of the present invention. When other optimal KaSo [meter compared to S Ken question is rather short, the relative deviation from the correct value was very small. Therefore, it is possible to achieving the highest molecular mechanics performance by using molecular modeling system of the present invention to provide an independent combination of selection of potentiometer Shall instrumentation S and £ i optimization device. Molecular modeling expert database device 1 1 1, to help the user of nonprofessionals, according to the model of the target that was intended to molecular data for the purpose (molecular data object), the optimal combination of molecular modeling device It makes it possible to determine the selection.
High accuracy and reliability by molecular potential generator instrumentation ffi, compared potential energy surface data provided to the input, used to be optimized or adapted each potentiometer Shall parameters of Isseki the set Bok, mutual individual molecules it can be accomplished by you to select the most appropriate potential device for explaining the operation. For example, one data Gariu 厶原 terminal and two energy surfaces describing the interaction of gay atom is also suitable and accurate means ¾ describing the atomic interaction in order to optimize the potentiometer down Shall parameters it can be most accurately approximated by selecting. And that means in the case of the embodiment, apparatus 1093 (FIG. 14) is a two rest potential device in example Steri changers  is © We over bar potentiation catcher Le device 10933 or potential instrumentation ffl 99/10939 of the present invention. Achieve With potential instrumentation E 99/10939 of Honhatsu叫 to together optimally KaSo S ten thousand nine hundred twentyeight and (15) of the present invention, Ne of the outermost j¾ optimized potential parameters Isseki the set, the reliability and performance can do. 13, the case of using other optimization apparatus such for the previous SL problem of generating molecular potential parameters using the potentiometer Nsharu device 99/10939 of the present invention, the relative deviation from the exact ab initio potential energy surface σ shows Yotsute, the performance of potential parameters Isseki optimized during the time and required relative CPU. 13, when the selection of ¾ optimization device is or restricted limited performance low  indicates that sul. In other words, simplex device 10921 (FIG. 15) or the use of normal standard optimization apparatus used, such as sudden descending F unit 10923, the optimization potential parameters set Bok accuracy and ^ dependability decreases when calculating Q is Kunar . The molecular potential parameters Isseki set best performance against the problem of 敁適 of can be achievement by using optimization device 10928 of the present invention. Therefore, optimization performance of the molecular potential parameters set of 设高 generates potential parameters Isseki, the molecular modeling system of the present invention to select a combination of potential instrumentation fi and optimizing device independently it can be achieved by taking advantage of.
Molecular modeling system processor 105 is connected to the common data / Adoresubasu 113 (FIG. 1), and a data address bus device 110 device is connected by 107, 108, 109 and 1 11, and these device can achieve higher computational efficiency because accessing data mass (data mass) storage device 115, network Bok work file system unit 116, the distributed processing unit 117 and the memory device 1 14. Because, since the processing unit in the apparatus 105 are each able to access the distributed calculations means most appropriate topically specific molecular modeling problems to be solved.
For example, in the case of problems of molecular dynamics containing argon rare gas atom and a semiconductor Gay atom, Renner de most suitable port Tensharu device to describe the interaction of argon rare gas atoms  Select Johns apparatus 10731, wherein selecting a semiconductor atoms Institut Ringer one © We over bar potential device relative 10733, and selects a bar Kkingamu type potential device 10732 to calculate the interaction of argon rare gas and Gay element. To make the best molecular dynamics modeling, distributed processing unit 117 dispersed Single Instruction Multiple Data location (Figure 16) in (Distributed Single Instruction Multi Data Unit) with 1171; memory devices 114 distributed co ^ memory device (distributed Shared Memorv unit) 1142 Lena one de  employed with Johns potential apparatus and back Kingham type potential device and multi instruction multiple data devices (multi instruction multi data unit) 1172; and dispersed non Kyo冇Me memory device (distributed unshared memory unit) 1143 Stille ringer one  Uweba  Ri most efficiently der be used with potential device, so that the tracking of the molecule coordinates (the propagation of molecular coordinates) is, the local memory device (local memory unit) 1141 together it can be carried out by using each of the track device (propagator unit). Figure 17 is the molecular dynamics models 閱 the CPU time required to compute the interaction between the atoms on when a device within the respective instrumentation S 114 and device 1 17 and most effectively selected to the shows the performance of of the problems, it is close to the value Chikaraku 1 relative CPU time shows the best performance. When using the Renner de one Johns potential device with dispersed nonshared memory device, it can be seen from Figure 17 that the value is lowered performance becomes 1.3. Constitution Ringer  © over bar potentiation catcher Le device choosing a different decentralized computing devices instead of 10733 performance is reduced to al. Molecular model Chemical kiss part database instrumentation S 111 is to assist the user of nonprofessionals, according to the model of the target that was intended to molecular data for the purpose _{ (mo l ecu l ar data object } ), molecular dynamics equipment, molecular modeling distributed processing apparatus, possible to determine the memory device, and the selection of the optimum combination of the data mass storage device.
In the molecular mechanics containing argon rare gas atoms problem, if the Ru with Lena one dough Johns potential device 10831 to explain the interaction of the atom, the descent optimization apparatus 10823, a single instruction multiple data processing instrumentation fft 1171 and it is most effective to use with distributed shared memory device 1142. With multiinstruction multiple data devices (Multiple Instruction Multiple data unit) 1172 with respect Lena one dough Johns potential apparatus, the value of the relative CPU time is reduced in molecular mechanics performance becomes 1.3 (Figure 18), distributed nonshared memory with the apparatus 1143 the performance is further decreased. Accordingly, with the present onset Ming molecular modeling system, it is Rukoto has limited the selection of the calculation means are Fukusa ¾ Insufficient computational performance causes. Molecular model modified Kisupa  Bok database instrumentation g 111 is to assist the user of the nonexpert, in accordance with the intended modeled target molecular data shall be the object (molecular data object), molecular g optimization device, molecular modeling distributed processing device makes it possible to determine the memory device, and select the best combination of data evening mass storage device. In the molecular potential di We Nere one Deployment issues Arugon noble gas source interacts  Ryo  the potential energy surface case to be adapted, for example, distributed single instruction multiple data devices 1171 and distributed both 苻 memory equipment 1142 together with Lena one dough Johns potential device 10,931, it is most efficient using a calculation with a single instruction multiple data processor 1171 distributed shared memory device 1142 dive optimizer 10923 using as well. Figure 19 is due to the most effective relative CPU time related to the a CPU time required by the selection of the device, showing the performance of the molecular potential generational sucrose emissions problem. 19, to the S optimization apparatus 10923 and potential device 10931, the predispersion different combinations of devices in the device 1 14 and 117 is used, performance may be observed to substantially reduced it can. It was but connexion, wherein for the entire potential instrumentation E and optimum KaSo ffi of a computational performance using only the same 1 means computing equipment and memory instrumentation a becomes insufficient. Therefore, by molecular modeling system of the present invention, selection of the calculation means poor computational performance caused to have been limited are overcome. Molecular model Chemical Kisupa Bok database apparatus 11 1, to help the user of nonexperts, in accordance with the intended model of y target the molecular Day evening for the purpose (molecular data object), the molecular potential di We Ne laser evening devices It makes it possible to determine the selection of ¾ optimal combination of molecular modeling distributed processing instrumentation s, memory instrumentation s :, and data mass ^ 憶 device.
The de Isseki input an operation input for the molecular modeling system, keyboards  de device 1011, the graphic table device 1012, a screen manpower instrumentation E 1013, Contact and mouse input device 1014 includes an input device 101; memory instrumentation ¾ 114; de Isseki mass storage device 115; and network Bok workpiece filesystem 11G; or may be provided by these combinations, input by said device, other devices molecular modeling system by a common data / address bus 1 13 sent to the class (Figure 1).
Data output of the molecular modeling system typically output device 106 (FIG. 1) as well as the individual output device 1074 (FIG. 20), is provided by the 1084 (FIG. 21) and 1094 (FIG. 22).
[Example 2]
Other embodiments of the present invention, FIG. 1, 2, 3, 7, 8, 9 and 12 refer to will be described below. Other than 107, elements of molecular modeling systems are described in Example 1.
Molecular Dynamics device 107 (FIG. 3) receives an input from the common data Z address bus 1 13 or, et al., As well as instrumentation S 108 and device 109 (FIG. 2). Molecular dynamics inputs are processed by the data input instrumentation ¾ 1071 (Figure 3). The input data of molecular dynamics device, parameters for the molecular dynamics is given, the calculation means is selected. The molecular dynamics, according to the expanded selection and Shikashirube intermolecular interactions, depending on the tracking device 1072 potential device 1073 (FIG. 3), to timely deploy the molecular input coordinates (propagate molecular input coordinates in time). The potential device 1073, to calculate the forces acting on the molecules, the calculated force, tracking device 1072 is deployed at suitable molecular coordinates using. The expanded coordinate, potential device 1073 is used to calculate the forces acting on the molecules at the time of deployment. Then the force acting on the molecules at the time it was developed, the tracking device 1072 further molecular coordinates using timely deployment, the further passes again instrumentation g 1073 the expanded molecules 应標. By the cycle of the force acting on the distributed by phase ¾ action meter ^ and then expand the molecule coordinates, the molecular coordinates can be timely deployment continues. Timely development of molecular coordinates is continued until the conditions defined by human power parameters Isseki of molecular dynamics is satisfied. The output of the molecular dynamics device is processed by the output instrumentation S 1074 in accordance with the input specifications apparatus 1071 provides.
Potential device 1073, a means of interatomic potential and the atomic force calculation of molecular and II Bei, Renner de one job lens ¾ 10731 (term stabilization proportional to the sixth power of the reciprocal of the interatomic distances and 1 2 between the resulting atomic potentiometer down sheet catcher Le consisting term destabilization proportional to multiplication and then atomic force) (E.Lennai'd Jones:.. Proc.Roy.Soc (Lond), 106A, 441 ( 1924);.. Proc.Roy.Soc (Lond), 106 a, 463 (1924)), instability estuary proportional to 1 square of the inverse of the original 7Q distance Buckingham type 10732 (Lennard Jones potential a potential and original while force resulting therefrom which recombinant can ffl an exponential function of the inverse of the interatomic distance), Institut Li Nga  Uweba type 10733 (were especially developed for molecular dynamics calculations Si, interatomic distance only not angle to make a 3 atoms (bond angle) also interatomic potential, and then the resulting atomic question including as a parameter Force), evening one software (Tersoff) 10734 (formally is represented by the sum of the potentials in which only distance between atoms as a parameter, then the interatomic potential was that its intensity varies depending peripheral 1 * 1 environment the resulting atomic Q. force), in addition to the conventional potential because representing embedded atom (embedded atom) type 10735 (atomic force in the bulk ^, atomic questions force at different surface and grain boundaries of the bulk ^ properly Sakai to determine the in place with a ¾ inch density of where nuclei are present, Po Te down sheet catcher Le and it Kaka et resulting et been Ru atoms query tailored consider energy when inserting the original f nuclei power) (MSDaw and MI Baskes: Phys.Rev.B, 29, (5443 (1984)), tightly coupled (tightbinding) type 1073G
(Crystal structure atomic of atoms on the question potential was out how to calculate the electronic state of the entire crystal by using the atomic orbitals and atomic force) (things 现学 dictionary edited by the Editorial Committee: physics dictionary (Baifukan, 1992) see section "bands theory"), molecular orbital type 10737 (only with the results of molecular orbital method is a method of calculating the electronic state of the molecule, interatomic potential determined without using the experimental values, and then give al atoms Q force) to (Kazuyuki Hirao, Kawamura Oiki: materials design by computer, P.64 68, (Tenohirahanabo, 1994)), custom Mei de (custom made) type 10738 (specific a specific substance physical properties between developed nuclear interatomic potential, and then the resulting atomic force so as to reproduce the experimental values with high accuracy) (Kazuyuki Hirao, Kawamura Oiki: PC material design by, P.5960, (palm Hanabo, 1994)) and type of potential device of the present invention 1073 9 or using the potential of a combination of the type is selected potential device (Fig. 7). By selecting the potential instrumentation E, the efficiency of the molecular modeling of molecular modeling systems, precision and a problem that reliability is insufficient is overcome. This is because available most suitable type of potentiometer Shall instrumentation a relative individual molecules question interaction between the components of molecular. For example the noble gas atoms and the semiconductor of the original  {■ molecular dynamics simulation of problems including, 15 t appropriate and accurate potential device of a rare gas atom (Lena one dough Johns potential I 10731 Ga举Ge be) selected, the most suitable and ίΗ 砘 potentials device to the semiconductor atoms (Instituto Li Nga triple potential device such as eleven Weber potential device 10733 and the like) is selected, and a rare gas atom and ^ conductive atoms it can be described most accurately by selecting the most suitable and accurate Potensha Le device described interaction (Buckinghamtype potential device 10732 may be mentioned). Selecting only one of the potential device to describe different types of interactions the, as seen in the second column 12, the maximum error is increased and reliability is lost. Killing also correct potential device for all the different interactions, that is, selects a potential device 10737 of molecular orbital or socalled ab initio, calculation costs unnecessarily very high (Figure 12 Column 3). For all the different interactions, selecting only the other union Align or one of the potential device potential device, the reliability of the molecular dynamics prediction is lost because accuracy is insufficient (Figure 9 ). Thus, by selecting individual most appropriate potential device, it is possible to achieve ^ have computational efficiency.
Tracking device 1072 has Bei means for calculating the molecular coordinates timely deployment (FIG. 8), Baretsu bets (Verlet) type 10721 (continuously you change position in reality, velocity, acceleration, small how to replace the equation of motion in the difference equation as changes stepwise at time intervals) (L.Verlet:. Phys.Rev, 159,98 (1967)), the predicted corrected (predictorcorrector) ¾ 10722 (movement direction ¾ formula recombinant Ru method) placed differencing equations (M.Haile:?. Molecular Dynamics Simulation, 159163 (John Wiley & Sons, in, 1992)), Te one Raeku Supanshiyon (TaylorExpansion) type 10723 (location, etc. If the ^ next derivative with respect to time of it is found analytically, finding a solution of the motion equations using the formula of the Taylor expansion) (H.Margenau and GMMurphy: T e Mathematics of Physics and Chemistry, P.483484 (D .van Nostrand Company, in '1956)), Runge one Kutta (RungeKutta) M 10724 (normal differential side Wherein the solution method) (H.Margenau and GMMurphy:. The Mathematics of Physics and Chemistry, P.486487 (D.van Nostrand Company, Inc, 1956)) and scum Tamumei de type and development of the combination of the components tracking instrumentation a using is selected. By selecting the tracking device, the efficiency of the molecular modeling of molecular modeling systems, accuracy and ^ Robustness is a problem that it is insufficient is further overcome. Because, in the case of the desired level of molecular dynamics in the study of the development of molecular coordinates is because it uses the most appropriate tracking device. The output device 1074 processes the molecular dynamics output data according to the parameters input device 1071 provides. Data and graphics to be printed is processed by apparatus 10741, data and graphics to be displayed are processed in unit 10742. Data and graphic and view is made printed becomes an output between time actual and processed by unit 10743, and the set time the output is processed by the apparatus 10744.
[Example 3]
Other embodiments of the present invention, FIG. 1, 2, 4, 10, 1 1, 12, 13 and 21 refer to will be described below. Elements other than 108 molecular modeling system is described in Example 1.
Molecular mechanics device 108 (FIG. 4) and a device 107 from the device 109 from the common data address bus 1 13 (FIG. 2) receives an input. Molecular mechanics inputs are processed by the data manually instrumentation fi 1081 (FIG. 4). Te cowpea in molecular mechanics input data, parameters for molecular mechanics calculations can be obtained, and calculation means is selected. Molecular mechanics instrumentation I £ 108 is by the potential apparatus 1083 an optimization device 1082, according to the selection and optimization of molecular interactions, to optimize the frequency T · coordinate input structure (Fig. 4). Molecular potential is calculated by the potential device 1083, from the molecular potential, and the means of optimizing Therefore, the force acting on the molecule, the molecular coordinates are optimized. Optimization device  by 1082, the optimized molecular coordinates are calculated. From the optimized pre _{t} coordinate, the force acting on the molecules in the case means molecules potentials, and optimization is required, is calculated in potential device 1083. The total ^ been potentials and the respective forces, optimizer 1082 to further optimize the molecular coordinates using, and then the optimum type and molecular coordinates of the interaction which is _{c} selected to pass the potential device 1083 that coordinates in accordance with a Potensha Le minute  this cycle to calculate the forces acting on f, it can be achieved by continued optimization of molecular coordinates. Optimization of molecular coordinates is continued until the conditions defined by molecular mechanics input parameters Isseki is satisfied. The output of the molecular mechanics device is processed by the output equipment 1084 according to the input specification of which is provided by the device 1081.
Potential device 1083 provides a means that to calculate the potential and forces between atoms in the molecule (Fig. 10), Lena one Dojiyonzu type 10831, Buckingham ¾ 10832, Institut Li Nga  Weber type 10833, Tasofu type 10834, buried atom type 10835, tightly coupled 10836, molecular orbital type 10837, Kas evening 厶Me I de type 10838, and potential device with a mold 10839 or potential combination of the Ingredients, the potential of the present invention are selected . By selecting the potential device efficiency of molecular mechanics of molecular mechanics system, a problem of low accuracy and reliability, with respect to the interaction of individual atoms between the components of the molecule, the most suitable type of potential instrumentation g it is possible to overcome it is possible to take advantage. For example problems of the semiconductor molecular mechanics of optimizing a molecular structure containing one gallium atoms and two gay atom may instrumentation S such potential instrumentation S 10839 of Institut ringer eleven Weber potential instrumentation K 10833 or the invention 1083 include triple potentiometer catcher Le device (Fig. 10). ½ it may also be most accurately described by selecting appropriate and accurate potential device. If only selecting one potential device for explaining a said different types of interactions, the maximum error is increased as a result, molecular structure molecular modeling KaSo R calculates very ίΐϊ Robustness is lowered. By selecting the most appropriate molecular modeling optimal KaSo Κ for each object of the structure β optimization, it is possible to JutsuNaru very large computational efficiency. Some optimum KaSo S is potential instrumentation is required to assemble the Sshianmatori box their respective to the primary and secondary derivative of potential for use in E, but the S optimization apparatus Lena one doughs John potentiometer Although it is suitable for use with tangential device 10831, it is the most unsuitable for use with the twobody potentiometer Shall device. Therefore, by selecting an optimization device for each potential device used, it is possible to achieve a greater computational efficiency. Figure 12 shows the performance that solve the problem of the molecular modeling. As a measure of its performance, showing the SiGaSi molecular structure optimization relative accuracy (the second column in FIG. 12). The relative accuracy indicates that achieve more excellent accuracy by selecting an appropriate potential instrumentation.
Optimizing device 1082 is provided with a means for calculating an optimized molecular coordinates (Fig. 11) Simplex (Simplex) type 10821 (2 divides the multidimensional space in Afain surface, searching the maximum value or minimum value direction, the procedure repeating, with the simplex method function to narrow the maximum / minimum area), generates a steepest descent (steepest descent) type 10822 (the objective function f (X) value point sequence reduced to less of one after another from the initial value of {Xn} in the socalled descent method to the differential direction of the objective function f (X) the point train generating direction, ▽ have opposite directions, i.e. ability to seek the minimum value taken in the steepest descent direction of f (X)) ( Masatoshi Sakakazu optimization nonlinear systems: P.73, Morikita publishing, (1986)), diving (Gradient descent) type 10823 (nonlinear equation which gives the optimality requirements, V f (X) = 0, the solution of Newton method to go directly obtained by Repetition rate law, quasiNewton method Which features that have a), with modifications thereto instead using simple gradient basevector as drop direction different from the conjugate gradient (conjugate gradient) type 10824 (steepest descent method, the by generating a conjugate vector direction has a function to continue to search the minimum point) (Masatoshi Sakakazu optimization of nonlinear systems: P.79, Morikita publication, (1986)), main door opening Polis' Monte Carlo (Metropolis MonteCarlo) 10825 ( metro Police San Carlo to probing the search direction based on the prering method) (S.Koonin, computational Physics, page 185: the Benjyamin / Cumming Publishing Company, Inc.), ι '' · ί speed Pena (Fast Penna ) type 10826 (main Bok Robo Risusanpuri packaging method improve, based on a shortened search time Penaarugorizu 厶) (TJPPenna, Phy.Rev., E51, R1 (1995)), custom Mei de! 10827 (user Your own Custom made optimizer) and optimization system using 屐適 of combinations of type 10828, and before Ϊ himself components optimization present invention is selected. By selecting the optimum KaSo S, the efficiency of the molecular dynamics of molecule modeling systems, will have the accuracy and reliability is insufficient problems are overcome. If what is because it is possible to utilize the most suitable type of optimization device for each object of Potential instrumentation H¾ Contact and modeling employed. By selecting the most appropriate molecular mechanics optimization system for each object of structural optimization, even only not select one optimization devices available, it is possible to achieve a greater computational efficiency. 13 before
, It shows the ability to solve 1d molecular modeling problem. It showed relative computational expense (12 column 3) using the optimum KaSo fi 10828 of the present invention (FIG. 1 1) as a measure of this performance, a more excellent efficiency by selecting an appropriate optimization device it is shown that can be achieved.
The output device 1084, the input So歡 1081 processes the evening molecular modeling output data according to the parameters provided. De Isseki a figure to be printed is processed by the Personal Protection for First Aid or Rescue Personnel 10841, data and graphics to be displayed are processed in unit 10842. Data and graphics to be printed and view is punished ¾ in device 10843 as a real time output, and processed by instrumentation fft 10844 as ¾ constant time output (Fig. 21).
[Example 4]
Other embodiments of the present invention, FIG. 1, 2, 5, 13, 14, 15 and 22 will be described below with more information. Elements of molecular modeling systems other than instrumentation S 109 is are described 突施 Example 1. Molecular potential generator device 109 (FIG. 5) receives an input from the common data / § address bus 1 13 and So匱 107 and 108 (FIG. 2). Molecular potential generator input is processed by the data input instrumentation fi 1091 (FIG. 5). By molecular potential generator input data, molecular port Tensharuji X Nere parameters evening calculations given and calculation means is selected. Molecular potential generator apparatus, to produce a molecule potential by optimizing the parameters Ichita potential interactions input Potensha Le inputs Isseki de potential device 1093 an optimization device 1092 thus the potential energy surface. Molecular potentiometer catcher Le is calculated by using the potential device 1093. The optimizer 1092, potential parameters one higher is optimized to the input potential E Nerugi one given. Its optimized potential parameters or al, molecular potential energy is calculated by the potential instrumentation fi 1093. The calculated potential energy, optimizing device 1092 is used to further optimize the input potential E Nerugi one given potential parameters Isseki, passing again the parameter Isseki the potential device 1093. This cycle of calculating the potential energy and optimization potential parameters Isseki can be achieved by continuing to optimize the potential parameters Isseki. ¾ optimization of potential parameters is continued until the conditions defined by the molecular potential di We Nere Isseki input parameters Isseki is satisfied. The output of the molecular potential generator device is processed by the output instrumentation R 1094 in accordance with human power specification unit 1091 is provided.
Potential device 1093 provide a means that to calculate the potential and forces between atoms in the molecule (Fig. 14), Lena one de In this device  John ^ 10,931, carbonochloridate Kkingamu type 10932, Institut Li Nga one  © E Ichiba first die 10933, data one source off ¾ 10934, embedded atom type 10935, tightly coupled 10936, molecular 軌追 type 10937, custom Mei de type 10938, and type 99/10939 also the potential of the present invention is a combination of the components potential device is selected using the potential. By selecting the potential device efficiency to adapt the molecular modeling potential of molecular modeling systems, the problem of insufficient accuracy and reliability are overcome. Because, for presenting the individual features of the molecular energy surface of the molecule Q interactions under consideration, ½ also because available potential device suitable type. For example, the potential energy surface de Isseki describing the interactions of the noble gas atoms, Lena  Dojiyo lens potential device is correctly fitted using 10,931, elements for example, a metal atomic potential energy surfaces de describing the interaction  evening is accurately fitted using embedded atomic potential device 10935. However, either adapt the metal atom energy data using the Lena one dough Johns potential device 10931, or adapt the noble gas energy de Isseki using embedded atomic potential device 10935, becomes insufficient accuracy, and potential the prediction function for performing the modeling of molecular dynamics and molecular mechanics is significantly limited.
Optimizing device 1092 provides a means for calculating the optimization potential parameters Isseki (FIG. 15), in the apparatus, the simplex type 10921, steepest descent type 10922, swoop type 10923, conjugate gradient type 10924, main Toroporisu Monte  power Norre port type 10925, fast Pena type 10926, the optimization device is selected using a mold 10928 and optimized combination of these components custom Mei de type 10927 and the onset Ming ¾ optimization. By selecting the optimum KaSo K, adaptation efficiency of molecular modeling potential of molecular modeling systems, ½ degree and dependable is a problem that it is insufficient is overcome. Since, in order to adapt the potential parameter for each potential device that will be employed is because it employs the optimization apparatus of the most suitable type. By selecting the most appropriate molecular potential generator les Isseki optimizing device for each object of the energy table surface adapted, other combinations of the optimizing device or available optimizer in selecting only once compared, it is possible to achieve a much greater computational efficiency. For example, one of the gallium atoms and the potential energy surface data describing the interaction of the two Kei atom may potentiometer Shall parameters Isseki optimization is most appropriate to explain the interaction between atoms for accurate question mark it can be most accurately approximated by selecting stages. And Such means, in the case of the above embodiment, threebody potential devices within device 1093 (FIG. 14), for example, stay ringer one Uweba  Potential apparatus of the potential device 10,933 or the present invention 99/10939 there Ru. With potential device 99/10939 of the present invention to be an optimized device 10928 of the present invention, the highest accuracy can be achieved set Bok optimization potentiometer interstitial parameters having reliability and performance. Figure 13 employ other optimization device to the problem of molecular potentiometer down Shall parameter generation, in the case of using the potential device 99/10939 of the present invention, the potential parameters  the evening performance optimization, accurate ab initio It is shown as a function of the relative CPU time relative deviation required person from potential energy table [. Figure 13 is a simple Rex apparatus 10921, using conventional standard optimization device used as dips device 10923, the selection of the optimizing device is limited or restricted, optimized been potential parameters set Bok of performance that precision andreliability indicating that the longer is and computation time low. Best performance of the molecular Potensha Ruparame Isseki set optimization problem can be achieved by using an optimization device 10928 of the present invention. 3Ί the highest molecular potential parameters Isseki set of optimized performance I or is independently a combination of potential devices and optimization device for generating a potential path lame Isseki
Can be 违成 by using molecular modeling system of the present invention to choose Te.
The output device 1094 wants to parameters provided by the input device 1091 for processing a Isseki molecular potential di ENE rate evening output data. Data and graphics you print is processed by instrumentation K 10941, data and graphics to be displayed are processed by device 10942, and the data and the figure for printing and display is processed by instrumentation K 10 943 as Outputs ¾ time and setting is processed by the apparatus 10944 as the Q output when (FIG. 22).
[Example 5]
Other embodiments of the present invention with reference to FIG. 1, 2, and 3 0 you described below. Elements of molecular modeling systems other than system 111 are described in Example 1.
Molecular models Kako Kisupa Bok database processing apparatus 111 (FIG. 2), the common de  evening address bus 113, and device 107, instrumentation S 108, and inputs the Isseki de than instrumentation fi L09 (Figure 2). Input processing, Ru is processed by the data manually apparatus 1111. (Figure 30) Molecular model modified Kisupa one Bok database apparatus 111, by means of expert database processing apparatus 1112 and expert database determination unit 1113, the molecular De and input object Isseki (molecular data object) and user input , according to the model of the target defined, it presents the best combination of molecular modeling device. Data input device 1111, the expert database determination unit 1113, the common data, give stearyl Isseki scan information for all devices in the molecular modeling system obtained through Z Adoresunokusu 113. Ekisupa one DOO database determination unit 1113, and input data from the input device 1111, Ekisupa one preparative database processing instrumentation  to handle more given data in 1112, to determine the ¾ optimal combination of molecular modeling system unit give a decision result to the data output instrumentation Te. Expert Database processor _{m 2} holds all the characteristic data of known chemical elements, and the characteristic data of the molecules molecular modeling is treated as a unit (single entities). Chemical element is held in the device 1 112, and the single  data about body molecule, atom chemical element names (the atomic chemical element name) and the symbol, molecule name (the molecular entity name) and its symbol Honoré, load (the electrical charge), magnetic 4 dogs status (the magnetic state) I O emissions state (the ionization state) energy, the polarity of the atoms and molecules (the atomic ana molecular entity polarization)> Hara   and molecules 軌坦 shape (the atomic and molecular entity orbital configuration) atoms and minutes Na semi ί St (the atomic and molecular entity radius), and Lee 匕学 elements, classical mechanics of molecular quantum mechanics relates characteristic data, and molecular modeling systems is all of the data concerning device. Ekisupa one DOO database determination unit 1113, molecular data to target, the user inputs, modeling goals defined, and each data unit 1112 retains described above, up receiving as input, Ekisupa one WINCH database in 1113 to determine the optimal combination of molecular modeling system unit by. Expert de in 1113  evening Ichisu is, Ekisuha Bok information (the expert information) and 1112 atomic elements that held by the original ^ chemical elements (atomic chemical elements) and molecules (molecular entities), molecular data, molecular modeling system de  based on evening, atom,  that the best way to model the interaction of the molecule is retained
[Example 6]
Other embodiments of the present invention with reference to FIGS. 12 and 23 will be described below. By molecular modeling method of the present invention, the problem of describing the nonuniform interaction polyatomic system as seen in the art of the present invention, terms and uneven threebody describing a heterogeneous twobody interaction the ¾ describing the interaction ^ be solved free using said Min, number form of interaction. Generics analysis functional of the present invention (generu; analytical functional) is convenient for use in the technical field of the present invention, without compromising in terms of computational cost, the other normal interaction functionals It compared a high accuracy.
The present invention, the problems in the art of the present invention arising from the functional of atomic phases JI effects commonly used, namely, the problem of high accuracy in describing the nonuniform interaction, calculates the interaction between atoms economic efficiency high Mel problems, and a new type contemplated the problem of designing a new interaction functional every interaction between heterogeneous atoms are overcome when. These problems are the input device 101, shall apply molecular modeling system comprising comprising a molecular modeling system processing unit 105 and molecular modeling system output device 106 (MMS), the molecular modeling system processing unit 105 forces molecules System molecular dynamics device 107, the molecular potential to produce structural molecular mechanics instrumentation S 108 to ft optimizing, and intermolecular interaction potential intermolecular interactions potentiometer Nsharuparame Isseki molecular system for tracking the question onset exhibition when any one "" 3 Jiwenere evening device 109 or molecular dynamics, becomes comprise a combination of apparatus molecular mechanics and molecular potential generator, the molecular dynamics ManabuSo E 107, intermolecular interactions means for selecting a question development tracking device when means and molecules for selecting potential or a combination of said means are provided, ^ serial fraction Ryo  mechanical instrumentation In 108, stage selecting means and molecular mechanics structure 敁適 apparatus for selecting intermolecular interaction potential, or a combination of said means, is set vignetting, the fraction? The potential generator instrumentation S 109, means for selecting the means and molecular potential parameters Isseki optimization So蹬 selecting the molecular question interaction potential device or £ Yu,  the combination of stages is provided, the molecular Q phase: ^ means for selecting an action potential device   one is selected and union Align similar and dissimilar types of illustrating the interaction of atoms three holiday heterogeneous atoms question interaction Potensha Le analyzer, or molecular question interaction potential device , and the one of the selection of highspeed nonpolynomial optimization apparatus optimizing means, or by molecular modeling system, which is a selection and combination of 铰適 apparatus, by utilizing the functional solve it Ru can be. Note the pan factor / _{2} of the twobody interaction functions and indicates the factor of threebody interactions in Figure 23, and the utility of these terms in overcoming individual terms and the problems of該汎function continue in this embodiment will be described in detail.
In M 23, the interaction term functional between the atoms (1), (2) and (3), HaraRyo  since only relies between distance represents the section of the twobody interaction. ¾ (1), because the physically meaningful exponentially represent repulsion between atoms cored (the inter atomic core repulsion), nonphysical representation of the atomic question repulsive force apparent Lena one Ton'yonzu type potential overcome It is. It parameters a, and bi is the intensity of the repulsive force, to determine the reduction of repulsive force due to an increase in interatomic distance. ^ Child between the attractive force, is dominated by the electrodynamic interaction with the quantum mechanical interaction of the electron core shells of atoms (the atoms' electron shell) (electrodynamical and quantum mechanical interac on). ¾ in FIG. 23 (2), since represents the polarization of the individual atoms polarization (atomic polarization) and the dispersed phase ¾ action (Dispersion interaction) by atomic shells (the atom shell), represents the main function of the twopiece original gamma! Attraction . Atomic polarization all SANYO specific to the type of individual single atom, which is generated by instantaneous relative rotation E of atoms shells electrostatic f (instantaneous relative displacement), is done by itself regarded major factor but should. Functional fullbody interaction (the total 2 body interaction functional), because the basic physics of twobody interaction ΓΗ probability than functional Lena one Dojonzu type cover Solo, quadrupole (quadrupole ) to be more consider the twobody interaction higher than the Buckingham type potential meter ^ gar layer economical. For different pairs of twobody interaction  a set of parameters Isseki _{ai,} b, and because _{Ci} is possible to specify the precision and adaptability to describe a heterogeneous phase II action is ensured. Section of FIG. 23 (3) beyond the two mutually acting atoms 1 and predetermined Chikara' Bok off radius depends on the type of j (cutoff radius), becomes zero when analyzed guarantee.
If the third atom is present, the change in strength of the twobody interaction, the triple phase of their respective 1: depending on the individual electronic properties of the on 3 侗 atoms to base that are participating in the action are doing.
A set of arrangement priority claim scissors angle of 3 atoms (Enclosing angle) (configurational preference) is square (5) with three arrangement angle being Inkurozu Te cowpea in each of the three atoms described by some angular dependence around (4), and the cosine of the angle are compared with the cosine values of the angular placement has priority (6). Only item (4) is essence i.e. preferential angle of threebody interactions generally a (the Preferred angle) cover one and, and thus to be varied angles, threebody interaction varies in some manner. However ^ the physical for twobody interaction in the presence of a third atom, the type of allatom, strengthened thirtytwo rest interaction holiday arrangement or weakened by distance and type of the third atoms since or are not able to cover one by an angle dependent term. Therefore, to ¾ each of the two other consider index ¾ distance for interaction atoms (7) is an original 7 and physical properties dependent changes to the distance of the triplephase S action. Specific function term including an activeoff radius (8) shows a limited range of the three rest interaction. When two atoms of threebody cross the action being discussed approaches, each of twobody interaction, as in the case of strong bond dimer holiday molecules (dimer Molecules), now contributes primarily as a result, Sankyusho action is likewise so is best described by pan 閱数 limiting the impact and intensity range of threebody interactions in the short distance, different types ::. body coupling the description can do. This limitation is described in the section of Figure 23 (9). Thus, fullKaradaHiroshi function (the total 3body functional) is Instituto ringer  © We problems caused by a bartype functional, that is, triple claim tends Ru distort twobody section , Ru can overcome the problem of causing additional accuracy problems in the accuracy of the functionals indicating their two bodies have been limited. All atoms question interaction functional of the present invention (the total atomic interaction functional) is different orders (order) at Institut Li Nga one  contains functions © E same general type as used for ICHIBA IchiHiroshi function cage, designated function is reused for different ¾ of the functional of the present invention, and the number of adjustable parameters like the stay Ringer one Webertype functional small, illustrating the interaction between heterogeneous nuclear higher precision for the functional to the computation cost can be achieved comparable to stay Li Nga eleven Webertype functional (Figure 12). Thus, the functional interaction between heterogeneous atoms present invention, it only its general form to a specific union Align the nonuniformity of interaction without or atoms to compromise on computational expense is valid without compromising provide high accuracy comparable to the interaction functionals are usually needed for.
[Example 7]
Other embodiments of the present invention, FIGS. 3, 7, you described below with reference to FIG. 12.
According to molecular modeling method of the present invention, the problem of describing the interaction between heterogeneous atoms Gariu 厶原 terminal and two Kei atom UNA one by are _¾ to the technical field of the present invention, the present invention It is solved in molecular dynamics device 107. Ie, by calculating the force acting on the molecule atom molecular dynamics tracking device 1072 of the molecular dynamics device 107 calculates the timely development of molecular 应標, potential treatment of the molecular dynamics device 107 repeating a cycle of calculating the forces acting by interaction jfl potentiometer catcher Le in molecular original T in instrumentation ίδ 1073, to calculate the time evolution of the molecular system by timely deploy molecular coordinates. During this calculation, calculate the total interaction energy between atoms is one triple uneven atomic interaction analysis apparatus _{10} 739 force similar and dissimilar types of atoms of selection means of the potential processing device 1073 to.
The total interaction energy between the atoms is calculated by twobody distance (r) dependent interaction and triple distance and angle dependent interaction. Said a twobody phase interactions, six functional between two arbitrary nuclear  has (r ^{6)} index and the inverse of the force versus distance dependence, the type of the parameter of the twobody interaction atoms depends on the exponent interaction represents an atomic repulsive force caused by each atom core was parameters Isseki of r ^{e} term denotes variance electronic wick Thus occurs the polarization of each atom, further , subtends an angledependent trigonometric functions that are weighted by the distancedependent exponential said triple function term is weighted distance is made, parameters Isseki of the trigonometric function and exponential function, of each atom scissors angle of choice of the structure of the mold and threebody, and set cutletoff radius exceeds the twobody and threebody which weighting has been performed by the distancedependent exponential for controlling the succession of Chikara'off analysis of the interaction It relies on functional That.
The total interaction energy of similar and dissimilar types of atoms, the sum of the threebody interaction for two bodies ^ interaction, the apparatus atoms 应標 and from the parameter ^ is JT  is provided in the input, All phases S acting energy U is the functional form of the following similar and dissimilar 'type atoms:
Shed = Σ /, + Σ (number ^{1)}
In the above formula, the term twobody sections and threebody is
Arm _{ = [a; exp (br; } )  c r '6  exp ((r y  / ·,) "') ( Number _{2) 3 = h {r ti} , r ik, ω ι) + h (r _{ ti, r jk, ω; ι } ,) + a _{ h (r ki, r kj, } ) ( Equation 3), in the section,
so
Oh λ
NOTE: Ri ((
hi r .., r eleven,.,, ω ... λ.. exp λ ,, r, exp
ij ik This λ ijk, UU ικ IK
This
2
cos ω ... + α. (number 4), \ ≠ l '
so,
γ .. = γ .. (5)
! J β
In it, omega atomic i calculated from the atomic coordinates of the input is enclose angle atoms j and k, r _{tj} is the distance between the〗 the atom i to be calculated from the atomic coordinates of the input , the parameters a ,, b ,, _{Ci} devices Ri types can each atom i, the parameter scan i 7 i, is determined by the type of atoms i and j, and the ij interaction setting force Tsu Bok off the radius, parameters ai are calculated by setting the angle of i that defines the structure selection of configuration of the atoms.
Wherein each input parameter Isseki, the type of the physical properties of a given atom, and its Ri Contact illustrates the interaction of these atoms with similar or dissimilar types of other atoms, the values of the input parameter range is represented by the following formula.
8
_{. 0.1,10 eV, b j G [} 0,10 A 1
^{ 10 4, 10 8 ■ νλ} 6 ( 6) 4
λ 10 ^{4,} 10 e les ', gamma one ^{io J, io A ' 1,250} ] A · ( 7)
U Li Li
The input dependent variable triple heterogeneous atoms question interaction analysis instrumentation 10739 comprises three all Θ enclosed by atomic Question distance and three interaction for atoms, a _{jk.}
ΙΪΪ Symbol three holiday uneven ^ child in i. Manpower parameters of action analysis 10739 is, two atoms Kei ^ a is ¾ If one of ¾ J is in Gariu 厶型, Gariu 厶 and gay ^ of the two holiday When the three holiday: child liijffl interaction river energy ill] Thus the work, ie Ά 1 Kiii or socalled ab initio energy illl  the ίϊί "is the optimized parameters, and the human power parameters were columns ^ below parameter  is the evening. Replacement sheets (Rule 26) (8)
The number 6, the range of parameters according to the number 7, depending on a range of parameters _{ai,} the overall size of the atomic repulsive force is possible for regulatory all kinds of atoms; thus the range of the parameter b> reduction of repulsive force is adjustable; parameters Isseki size of the dispersed by the scope of the ci enters the physically significant limitations; by the scope parameter lambda, triple the original between the repulsive force magnitude, and three priority twobody molecules formed in the body arrangement, authored by pure twobody interaction terms  Ku can be revealed without interference; by the scope of the parameters Ta, and increase of the force in the threebody interaction reduction Ri adaptive der; and the range of the interaction by the scope parameter ¾ is a range that can reveal the internal properties which occurs in a long range of processes and atoms of a single atom.
As described above, nonuniform  it is possible to overcome the appropriate parameters Isseki relative atomic threebody interactions select cell Tsu preparative find problems. Function form shown by _{c} number to several 5 that can explain the interaction between heterogeneous atoms of a single gallium atom and two gay atom, the number of举comparable to that of the stay Ringauwe one bar since it is necessary to compute the function (single function), functional on mutual reaction of the gallium and gay containing a functional of the present invention does not compromise on computational efficiency. Third column of FIG. 12 shows the efficiency of the calculation by referring to the efficiency of the functional of the present invention. Functional of the present invention has the same efficiency and stay Ringau Eba functional, a pure twobody half the efficiency of the Lena one Dogi Yonzu and Buckingham quadratic functionals, and first principles calculations it is about 1000 times the efficiency of the (first principle calculation). The parameter set Bok shown between the number of shape and number 8 shown in equations 1 to 5, it is possible to achieve high accuracy in describing the interaction prior ¾ gallium and Ke I elements. In the second column of Figure 12, the accuracy of the functional of the present invention, is illustrated with reference to the first principle calculation, or socalled ab initio calculations. Accuracy of the present invention, 3fold compared with the Institut Li Nga eleven Weber type functional high and the net frame twobody pan 閱数 by remote 510 times higher. Accordingly, the functional form of the present invention, while maintaining the computational efficiency high t, the problem of imparting the accuracy is overcome.
[Example 8]
Furthermore, a description will be given of another embodiment of the present invention below with reference to FIG. 24. By molecular modeling method of the present invention, the path lame one data space when optimizing Isseki parameters as possible out to overcome the problem of local minimum point occurs. Since the optimization algorithm to overcome the problems fall into local minimum point, the present invention uses Aniru method (the simulated annealing technique). Since § Neil method converges Yutsuku litho or overcomming clothes from low efficiency problem other methods, Cauchy b les Ntsu distribution relative Bijite ranging distribution of these Nog ra chromatography data (visiting distribution) (CauchyLorentz distribution) using the convergence from being proportional to the logarithm of Aniru temperature, improves in proportion to the inverse of Aniru temperature. Said receiving of parameters are selected from the distribution Isseki is controlled by the general tortoise Bok port Indianapolis algorithm, Aniru temperature of the algorithm is disconnected from Aniru temperature Bijiti in g distribution parameter Isseki (decouple). Since all parameters Isseki overcomes problem of Bijiti ring distribution of the same, these parameters  evening is selected from copolyesters one sheet eleven Lorentz type distributions, initial of the initial temperature distribution in the parameter Isseki since the clear function test values, all parameters Isseki performs its individual distribution. The pansensitivity function is optimized (the sensitivity of the functional to be optimized) is, ί solid people of Roh, of the parameters / ", different for the fluctuations, and of the individual parameters Isseki within space since the number and depth of the local minimum points that are cover one by the individual jump path lame Isseki is different from the other parameters, the algorithm of the present invention, Te to base a single parameter, the respective parameters of Isseki so that according to parameters taken topologies are brother from a perspective view. This feature, overall efficiency of the reliability and optimization ¾ optimization increases. This feature thus be achieved thereby proportional Bijiti ring distribution parameters to each jump success rate (jump success rate). As a result, for each parameter, independent of § Neil sequence of other parameters and other Roh. Since Bijiti ring distribution or we separate § Neil sequence of parameters is performed, for each parameter, each Kuen quenching time scale of that (quenching time scale) is Ru performed. As a result, some parameters are highly efficient when examining becomes therefore parameters Isseki space to converge much faster than other parameters Isseki, because it provides more space other parameters Isseki , and the procedure of g optimization becomes better crab responsiveness spring to the feature (global features) of spherical parameter space, and many of the parameters and various types of even handle complex functionals of different functions to become, and general purpose optimization ffl as used Applications  becomes scrap powerful. Its Arugori efficiency of rhythm and intensity (robustness) in order to further improve, for each lame Isseki, its individual generalized main Bok port is initialized by a function that depends on the initial test value of parameter one data since according to the police algorithm, so they respond better to different roles of parameters Isseki in to be optimized functional. Strength and efficiency of the algorithm 厶 Furthermore, by the by connexion visiting distribution function on the convergence behavior of the individual parameters are adjusted, further increasing. This feature, to execute the optimization algorithm 厶 a loop to adjust the distribution function by starting at a current optimum parameters by refreshed Aniru temperature is the current optimal parameter set function Bok, therefore parameters  that the evening large and deep local minimum point let more efficiently examined local parameters Isseki space even be realized by allowing overcome. If the distribution function is adjusted in a loop parameter Isseki is because it is possible to scan one layer faster parameter space can converge faster than the case of no loops. Strength of the optimization algorithm is increased by the performance of the loop, since quickly overcome local minimum point without external intervention, the optimal Kaa Rugorizu 厶 is selfregulating.
By way each item, Bijiti ring probability distribution of Cauchy one bite Lenz indicated by閱数the probability of changing from the test value X parameter to X _{t + 1} is I衷by the following formula.
^{1} XVX
_{ . (■ l + (/ r }  l) j + [(Dl) / 2]
(,  l)) 7 ( / x) 2 / (3  1v ( number 9) where T _{x} (t,) is, of all, lame Isseki and distribution fffi base click Bokuru ^ . Λ 'is its ί I_々 time scale of indicates that the S protect, and D is of come people of parameters Isseki scan Bae one scan,
! Two's, _{+ }  (Number 1 0) is of can person of each parameter Isseki test jump, and q _{v} controls the form of code replacement sheet (Rule 26) System bite Lenz distribution. Individual time scale t, is set according to the flow diagram of the algorithm of FIG. 24. The probability of the individual general turtle Bok port Indianapolis parameter acceptor is represented by the following formula.
, 1 / (1
Px (t, = minJl,  () (y (x tr + \) . Y [x,  ... ( Equation 1 1) where y _{(Xu)} is in parameter Isseki setpoint _{Xtx} at time t _{x} it is a sum of squared errors. I, is the individual general turtle Toro Indianapolis parameters § Neil.
parameter range of q _{a} avoids instability given and numerical sufficient adaptability fast Aniru
q _{a} e [10, a l) (number 12).
[Example 9]
Furthermore, other 突施 aspect of the present invention with reference to FIG. 25 as in Example 8 to explain below.
By molecular modeling method of the present invention, the local extreme small point parameter one data space is, it is possible to overcome the problem that caused the convergence speed and accuracy Aniru. To optimize the parameters to overcome the problem that is trapped in a local minimum point, the probability of the parameter distribution is quenched according to individual parameters Isseki Aniru temperature Tx that by the following equation.
7 () Double  .gamma. (number 13)
1
Here Tx (t,), all parameters and distribution iS base click Bokuru ^  X indicates that Megumori its individual Yui厶 scale t, a and Tx "respectively HatsuTomo 试験 parameters is an individual output ¾ which parameter of Isseki a of 1¾] number Isseki. parameter q _{v} controls the eleven Lorenz worth particular Cauchy. _{q,.} the parameter Isseki Tsutsushima,
v [2,3) (number 14)
In and, q,. = 2 indicates a cone eleven Lorenz min ^, q,> '2 is Ikubunhi replacement sheets (Rule 26) 41
Zunda co one sheet shows an bite Lenz distribution, numbers in parameter Isseki the range 两内 is stable.
Parameter Isseki individual distribution probability are distributed Aniru temperature and q,. By co one sheet one bite Lenz distribution mode control, and the Π himself regulation loop and loop initialization test Roh, a function of ° lame Isseki the Cauchy by a bite Lenz distribution § Neel temperature and generally turtle Toro Indianapolis § Neel temperature, optimum catheter order is strongly reliably performed and the local causes problems are overcome. Figure 25 shows the strength and efficiency of the optimization method of the present invention as compared to the highspeed algorithm of Pena. Functional: y (x) = At smiBi X) four parameters _{ A 2 cos (B 2 x) } , the previous functional: y (x) = 21 sin (3.3x) + 3.2 cos (l.lx ) (optimization was carried out for 100 function values equal) to all starting test parameters 1. This calculation was carried out in IBM compatible 486 / 100MH Z Personal Computer. Further Figure 25 shows the optimization of the accuracy of the function of the relative deviation as a function of the CPU time required is achieved, the relative deviation are then Bok plotted in logarithmic scale. The Pena method is converges fast, even in the case of normal functional used triangular function number, the accuracy is limited, a limited accuracy is maintained for a long time. The behavior of Pena method is due to the fact that capability to overcome the local minimum point is limited. 25, the optimization method of this invention, and indicates that converges quickly to parameter one Tase' Bok three digits of accuracy. Therefore, the optimization method of the present invention is to overcome the problems local parameter Isseki minimum point causes, and can be effectively applied to functional of trigonometric functions. Since the optimization method is effectively overcome the local minimum point of the present invention, user interaction is negligible and well so the user one skilled is not required very little when using this method.
[Example 1 0]
Further another embodiment of the present invention will be described below with reference to FIG. 26 and 突施 embodiments 8 and 9.
By molecular modeling method of the present invention, further it can overcome the local extreme small point in the parameter one data space is a problem of causing relative convergence speed. Since the convergence of Aniru optimization method overcomes the problem of slow, parameters to be optimized  each evening, the force was the individual receiving probability Aniru temperature (acceptance probability annealing temperature), 'connexion, individual general turtle Toroporisu receptor probability is assigned. With distribution Aniru temperature generally turtle Toro Indianapolis § Neel temperature of the individual Cauchy one bite Lenz is a function of loop initialization test parameters, from Rukoto high efficiency is obtained in selecting the test parameters Isseki, You can achieve the convergence of high efficiency. Figure 26, a highspeed convergence and accuracy of the ¾ suitable process of this invention, in comparison with highspeed algorithm Pena. Gaussian functional:! Y (x) = A exp ( the two parameters of x ^{2} / (2B0), the original functional: y (x) = 210 exp (x 2 / (2 1 1.45) ) (optimization was carried out for 100 function values equal) to all starting test parameter Isseki one. this calculation was performed in IBM compatible 486 / 100MH Z Personal Computer. Figure 26 is a fragmentary the CPU time shows the accuracy and convergence speed of the optimized achieved are as a function of the relative deviation beauty as a function, the relative deviation is shown on a logarithmic scale. this CPU initial error according to the optimization methods is within 20 seconds of the invention time 5digit decrease Suruga, the fast Pena algorithm, the initial error is only to a small threedigit decrease in the CPU time. before ¾L 20 seconds after, both methods have converged to its final accuracy. Accordingly the present invention the optimization method of effectively the functional functionals and exponential function of overcoming vital Gaussian low computation problems The optimization method of use as possible. The present invention performs very efficiently computed to achieve high accuracy in a short time, and so overcome the local minimum point in 冇効 skilled users who need the ½ Small in is suitable for use as an optimization tool can be used.
EXAMPLE 1 1] Further, the embodiments of the present invention FIG. 27, described below with reference to the embodiment 8, 9 and 1 0.
By molecular modeling method of the present invention, when applying the optimization method to complex functional, Ru can overcome the problem of convergence is slow and low precision. The optimization method, when applied to complex functionals include a number of different functions and the optimization number of parameters one should evening, to overcome the problem of low convergence is slow precision, the optimization method of the present invention molecular modeling system for implementing the optimal molecular mechanics device 108, one molecule coordinates using an optimizer molecular question interaction potential was used and optimizing device 1082 from molecular mechanics potentials device 1083 for molecular energy the optimized structure of molecular systems Te cowpea to reduction calculated; and the 设適 apparatus
1082 One of the optimizing device of the individual optimized parameters timescale Contact and Cauchy one  to calculate the optimized structure of the mouth Lenz test parameters optimized distribution by connexion molecular systems to individual parameters Aniru temperature optimum of necessity is calculated by the general tortoise Bok port Indianapolis acceptance probability, have but the generalized receptor probability has individual § twentyone Le order for each initial input optimization parameters, the Aniru optimization test parameter distribution It is controlled by the general turtle Toroporisuarugo the rhythm, and co one sheet one bite Lenz test parameters Isseki distribution Aniru is Ri by the individual test parameters timescale and Aniru order to recover an optimized initial test parameters .
The individual Aniru temperature of Cauchy one bite Lenz distribution, is quenched in proportion to the force of each of the paths lame reciprocal of Isseki timescale initial § two Lumpur temperature initial optimization study parameters Isseki function , and the quenching of the parameter distribution occurs in the general turtle Bok Ropo squirrel algorithm parameters Isseki received within autoregulatory parameter distribution loop, and generally turtle Toroporisu Arugorizumu individual parameters Isseki Aniru temperature of the individual parameters Thai It is proportional to the force of the reciprocal of the Musukeru.
Said optimized initial test parameters is in the selfregulating parameter distribution loop, as well as the individual Cauchy one bite Lenz distribution § two Lumpur temperature and individual general turtle Toro Indianapolis § Neel temperature and acceptance probability O connexion, it is possible to overcome the problem of low convergence is slow and accuracy in the case of applying the optimization method in the complex functional.
In Figure 27, the accuracy of the optimization method of the present invention the computational efficiency, optimize the ab initio data set Bok with 600 de Isseki point parameters of complex functional according Te Example 7 Odor compared to other methods in the examples. Incidentally, in this embodiment, initial parameter Isseki departed as follows.
._{ G "= 10000 eV, b Co } = 10 ', c c" = l eVA 6,
_{ a Si = 10000 eV, b Si } = 10 1, c Si = \ 00 e VA 6,
(Number 1 to 5)
, = 1 0 I Y _{Sl} .s, = 1.0 I = 6.1 do recalculation was performed at IBM compatible 486 / 100MH Z Personal Computer . Figure 27 shows the convergence and accuracy of the resulting optimized as a function of the relative variance (5 as a function of time. Note (simplex method how of Aru. Present invention shown on a logarithmic scale, descent method, compared to classical Monte Carlo method and Pena method . simplex method and descent method, the optimization parameter that is a classic seen from FIG. 27 is less precision and strength because they do not easily captured by local minimum point recovery. main Toroporisu optimization method is somewhat overcome the problem of the local minimum point converges slow. Pena method than Shinpuretsukusu method converges fast to Luke. However descent method and the classical main Toroporisu method likewise local minimum point is limited in accuracy because occurs after a short time. before Symbol optimization method of the present invention is to converge faster than the method of Te to base listed above, and other methods of the are in to Have improved the accuracy and held after a period encountered problems of local minimum point, steadily converged to. Therefore the optimization method of the present invention, calculation efficiency of problem and local minimum point occurs less to overcome the problem, or one many can be applied to 冇効 complex functional with parameters Isseki be different functions and many optimizations. Therefore, the optimization method of the present invention, the user since the intervention of a skilled can be applied β have efficient complex functional with little need be used as a general optimization and calibration tool.
[Example 12]
Other embodiments of the present invention with reference to FIGS. 11 and 28 as well as the embodiment 8, 9, 10 Contact and 11 will be described below.
Ballistic, as seen in a method for predicting the trajectory control, some of the dependent variable to optimize the unknown functional initially having the problem, the initial W are optimized device unknown functional in (400 ) be solved by. This problem is solved by utilizing a parameter optimization method of the present invention a parameter optimizing apparatus (Figure 1 1) to the device 400. Data sensor 401 forces ";., Early, such as missile ballistic To find characteristic data of the process with unknown functional dependency related data sensor (releted data sensor) 402 is a fast optimization, the control and prediction Search for useful relative parameter Isseki data. Isseki data de that sensor 401 is collected is placed in the data buffer 403, data collected related data sensor 402 is placed in the associated data buffer 404. function analysis instrumentation S 405 force, 'by set Bok primitives device of the apparatus 405, to approximate the data set of de Isseki buffer 403. in this case, function need use in the process of optimization and control using the optimum KaSo »1093 optimized, and apparatus 403, 405 and 1093 are connected by the data Z Adoresubasu device 409. in the following, the data sensor 401 inputs the initial unknown functional dependencies functional form, associated de One evening used with data from the buffer 404, to predict the data value in the setting 捋来 during interrogation of the process being monitored (preset future time) by the sensor device 401 and 402. Then, before controlling the predicted data value using the apparatus 407. as controller 407 by treatment ¾ front ¾? measurement data value, the operation control system 408 output device or the predicted data value to be controlled to a predetermined ¾ exactly the process 41 to 1 can Oku违 in. the operation control system 408 is to by connexion connected to the control data / Adoresubasu instrumentation IE 410, to deliver the output Ru piece j Irare to control a subsequent process or direct process . in the realtime control process such as emergency aircraft control systems or antimissile system, the initial unknown functions 従厲 resistance, very fast and very accurate process It is essential to provide a control means. Online optimization system process or control, if not provide sufficient optimization convergence speed, it also predicts that instruct proper respective control and the optimization process not even. If the initial unknown functional dependency parameter prediction is not sufficiently accurate, by the optimum KaSo location of the process control is sometimes it is not possible to instruct proper dangerous processing result occurs. the present invention, an initial unknown can achieve very ^ have the speed and accuracy of the functional dependency parameters Isseki prediction, it is possible to overcome the problems occur because the convergence speed and accuracy of 硗適 reduction is insufficient.
[Example 13]
Other ¾ embodiments with the present invention, FIGS. 11 and 29 as well as the embodiment 8 will be described below with reference to (J, 10, 1 1 and 12.
'Kaguri head load prediction and as is 兌 the respective control processes, some problems of optimizing the parameters of the initial known functional of dependent variables, for optimizing device of the initial known function ffl (500) It is solved by the optimal KaSo S. This problem is solved by utilizing the method of the present invention a parameter ½ optimization device (Fig. 1 1). Data sensor 501 Ca ^ E Kaguri head ^ find the characteristic de one other known functions 従厲 of the process, such as load patterns. Related data sensor 502, speed optimization, searching related parameters Isseki data 冇用 to control and predict. Data sensor 501 Jie of Xia meth de Isseki is placed in a data buffer, data related data sensor 502 is collected takes into related data buffer. According to the data provided by the data / Adoresubasu device 509 device 503 and 504 via the optimum KaSo location 1093 to optimize the parameters Isseki known functional dependencies provided by the function device 505. The output of the optimizing device 1093 provides data to the controller 507, then the controller 507 calls the prediction control process and the process by delivering each de Isseki output to the operation control system 508, Mr. Su 508 Te is connected to the control data / Adoresubasu device 510, it delivers an output used to control the subsequent process or direct process. In the realtime control process for example power Dali head control system or conventional reactor if Ku fusion reactor, it is essential with a very accurate means very fast known functional dependence of the process control. If the online process if clause optimal KaSo 5Ϊ of control does not provide a sufficient optimization convergence speed, the process of control and optimization of their respective, process is not performed also be properly instructed also predicted. If known functional dependence of parameter prediction is not sufficiently accurate, process control may Rukoto properly because instruction can not be dangerous processing result Oko. Optimization apparatus of the present invention, it is possible to achieve very high speed and accuracy of the known functional dependency parameters Isseki prediction and optimization of the convergence speed and accuracy occurs because insufficient problem is overcome be able to. Optimization apparatus of the present invention, it is possible to achieve high speed and accuracy the initial unknown functional dependencies parameter prediction, and the convergence speed and accuracy of the optimization takes place due to insufficient problems can be overcome . Select interaction potential between proper 3 molecules relative availability individual interactions industrial, and selects the deployment scheme of more hours and molecular coordinates BunRyo  depending on adopting power, 'the molecular modeling systems and modeling methods invention, other conventional ffl I, its dependent molecular modeling system and Takashi from molecular modeling techniques, adaptive, to achieve efficiency and reliability. Molecular Question interaction potential apparatus and method of the present invention, through its design, adaptability molecular dynamics, efficiency and reliability is further improved, processing with high accuracy and efficiency interactions heterogeneous molecule. Select the appropriate molecular question interaction potential describing the interaction between the individual molecules, and by take advantage molecular mechanics to select the appropriate highspeed molecular coordinate optimization scheme, molecular modeling systems and the molecules of the invention modeling method, molecular modeling systems and molecular modeling methods higher efficiency is another commonly used to achieve accuracy and reliability. Fasteffective optimization apparatus and optimal Kaho of the present invention further, the accuracy of molecular modeling of the present invention to improve the efficiency and reliability. Optimization apparatus and optimization method of the present invention may be utilized other than a problem of molecular mechanics and molecular modeling optimization, for example, known and unknown function number 従厲 of process control and process prediction. By selecting an appropriate intermolecular interaction potential describing the interaction energy surfaces individually, and utilizing molecular potential Genet rate evening to select the appropriate highspeed potential parameters Isseki optimization scheme, the molecules of the invention model systems and molecular modeling methods, and molecular modeling system used other normal minute? Higher than the modeling method efficiency and that to achieve the adaptability. By fasteffective optimization apparatus and optimization method of the present invention, the accuracy of molecular potentiometer Shall generator, efficiency and reliability is further improved, the reliability and accuracy of the potential parameters one had this generator is small total effort it becomes possible to provide the evening. Individual molecules Q phase: ft action Potensha le co one de writing before ^ Potential and Intafu I scan through to connect to the molecular modeling system, the molecular modeling system and the molecular modeling of the present invention since completely eliminated by the molecular potential di we Ne laser evening is part of the law, significantly reduced tumorbearing food for users trying'll solve molecular modeling problem.
The same molecular modeling systems and molecular modeling methods, molecular dynamics, by using molecular potential di We Ne laser evening molecular mechanics and the invention, molecular modeling systems and molecular modeling methods of the invention further to achieve the convenience for greater efficiency and user scratch. Since, it is any part or al may suitable designed with Flip in molecular modeling system using the appropriate knowledge with respect to the center of the modeling problems and user one problem molecular Question interactions.
Interaction generic Hiroshiseki number of interaction potential between devices heterogeneous molecules, and a method comprising a small number of functions different types with a small number of adjustable parameters using the present invention, and twobody and threebody atoms If different physical and chemical during interaction is described through a special form factor of a heterogeneous twobody factor and threebody of the functional, it is needed for normal to describe the interaction between atoms other much higher accuracy than the functionals are, can be achieved without compromising in terms of computational cost. Therefore, problems arise in the technical field of the present invention can be by a very high accuracy of all test results obtained, resolving without compromising in terms of computational cost. Furthermore, this by interaction generic function, because the cost to develop the interaction functional of cost to obtain a high accuracy and case design (Case the Designed) is applied, is going cane or supplement or as you'll conventional resolution without problems can and solve child. Moreover, problems arise when the semiconductor modeling of the interaction gallium and gay arsenide, precisely and also axioms are currently available can be calculated by cost efficiency have far from the socalled ab initio methods that. Heterogeneous matter Ryo of the present invention that Zui ¾ uneven molecular question interaction method of the present invention  by the potential device, molecular modeling systems and molecular model adaptability and accuracy of the Dell process of this invention is increased, the user It will be able to resolve the problems could not be resolved prior.
Furthermore, by himself adjustable optimization method of high speed of the present invention for analyzing any functional, the conformance tool that requires little user one intervention and user skill, Sofutouweatsu one Le such example Kaleida Graph (TM) , Cricket Graph (registered trademark), Dalta Graph (registered trademark), etc. Graphical User Interface (GUI) tools, and in the Wye Dosupuretsu de (Wide spread) nonGUI mainframe optimization tool, Shin currently used in common dependable low simplex method and descent method  instead of (personal computer, Wa seen in Kusutesho down), can be achieved by inserting into the graphical data display software Towe §.
Furthermore, a complex multiparameter Ichita functional device and method of the current adaptation optimization that could not be subjected to conventional optimization Because efficiency is insufficient, f ii Kikei 济的 optimization of the present invention it can be optimized by the device and optimization methods. Further optimization apparatus and optimization method of the selfadjusting fast of the present invention, HaraRyo  furnace, seen aircraft, missile tracking So匿, ¾ Kaguri head, in various ^ 3 Dai prediction and control of satellite such can be utilized for known and process control of real time interrogation functions 従屈 of Suechi and process prediction.
Claims
Priority Applications (2)
Application Number  Priority Date  Filing Date  Title 

JP14543096  19960607  
JP8/145430  19960607 
Publications (1)
Publication Number  Publication Date 

WO1997046949A1 true WO1997046949A1 (en)  19971211 
Family
ID=15385071
Family Applications (1)
Application Number  Title  Priority Date  Filing Date 

PCT/JP1997/001951 WO1997046949A1 (en)  19960607  19970609  Molecular modeling system and molecular modeling method 
Country Status (1)
Country  Link 

WO (1)  WO1997046949A1 (en) 
Cited By (3)
Publication number  Priority date  Publication date  Assignee  Title 

JP2005034843A (en) *  19980626  20050210  Toshiba Corp  Calculation method of reaction mechanism and recording medium with the calculation method recorded thereon 
WO2006102228A1 (en) *  20050318  20060928  Eve Zoebisch  Molecular modeling method and system 
JPWO2008041304A1 (en) *  20060929  20100128  富士通株式会社  Molecular force field allocation method, molecular force field allocation device, and molecular force field allocation program 
Citations (1)
Publication number  Priority date  Publication date  Assignee  Title 

JPH08161290A (en) *  19941206  19960621  Fujitsu Ltd  Potential library device 

1997
 19970609 WO PCT/JP1997/001951 patent/WO1997046949A1/en active Application Filing
Patent Citations (1)
Publication number  Priority date  Publication date  Assignee  Title 

JPH08161290A (en) *  19941206  19960621  Fujitsu Ltd  Potential library device 
NonPatent Citations (1)
Title 

PHYSICAL REVIEW E, Vol. 53, No. 2, February 1996, JURGEN SCHULTE, "Nonpolynomial Fitting of Multiparameter Functions", p. R1348R1350. * 
Cited By (5)
Publication number  Priority date  Publication date  Assignee  Title 

JP2005034843A (en) *  19980626  20050210  Toshiba Corp  Calculation method of reaction mechanism and recording medium with the calculation method recorded thereon 
JP4713856B2 (en) *  19980626  20110629  株式会社東芝  Reaction mechanism calculation apparatus and reaction mechanism calculation method 
WO2006102228A1 (en) *  20050318  20060928  Eve Zoebisch  Molecular modeling method and system 
US7797144B2 (en)  20050318  20100914  Eve Zoebisch  Molecular modeling method and system 
JPWO2008041304A1 (en) *  20060929  20100128  富士通株式会社  Molecular force field allocation method, molecular force field allocation device, and molecular force field allocation program 
Similar Documents
Publication  Publication Date  Title 

Pitzer  Bonding in xenon fluorides and halogen fluorides  
Bedaque et al.  Effective field theory for fewnucleon systems  
Groth et al.  Global three‐dimensional MHD simulation of a space weather event: CME formation, interplanetary propagation, and interaction with the magnetosphere  
Silverman et al.  The bootstrap: to smooth or not to smooth?  
Wetzel et al.  Reconciling dwarf galaxies with ΛCDM cosmology: simulating a realistic population of satellites around a Milky Way–mass galaxy  
Lattimer et al.  Neutron star masses and radii from quiescent lowmass xray binaries  
Kim et al.  Phase field modeling and simulation of threephase flows  
Neale et al.  Description of the NCAR community atmosphere model (CAM 5.0)  
Isukapalli et al.  Stochastic response surface methods (SRSMs) for uncertainty propagation: application to environmental and biological systems  
Cranmer et al.  Empirical constraints on proton and electron heating in the fast solar wind  
Paxton et al.  Modules for Experiments in Stellar Astrophysics (): Convective Boundaries, Element Diffusion, and Massive Star Explosions  
Claire et al.  The evolution of solar flux from 0.1 nm to 160 μm: quantitative estimates for planetary studies  
George et al.  A parallel 3D dendritic growth simulator using the phasefield method  
Blanchet et al.  Gravitational field and equations of motion of compact binaries to 5/2 postNewtonian order  
Astier  Can luminosity distance measurements probe the equation of state of dark energy?  
Garraffo et al.  Lagrangian data in a highresolution numerical simulation of the North Atlantic: I. Comparison with in situ drifter data  
Bourlioux et al.  Highorder multiimplicit spectral deferred correction methods for problems of reactive flow  
Henry et al.  Abundances of galactic anticenter planetary nebulae and the oxygen abundance gradient in the galactic disk  
Farooq et al.  Hubble parameter measurement constraints on the redshift of the deceleration–acceleration transition, dynamical dark energy, and space curvature  
Aubert et al.  A radiative transfer scheme for cosmological reionization based on a local Eddington tensor  
Jallouli et al.  Relativistic effects in the pionium lifetime  
Dasgupta  Interpolants within convex polygons: Wachspress’ shape functions  
Heitmann et al.  The coyote universe. I. Precision determination of the nonlinear matter power spectrum  
Morton et al.  The radius distribution of planets around cool stars  
Bovy et al.  A direct dynamical measurement of the Milky Way's disk surface density profile, disk scale length, and dark matter profile at 4 kpc≲ R≲ 9 kpc 
Legal Events
Date  Code  Title  Description 

AK  Designated states 
Kind code of ref document: A1 Designated state(s): CN JP KR US 

AL  Designated countries for regional patents 
Kind code of ref document: A1 Designated state(s): AT BE CH DE DK ES FI FR GB GR IE IT LU MC NL PT SE 

DFPE  Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101)  
121  Ep: the epo has been informed by wipo that ep was designated in this application  
122  Ep: pct application nonentry in european phase 