Simulation parameter determination method

Abstract

Provided is a method for reduction of calculation amount required for parameter determination, and for reduction of calculation amount required for parameter determination satisfying desired accuracy. It is a method for preferentially selecting analysis objects, which are capable of providing parameter determination in high degree of certainty, and for controlling calculation order of the selected analysis objects using a calculation order list.

Description

INCORPORATION BY REFERENCE

The present application claims priority from Japanese application JP2006-149108 filed on May 30, 2006, the content of which is hereby incorporated by reference into this application.

BACKGROUND OF THE INVENTION

This invention relates to a simulation parameter determination method, in particular, relates to the parameter determination method necessary for prediction of experimental value by simulation calculation.

With recent progress of a computer, computer simulation has increasingly been utilized in many opportunities, in science field. Simulation of molecular level chemical reactions occurring in the body is one example thereof; in particular, simulation of binding reactions between proteins and low molecular weight compounds is expected in application to drug discovery. Because water occupies about 60% of composition, in binding reactions of molecules in the body, not only molecules directly participating in binding but also a water solvent present around the molecules largely contribute to binding energy. Therefore, highly precise calculation of binding energy in a water solvent is required for computer simulation. In the case where computer simulation is utilized in drug discovery, such calculation accuracy is desired that energy error of calculated value relative to experimental value is within 1.4 kcal/mol. This value corresponds to 10 times error in binding constant obtained as experimental measurement result of binding between proteins and low molecular weight compounds.

In highly precise calculation of solvation energy, defined as energy difference before and after introduction of effect of water, use of a calculation model with high theoretical exactitude is necessary. Explanation will be given on a polarizable continuum model based on the first principle quantum chemistry calculation, which is one of such a model, and parameters introduced therein, with reference to FIG. 1. This model explicitly deals with atomic nucleus and electrons of the calculation object molecule 101, however, the water solvent 102 is dealt with as dielectrics including the calculation object molecule 101. In addition, as shown by an outline zone in the drawing, definition of a boundary surface 104 between the space 103 occupied by a molecule and other region, namely the space occupied by dielectrics, is required. The boundary surface can be defined as the surface formed by overlap of the sphere 105 arranged on all atoms constituting the molecule. Approximately, no arrangement of a sphere on a hydrogen atom may be selected. Solvation energy is calculated based on charge present on the boundary surface, as a result of interaction between a molecule and a water solvent. Therefore, definition of the boundary surface has direct influence on accuracy in calculation. Accordingly, considering the radius 106 of a sphere arranged on an atom, as a parameter specifying the boundary surface between a solute and a solvent, determination of this value is capable of providing highly accurate calculation of solvation energy. Hereafter, this radius is referred to as “an atomic parameter”.

As one of the atomic parameter determination method, there is a method for using a gradient matrix. Explanation will be given below on the atomic parameter determination procedure using a gradient matrix, with reference to FIG. 2 and FIG. 3.

FIG. 2 is a flowchart of the atomic parameter determination method using a gradient matrix. First of all, the energy convergence threshold T 2011, the experimental value 2012 of three-dimensional coordinate information and salvation energy of a molecule, and the atomic type 2013, which is a classification based on chemical similarity of each atom constituting the molecule, are input from the input apparatus 201. Subsequently, in the atomic parameter initialization step 202, values of atomic parameters set by each of atomic types are initialized. Then, in the matrix generation step 203, a matrix that stores salvation energies of all molecules and salvation energy gradient of all molecules to change in each atomic parameter, is generated. Next, in the matrix-equation solving step 204, update value vector x is solved, so as to minimize the objective function (expression 1), where A represents a matrix generated; x represents update value vector of atomic parameter; and b represents experimental value vector of solvation energy. g=Σ|Ax−b|²   (Expression 1) Next, in the atomic parameters update step 205, atomic parameter value is updated based on the update value vector. Next, in the atomic parameter calculation termination judging step 206, whether the the amount of change in residual sum of squares of solvation energy is below τ or not, is judged. In the case where the the amount of change is equal to or larger than τ, the matrix generation step 203, the matrix-equation solving step 204 and the atomic parameters update step 205 are repeated. In the case where the the amount of change of the atomic parameter is below τ, the converged atomic parameter value 2071 is output by the output apparatus 207.

FIG. 3 is details of processes in the matrix-generating step, the matrix-equation solving step and the atomic parameters update step, in the atomic parameter determination method using a gradient matrix. The process 302 to be executed in the matrix generation step 301 is calculation of solvation energies 3021 of all molecules, and solvation energy gradient 3022 of all analysis objects to change in each atomic parameter. The calculation results are stored in the matrix 3023. The amount of calculation required for each one of the matrix element 3024 shown by open circle marks in the matrix corresponds to one-time-evaluation of salvation energy using the polarizable continuum model. The process 304 to be executed in the matrix-equation solving step 303 is the minimization calculation of objective function g, which can be defined by a matrix-equation composed of the matrix 3041, in which the number of line generated in the matrix generation step 301 is the number of molecule, and the number of column is (the number of atomic type +1); the solution vector 3042 having the number of element of (the number of atomic type +1); and salvation energy experimental value vector 3043 having the same element number as the number of molecule. The matrix 3041 and the salvation energy experimental value vector 3043 are known, and the solution vector 3042 is unknown amount as an object to be solved. The process 306 to be executed in the atomic parameters update step 305 is to update the new atomic parameter vector 3061 by summing the atomic parameter vector 3062 before update and the vector 3063 deleted only top element from the solution vector obtained in the matrix equation solving step 303. Each of the elements of the vector 3063 represents the amount of change in atomic parameter value in the next loop. Because updating of the atomic parameter results in change in salvation energy or solvation energy gradient to change in atomic parameter, calculations are repeated from the matrix generation step to the atomic parameters update step until g<τ is satisfied.

Such the atomic parameter determination method has two problems.

The first problem is vast amounts of calculation required for the calculation 3021 of solvation energies of all molecules and the calculation 3022 of salvation energy gradient of all molecules relative to change in each atomic parameter, in the matrix generation step 301. It is because, in the first principle quantum chemistry calculation, molecular integration is calculated, which increases the amount of calculation in the order of m power of atomic number (m=3, 4, 5, . . . , depending on calculation methods), to express inter-electron interactions. When the amount of calculation required for calculation of salvation energy in one time is 1, the amount of calculation required for calculation of salvation energy gradient in one time in a finite difference method is also 1; therefore, total amount of calculation required for matrix generation equals to the number of the matrix element 3024 shown by open circle marks in the matrix 3023. Because the matrix generation step is repeated until convergence of solvation energy, total amount of calculation required for parameter determination is (the number of matrix element x the number of iteration).

In more generally, N represents the number of molecules used in atomic parameter determination; A(n) represents the number of atoms included in the molecule n; f(A(n)) represents the amount of calculation required for salvation energy calculation of the molecule n. As described above, f(A(n)) is a power function of A(n). In addition, p(n) represents the number of atomic types included in the molecule n; and T represents the amount of calculation.

In the prior art using a gradient matrix, calculation is repeated until the amount of change in atomic parameter is below energy convergence threshold. When the number of iteration is expressed by I, then the amount of calculation is expressed by (expression 2). $\begin{array}{cc}\begin{array}{c}{T}_{\mathrm{conventional}}=I\sum _{n=1}^N\left(1+p\left(n\right)\right)f\left(A\left(n\right)\right)\\ =\sum _{n=1}^N\left(I+I·p\left(n\right)\right)f\left(A\left(n\right)\right)\end{array}& \left(\mathrm{Expression}2\right)\end{array}$

The subsequent term after summation symbol at the right-hand side of the first line of expression 2 is the amount of calculation required for matrix generation in one time, and “1” in the summation symbol corresponds to solvation energy calculation, and p(n) corresponds to gradient calculation of solvation energy relative to change in atomic parameter value.

Supplementary explanation will be given on the first problem using the specific examples. FIG. 4 is an application example of the atomic parameter determination method using a gradient matrix to 10 molecules. In this drawing, the atomic type 401, the resultant atomic parameter value 402, and the atomic type 403, which each molecule contains, are shown. As the atomic type 401, the following 6 kinds were defined: CH₃(C), CH₃(N), CH₂(CC), CH₂(CN), NH₂(C), and NH₂(N). Atom(s) shown in the parenthesis shows atom(s) of bonding partner. For example, the atomic type CH₂(CN) represents carbon atom of CH₂ having bond with one carbon atom and one nitrogen atom. The atomic type 403 corresponding to each molecule was shown by open circle marks. For example, in CH₃—CH₂—NH₂ of the sixth molecule, CH₃ belongs to the CH₃(C) atomic type, because it bonds to CH₂; CH₂ belongs to the CH₂ (CN) atomic type, because it bonds to CH₃ and NH₂; and NH₂ belongs to the NH₂(C) atomic type, because it bonds to CH₂. As a result, open circle marks are provided at three sites in this molecule. As an exception in the present atomic parameter determination method, for the non-methyl carbon atom of (CH₃)₄—C of the ninth molecule, a fixed value of 1.7 was used; because this atom is surrounded by 4 methyl groups and thus it is difficult to contact with water molecules and is thought that contribution to salvation energy is small.

As for initial values of atomic parameters, 1.7 was used for 4 kinds of CH₃(C), CH₃(N), CH₂(CC) and CH₂ (CN), and 1.4 for 2 kinds of NH₂(C) and NH₂(N). As a specific method for solving, a solution vector was approximately determined every iteration time, by a steepest gradient method to be described below, to update atomic parameter. In addition, to accelerate atomic parameter convergence, atomic parameter was updated using the solution of past iterations, once four iterations. Details will be given below.

• 1) An update method every iteration
• 1-1) As for each atom “a” not equivalent in view of symmetry, which composes each molecule n, Taylor expansion of solvation energy En is terminated at the first order term, and based on the resultant equation and energy residue Z_n, atomic parameter, r_n,a,t, and weight thereof, W_n,a,t, are calculated; (Expression 3) and (Expression 4). $\begin{array}{cc}{r}_{n,a,t}=r_{n,a,t\left(\mathrm{old}\right)}-\frac{z_n}{\frac{\partial E_n}{\partial r_{n,a,}}·w_{n,a,t}}& \left(\mathrm{Expression}3\right)\\ w_{n,a,t}=\frac{{\left(\frac{\partial E_n}{\partial r_{n,a,t}}\right)}^2}{\sum _t{\left(\frac{\partial E_n}{\partial r_{n,a,t}}\right)}^2}& \left(\mathrm{Expression}4\right)\end{array}$
• 1-2) Updated value r_t of the atomic parameter is calculated as weighted average of r_n,a,t, obtained for each molecule n. $\begin{array}{cc}{r}_t=\frac{\sum _{n,a}{r}_{n,a,t}·w_{n,a,t}}{\sum _{n,a}{w}_{n,a,t}}& \left(\mathrm{Expression}5\right)\end{array}$
• 2) An update method every four iterations
• 2-1) As for each atom “a” composing each molecule n and not equivalent in view of symmetry, a straight line z_n=A r_n,a,t+B, which approximates hysteresis of the atomic parameters r_n,a,t(I), and residue z_n,(I) at I=4i−3, 4i−2, 4i−1, and 4i, is determined to calculate the atomic parameter value in which the residue becomes 0, as r_n,a,t=−B/A, assuming that I is the number of iteration; and I is an integer.
• 2-2) As for each atomic type t, updated value r_t of the atomic parameter is calculated as average of the resultant r_n,a,t. Sum on delta function of the denominator is the number of atoms belonging to atomic type t, among atoms composing molecule n and not equivalent in view of symmetry. (Expression 6) $\begin{array}{cc}{r}_t=\frac{\sum _{n,a}{r}_{n,a,t}}{\sum _{n,a}{\delta }_{t_{n,a},t}}& \left(\mathrm{Expression}6\right)\end{array}$

FIG. 5 shows the amount of calculation and calculation error obtained by the application of the atomic parameter determination method using the above-explained gradient matrix. In this drawing, the abscissa axis is the amount of calculation, which was defined as the calculation time 501 required for atomic parameter determination using one computer, and the longitudinal axis is the calculation error 502, which was defined as average value of absolute calculation errors of each of the molecules. The plot 503 of lozenge marks shows relation between the amount of calculation and calculation error. Because this method repeats solving of a matrix-equation, the calculation error becomes smaller with increase in the number of iteration. To attain a calculation error of 0.1 kcal/mol, the amount of calculation of about 500 minutes is required. As shown in this example, the amount of calculation required is vast, even in atomic parameter determination using relatively small number of molecules.

The second problem is that calculation accuracy is unpredictable in advance, when the resultant atomic parameter is applied to predict salvation energy. In general, calculation accuracy is considered to depend on definition of an atomic type. Therefore, in the case where the calculation accuracy is below desired accuracy, re-definition of the atomic type, and re-execution of whole procedures of atomic parameter determination are required. In this case, iteration of atomic parameter determination procedure itself is required, which further expands the amount of calculation.

This atomic parameter determination method can be generalized with the following substitutions: molecules with “analysis objects”; atoms composing molecules with “elements composing analysis objects”; atomic parameter with “elemental parameter”; experimental values of solvation energy simply with “experimental values”; and calculated values of solvation energy simply with “calculated values”.

FIG. 6 is a flowchart of the parameter determination method using a gradient matrix. First of all, the convergence threshold τ 6011, the necessary information and experimental value 6012 for calculation of analysis object, and the elemental type 6013 for each element composing an analysis object are input by the input apparatus 601. Subsequently, in the elemental parameter initialization step 602, elemental parameter set by each of elemental types is initialized. Then, in the matrix generation step 603, a matrix, which stores the calculated value of all analysis objects, and the gradient of calculated value of all analysis objects relative to change in each elemental parameter is generated. Then in the matrix-equation solving step 604, such an updated value vector x, that minimizes the objective function g=Σ|Ax−b|², is solved, assuming that A represents a matrix generated; x represents updated value vector of elemental parameter; and b represents experimental value vector. Next, in the elemental parameters update step 605, value of elemental parameter is updated based on update value vector. Next, in the elemental parameter calculation termination judging step 606, whether the amount of change in residual sum of squares is below τ or not, is judged. In the case where the amount of change is equal to or larger than τ, the matrix generation step 603, the matrix-equation solving step 604 and the elemental parameters update step 605 are repeated. In the case where the amount of change of the elemental parameter is below τ, the converged elemental parameter 6071 is output by the output apparatus 607.

FIG. 7 is the details of processes in the matrix generation step, matrix-equation solving step and elemental parameters update step in the parameter determination method using a gradient matrix. The process 702 to be executed in the matrix generation step 701 is calculation of the calculated value 7021 of all analysis objects, and the gradients 7022 of calculated value of all analysis objects relative to change in each elemental parameter. The calculation results are stored in the matrix 7023. The amount of calculation required for each of the matrix elements 7024, shown by open circle marks in the matrix, corresponds to one-time evaluation of calculated value. The process 704 to be executed in the matrix-equation solving step 703 is the calculation to minimize the objective function g, which can be defined by a matrix-equation composed of the matrix 7041 composed of analysis object number as the number of line, and (elemental type number +1) as the number of column, generated in the matrix generation step 701, and the solution vector 7042 having the number of element of (elemental type number +1) and the experimental value vector 7043 having the same elemental number as analysis object number. The matrix 7041 and the experimental value vector 7043 are known, and the solution vector 7042 is unknown amount as an object to be solved. A plurality of solving methods are present for the matrix-equation, however, it is premised on that rate determining step of calculation is the matrix generation step and not the matrix-equation solving step, therefore, a specific method for solving may be arbitrary here. The process 706 to be executed in the elemental parameters update step 705 is to update the new elemental parameter vector 7061 by sum of the elemental parameter vector 7062 before update and the vector 7063 deleted only top element from the solution vector obtained in the matrix-equation solving step 703. Each element of the vectors 7063 represents the amount of change in the next loop of value of elemental parameter. When the elemental parameter is updated, calculation value or gradient of calculation value relative to elemental parameter will change, therefore, calculations are repeated again from the matrix generation step.

In the generalized parameter determination method, in the case where matrix element calculation occupies a large portion in total calculation, it raises a problem that the amount of calculation required for repeated solving of the matrix-equation becomes vast. The amount of calculation is given by expression 2. However, function form of f(A(n)) differs depending on analysis objects. In addition, in the case where parameter determination result does not satisfy desired accuracy, re-definition of the parameter and iteration of the parameter determination procedure itself are required, which further expands the amount of calculation required.

To solve the above first problem, it is an object of this invention to reduce the amount of calculation required for parameter determination.

In addition, it is a second object of this invention to reduce the amount of calculation required for parameter determination satisfying desired accuracy, to solve the above second problem.

SUMMARY OF THE INVENTION

To solve the above first problem, in the parameter determination method relevant to the embodiment of this invention, only molecules having the number of undetermined atomic parameter of equal to or less than 1 are selected, and thus selected molecules are stored in a calculation order control list; and based on this list, the undetermined atomic parameters are sequentially determined. Selection of molecules is repeated until all atomic parameters are determined.

At the first round of atomic parameter determination procedure by iteration, a molecule having the number of undetermined atomic parameter of 1 is selected. In order to reproduce, by calculation, experimental value of salvation energy of such a molecule, determination of only one atomic parameter value is needed. Therefore, atomic parameter can unambiguously be determined.

Then, a molecule having the number of undetermined atomic parameters of 2 is considered. When one of the two undetermined atomic parameters can be tentatively determined based on atomic parameter value determined in the first round of iteration, then the number of undetermined atomic parameter becomes 1, and determination of only this one atomic parameter is left. In the case where reliability of the tentatively determined atomic parameter is high, newly determined atomic parameter is also determined with high degree of certainty.

In order to determine tentatively atomic parameter in high reliability, atomic charge and averaged bond distance are calculated in advance on each of the atoms. Then, the atomic parameter is expressed as a function with atomic charge and averaged bond distance as variables. Atomic charge is quantity relating to Coulomb interaction strength between molecule and water solvent, and averaged bond distance is quantity reflecting bond environment such as the number of bonded atoms of a molecule, bond order, atoms of bonding partner and the like. Therefore, atoms having similar atomic charge and averaged bond distance are considered to be similar also in values of atomic parameters.

However, atomic parameter obtained by the above method has no guarantee to be capable of giving determination of salvation energy in desired accuracy, for example, 1.4 kcal/mol. Therefore, to solve the above second problem, in the parameter determination method relevant to the embodiment this invention, first of all, acceptable energy error is set, which is defined as maximal acceptable value of difference between calculated value of solvation energy obtained by simulation calculation, and experimental value of solvation energy. Then, whether absolute value of calculation error, defined by difference between calculated value and experimental value, is within acceptable error or not, is judged on each of the molecules, in sequential determination process of undetermined atomic parameter based on the calculation order control list.

In the case where absolute value of calculation error is larger than acceptable error, one atomic parameter, which gives largest contribution to salvation energy, is noticed among a plurality of atomic parameters contained in a molecule, so as to make calculated value closer to experimental value. This atomic parameter value, although tentatively determined, is re-set to be undermined, to determine newly atomic parameter value. In this method, acceptable error is fixed instead of the number of atomic parameter.

It was explained above that atomic parameter value was expressed as a function with atomic charge and averaged bond distance as variables. As shown in the above example, in the case where tentatively determined atomic parameter based on a simple function is re-set newly as being undermined to execute re-determination process, the function that expresses atomic parameter value becomes more complicated.

An example will be given on the case where 2 atomic parameters are already determined on a certain atomic type, and values of the atomic parameters are expressed with a linear function of atomic charge as a variable. Atomic parameter value of other atom belonging to the same atomic type can be tentatively determined by substitution of atomic charge of the atom into the linear function expressing atomic parameter value. In the case where the tentatively determined atomic parameter value was re-set as being undermined in a process of atomic parameter determination, and the value was re-determined, value of the atomic parameter of three atoms belonging to the noticed atomic type cannot be expressed by a linear function with atomic charge as a variable. However, the atomic parameter value can be expressed by using a linear function with both atomic charge and averaged bond distance as variables, or more complicated functions like two kinds of linear functions with atomic charge as a variable. Therefore, in this invention, a function expressing value of each of atomic parameters is made complicated in atomic parameter determination process, so that absolute value of calculation error is within acceptable error.

Explanation was given above on the atomic parameter determination method used for calculation of salvation energy of a molecule, as an example. In generalization of this example to analysis objects such as materials, parts, structures and the like instead of molecules, a means for controlling of calculation order of analysis objects using calculation order list is considered to be effective. Applicable scope of this invention is the case where a calculation object can be divided to element and attribute value can be defined to each element, and also the case where evaluation of calculated value or gradient relative to parameter change of calculated value is a rate-determining step. In addition, such a means is also considered effective that value of elemental parameter is expressed by a function of value of elemental property, and, if necessary, a function form is made complicated, so that difference between experimental value such as characteristic value or property measured on a calculation object, and calculated value obtained by simulation is within acceptable error. The applicable scope of this means is the case where increase in the number of parameter is allowed, which is used to fit experimental value and calculated value.

A means for selection of only a molecule having the number of undetermined atomic parameter equal to or less than 1, has effect to determine preferentially atomic parameter which can be determined with high degree of certainty. For example, a benzene molecule is composed of 6 carbon atoms, and they are equivalent in view of symmetry, therefore the number of atomic parameter can be said to be 1. Therefore, atomic parameter value, which reproduces experimental value of salvation energy of benzene, is determination with high degree of certainty, and in this case, unambiguously.

In addition, a means for storing the calculation order control list, and for iteration of one-variable equation solving based on the list to sequentially determine undermined atomic parameter does not require the matrix generation step as in a conventional technique. Therefore, this method has effect of reducing the amount of calculation required for atomic parameter determination. In a conventional technique, calculation was iterated on solvation energy of all molecules and salvation energy gradient of all molecules relative to change in each of the atomic parameters. On the other hand, embodiments of this invention determine calculation order of molecules, and calculation is iterated on salvation energy of each of the molecules and salvation energy gradient relative to change in one atomic parameter.

In addition, a means for re-setting the tentatively determined atomic parameter value on a certain molecule, as being undermined, to determine newly atomic parameter value, in a process of sequential determination of undetermined atomic parameter by iteration of solving of a one-variable equation, based on the calculation order control list, has effect of reproducing solvation energy in desired accuracy. It is because new re-determination of the tentatively determined atomic parameter value is capable of improving calculation accuracy of salvation energy. Under the condition that increase in the number of parameter used to reproduce experimental value of salvation energy is acceptable, re-definition of atomic type and re-execution of whole procedure of atomic parameter determination can be avoided using this means.

The above effect is similarly obtained also in the case of considering analysis objects such as materials, parts, structures and the like instead of molecules. A means for selecting only one molecule having the number of undetermined atomic parameter equal to or less than 1, has effect to determine preferentially elemental parameter that can be determined in high degree of certainty. In addition, a means for storing the selected analysis object into the calculation order control list, and for iteration of one-variable equation solving based on the list to sequentially determine undermined elemental parameter does not require the matrix generation step as in a conventional technique. Furthermore, a means for re-setting the tentatively determined elemental parameter value on a certain analysis object, as being undetermined, to determine newly elemental parameter value, in a process of sequential determination of undetermined elemental parameter by iteration of solving of a one-variable equation, based on the calculation order control list, has effect to reproduce calculated value in desired accuracy.

According to this invention, the amount of calculation can be reduced as compared with the prior art.

Other objects, features and advantages of the invention will become apparent from the following description of the embodiments of the invention taken in conjunction with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 shows a polarizable continuum model and parameters used in the model.

FIG. 2 is a flowchart of the atomic parameters determination method using a gradient matrix.

FIG. 3 shows details of processes in the matrix generating step, matrix-equation solving step, and atomic parameters update step in the atomic parameter determination method using a gradient matrix.

FIG. 4 shows an atomic type definition, determined parameters values, and atomic types corresponding to each molecule in the atomic parameter determination method using a gradient matrix.

FIG. 5 shows an amount of calculation and calculation error in the atomic parameter determination method using a gradient matrix.

FIG. 6 is a flowchart of the parameters determination method using a gradient matrix.

FIG. 7 shows details of processes in the matrix generation step, matrix-equation solving step, and elemental parameters update step in the parameter determination method using a gradient matrix.

FIG. 8 is a flowchart of the atomic parameters determination method.

FIG. 9 shows details of processes in calculation order list generation step, the one-variable equation solving step, and the atomic parameters update step in the atomic parameter determination method.

FIG. 10 shows an input of molecular information.

FIG. 11 shows an input of atomic information.

FIG. 12 shows an initialization of atomic parameters.

FIG. 13 shows a hardware control approach in the one-variable equation solving step.

FIG. 14 shows an example of atomic parameters update in the case that value of atomic parameter is expressed by a linear function of a value of an atomic attribute.

FIG. 15 shows an example of atomic parameters update in the case that value of atomic parameter is expressed by a linear function of values of two atomic attributes.

FIG. 16A shows an atomic type definition used in the application of this invention, process of atomic parameter values determination, and atomic type corresponding to each molecule.

FIG. 16B shows an atomic type definition used in the application of this invention, process of atomic parameter values determination, and atomic type corresponding to each molecule.

FIG. 16C shows an atomic type definition used in the application of this invention, process of atomic parameter values determination, and atomic type corresponding to each molecule.

FIG. 16D shows an atomic type definition used in the application of this invention, process of atomic parameter values determination, and atomic type corresponding to each molecule.

FIG. 17 shows a comparison of amount of calculation and calculation error between this invention and the prior art.

FIG. 18 shows a comparison of solvation energy prediction accuracy.

FIG. 19 is a flowchart of the parameter determination method.

FIG. 20 shows details of processes in calculation order list generation step, one-variable equation solving step, and elemental parameters update step in the parameter determination method.

FIG. 21 is a flowchart of the atomic parameter determination method under the fixed acceptable error condition.

FIG. 22 shows details of processes in calculation order list generation step, one-variable equation solving step, and atomic parameters update step, in the atomic parameter determination method under the fixed acceptable error condition.

FIG. 23 shows details of calculation order list update step.

FIG. 24A shows an example of atomic parameter update in the case that value of atomic parameter is expressed by a linear function of value of an atomic attribute.

FIG. 24B shows an example of atomic parameter update in the case that value of atomic parameter is expressed by a linear function of value of an atomic attribute.

FIG. 25A shows an example of atomic parameters update in the case that value of atomic parameter is expressed by a linear function of values of two atomic attributes.

FIG. 25B shows an example of atomic parameters update in the case that value of atomic parameter is expressed by a linear function of values of two atomic attributes.

FIG. 26 shows a comparison of the amount of calculation when the number of molecules N used in the atomic parameter determination is sufficiently large.

FIG. 27 shows a comparison of the amount of calculation between the prior art and this invention in the atomic parameter determination method using 63 molecules.

FIG. 28 is a flowchart of simulation parameter determination method under the fixed acceptable error condition.

FIG. 29 shows details of processes in calculation order list generation step, the one-variable equation solving step, and the elemental parameters update step, in the parameter determination method under the fixed acceptable error condition.

FIG. 30 is a flowchart of atomic parameter determination method including information display step.

FIG. 31 shows an example of screen display in calculation order list display step.

FIG. 32 shows an example of screen display in the calculation progress display step.

FIG. 33 shows an example of data view at the atomic parameters display step.

DESCRIPTION OF THE INVENTION

Explanation will be given below on embodiments of this invention with reference to drawings.

Embodiment 1

Explanation will be given on the atomic parameter determination method of this invention, with reference to FIG. 8 and FIG. 9.

FIG. 8 is a flowchart of the atomic parameter determination method. First of all, the energy convergence threshold τ 8011, the three-dimensional coordinate information determined by calculation or experiment, and experimental value 8012 of solvation energy for a molecule, atomic type of each atom composing a molecule, and the atomic type, atomic charge, and averaged bond distance 8013 for each atom are input by the input apparatus 801. Subsequently, in the initialization step 802, the atomic parameter initialization step 8021 set by each of atomic types and the sorting up step 8022 of all molecules by the number of undetermined atomic parameters are executed. Then, in calculation order list generation step 803, the selection step 8031 of a molecule, having the number of undetermined atomic parameters is 1, from a molecule containing undetermined atomic parameter and the calculation order list generation step 8032, which stores sorting up result of selected molecules by number of atomic parameters, are executed. Then in the one-variable equation solving step 804, solving of a one-valuable equation is iterated on a molecule on the top of calculation order list, until the amount of change of calculated value of salvation energy becomes below τ, and the undetermined atomic parameter determination step 8041 and calculation order list update step 8042 are executed. In calculation order list termination judging step 8043, in the case where a molecule is still present on the top of calculation order list, procedure is returned to the step 8041; in the case where such a molecule is not present any more, procedure is forwarded to the atomic parameters update step 805.

Next, in the atomic parameters update step 805, on an atomic type corresponding to determined atomic parameter, the atomic parameter expression formula update step 8051, which equation being a function of atomic charge and averaged bond distance, the sorting up step 8052 of undetermined molecules by the number of undetermined atomic parameters, the sorting down step 8053 of molecules with the identical number of undetermined atomic parameters by the number of atomic parameters, and the undetermined molecule update step 8054 are executed. In the atomic parameter calculation termination judging step 806, whether an undetermined molecule is present or not, is judged. In the case where the undetermined molecule is present, calculation order list generation step 803, the one-variable equation solving step 804 and the atomic parameters update step 805 are iterated. In the case where the undetermined molecule is not present, calculation error of each molecule is calculated using the resultant atomic parameter, in the calculation error computing step 807, and the determined atomic parameter value 8081 is output by the output apparatus 808.

FIG. 9 is details of processes in calculation order list generation step, the one-variable equation solving step, and the atomic parameters update step in the atomic parameter determination method. As an example, explanation will be given on the first and the second rounds of iterations of each of the steps, in the case of the number of molecules of 9, and the number of atomic types of 8. In the first round process 902 to be executed in calculation order list generation step 901, the molecule having the number of undetermined atomic parameter of 1, the molecule 9022 (molecule 8, molecule 3 and molecule 4) having the number of undetermined atomic parameter of 1, are selected from molecules sorted up by the number 9021 of undetermined atomic parameters in advance. Calculation order of these molecules is molecule 8, molecule 3 and molecule 4, namely the same as the arrangement order of the molecules, and this order is stored in calculation order list 9023. In the first round process 904 to be executed in the one-variable equation solving step 903, undetermined atomic parameter is determined based on calculation order list. Because the first calculation is on molecule 8, salvation energy is calculated first, using atomic parameter r_a,8, and value of r_a,8 is adjusted so as to reproduce experimental value E8. In this process, value that can reproduce experimental value is searched while changing r_a,8, therefore, iterated solving of a one-variable equation is required. Similar process is executed on the two selected molecules left. In the first round process 906 to be executed in the atomic parameters update step 905, expression formula of atomic parameter as a function of atomic charge and averaged bond distance, is updated on a, b and h that are the atomic type 9061 corresponding to determined atomic parameter. Expression formula of atomic parameter may be a function of either one of atomic charge or averaged bond distance. For example, the atomic parameter 9062 of an atomic type a can be expressed as r_a(q)=c₀+c₁q; where c₀ and c₁ are coefficients. Then, undetermined atomic parameter value classified to atomic type “a” is obtained by substitution of atomic charge q into r_a(q)=c₀+c₁q. However, on the atom 9063, which atomic charge is apart that of other atom belonging to the same atomic type by certain threshold or more from, there may be an option of not setting atomic parameter value. The “value assigned” atom 910 is shown by a filled circle, and the value not assigned, namely the “value not assigned” (undetermined) atom 911 is shown by an open circle. Then, at this time, 6 molecules, namely molecules 5, 6, 7, 2, 1 and 9, are sorted up by the number 9064 of undetermined atomic parameters. In addition, in the case where number of undetermined parameters is the same, the molecules are arranged in sort down of the number 9065 of atomic parameters. Here, molecules having the number of undetermined parameters of equal to or larger than 1 become new undetermined molecules. Molecule 2 has the number of undetermined parameters of 0 at this moment, thus is not used in atomic parameter determination. In the second round process 907 to be executed in calculation order list generation step 901, the molecule 9072 having the number of undetermined atomic parameters 9071 of 1 (molecule 5, and molecule 6) are selected from molecules arranged in advance. Calculation order of these molecules is molecule 5, and molecule 6, and this order is stored in calculation order list 9073. In the second round process 908 to be executed in the one-variable equation solving step 903, undetermined atomic parameter is determined based on calculation order list. Solvation energy is calculated first, using atomic parameter r_a,5 under fixed value condition, and undetermined atomic parameter r_d,5, and value of r_d,5 are determined so as to reproduce experimental value E₅. In this process, value that can reproduce experimental value is searched while changing only r_d,5, therefore, process to be executed is iterated solving of a one-variable equation, even for a molecule composed of 2 atomic parameters. Similar process is executed on the 2 selected molecules left. In the second round process 909 to be executed in the atomic parameters update step 905, expression formula of atomic parameter, as a function of atomic charge and averaged bond distance, is updated on a, b, h, d and e that are the atomic type 9091 corresponding to determined atomic parameter. For example, the atomic parameter 9092 of atomic type “a” is updated from r_a(q)=c₀+c₁q to r_a(q)=c′₀+c′₁q. After updating expression formula and determination of atomic parameter value of an atom classified to the corresponding atomic type, re-arrangement operation of undetermined molecule is iterated in the same manner as the first round of iteration. This process is iterated until undetermined molecules are absent.

Explanation will specifically be given on embodiment of process in relation to a part of the step of a flowchart of the atomic parameter determination method shown by FIG. 8.

An example of the input 8012 of three-dimensional coordinate information of a molecule and experimental value of salvation energy is shown in FIG. 10. In addition, an example of the input 8013 of an atomic type t, atomic charge q, and averaged bond distance 1 for each atom composing a molecule, is shown in FIG. 11. Because these inputs are mutually related, explanation will be given on both of FIG. 10 and FIG. 11 at the same time. First of all, input information can be classified into molecular information and atomic information. Input information attributed to molecule includes the three-dimensional coordinate information 1001 of a molecule. As an example, the structure 1002 of para-xylene molecule was shown. Actual input is three-dimensional coordinates of each of the atoms composing the molecule. In addition, input information attributed to molecule includes the experimental value 1003 of solvation energy. The experimental value 1004 of solvation energy of a para-xylene molecule is −0.80 kcal/mol. The number 1005 of atomic parameter to be determined are 3 kinds, open circle, filled circle and hatched circle, using the example 1006 of a para-xylene molecule. Because the atoms with the same kind of mark are equivalent in view of symmetry, they have the same atomic parameter value. This information is not essential to parameter determination calculation, however, preparation thereof in advance provides convenience. Atomic information includes an atomic type 1101, an atomic charge 1102 and an averaged bond distance 1103. A molecule of para-xylene has 3 kinds of atomic parameters, and the atomic type, the atomic charge and the averaged bond distance are input on each thereof. Specifically, the atomic charge was determined by charge assignment algorism on each of the atomic nuclei composing the molecule, so as to provide best approximation of static potential of the molecule. The averaged bond distance was calculated as average value of bond distance for the atom to which each of the atoms directly bonds.

FIG. 12 shows an example of initialization 8021 of the atomic parameter r_t(q, 1), set by each of the atomic types. It is also convenient here to classify information to be handled into molecular information and atomic information. The atomic information 1201 includes the value 1202 of each atomic parameter and the status 1203 of atomic parameter. The status of atomic parameter is either the “value assigned” 1204 or the “value not assigned” 1205. As initialization value, values are not defined for all atomic parameters, and thus the status is treated as “value not assigned”. The molecular information 1206 can be calculated by summation of sets of atomic information. The number 1207 of “value assigned” atomic parameters contained in a molecule is the number of atomic parameter where an atomic parameter status is “value assigned”, and the number 1208 of “value not assigned” atomic parameters also can similarly be calculated. The calculation status 1209 of a molecule is either “unfinished” or “finished”; in the case where all of the atomic parameters are “value assigned”, the calculation status 1209 of a molecule thereof becomes “finished”. The initial value 1210 of molecular information is only summation result of initial value sets of atomic information. In the case of para-xylene, the initial value 1211 of number of “value assigned” atomic parameters contained in a molecule is 0, and the number 1212 of “value not assigned” atomic parameters is 3, the same as number of atomic parameters. The initial value 1213 of a calculation status is “unfinished”.

Embodiment of this invention is characterized by having the step that controls calculation order of solvation energy of a molecule using calculation order list. Explanation will be given on a hardware control approach in the one-variable equation solving step 804, using FIG. 13.

The following process is executed on a molecule listed on the top of calculation order list 1301. First of all, in the input file generation step 1302, the input file 1303 for salvation energy calculation is prepared based on three-dimensional coordinate information of the molecule that was input in advance, and information of “value assigned” atomic parameter. For “value not assigned” atomic parameter, setting of suitable value is necessary. For example, van der Waals' radius value specified by each element may be used. In the solvation energy calculation job input step 1304, the input file 1303 is delivered to the salvation energy calculation program 1306 that is installed in the computer 1305 to calculate solvation energy. In the calculation result outputting step 1307, the calculation result is written in the output file 1308 by the salvation energy calculation program 1306. In the output result processing step 1309, based on comparison between experimental value of salvation energy of a molecule already input, and calculated value, atomic parameter is updated. Procedures from the input file generation step 1302 to the output result processing step 1309 are iterated until the amount of change of calculation value of solvation energy becomes below energy convergence threshold. At the termination of the iteration, the molecule on the top of calculation order list is deleted and the next top and lower molecules in the calculation order are sequentially moved up. The above procedure is continued until calculation order list is empty.

Explanation will be given on embodiment of the atomic parameter, r_t(q, 1), update step 8051 of atomic type corresponding to determined atomic parameter, using FIG. 14 and FIG. 15.

FIG. 14 is an example of atomic parameter update in the case that the atomic parameter value 1401 is expressed by a linear function of the value 1402 of an atomic attribute such as atomic charge as a variable. A filled circle in the drawing shows the atom 1403 having “value assigned” atomic parameter, while an open circle shows the atom 1404 having “value not assigned” atomic parameter. First of all, the linear function 1405 correlating atomic parameter value with value of an atomic attribute is defined using the “value assigned” atom 1403. Then, the value 1406 of an atomic attribute of a “value not assigned” atom is substituted in a linear function to assign atomic parameter value. Here, value assigned to the “value not assigned” atom 1407 having value of an atomic attribute far apart from value of an atomic attribute of a “value assigned” atom, may not necessarily be highly precise. Atomic parameter of such a molecule is determined as the last atomic parameter of a molecule to which the atom belongs.

FIG. 15 is an example of atomic parameter update in the case that the atomic parameter value 1501 is expressed by a linear function of two values of atomic attribute such as the atomic charge 1502 and the averaged bond distance 1503 as variables. A filled circle (or a gray circle) in the drawing shows the atom 1504 having “value assigned” atomic parameter, while an open circle shows the atom 1505 having “value not assigned” atomic parameter. First of all, the linear function 1506 correlating atomic parameter value with value of an atomic attribute is defined using the “value assigned” atom 1504. Then, the atomic charge and averaged bond distance 1507 of a “value not assigned” atom is substituted into a linear function to assign atomic parameter value. Here, value assigned to the “value not assigned” atom having value of an atomic attribute far apart from value of an atomic attribute of a “value assigned” atom, 1508 may not necessarily be highly precise. Atomic parameter of such a molecule is determined as the last atomic parameter of a molecule to which the atom belongs.

Finally, the amount of calculations obtained by a conventional determination method for atomic parameter, and the atomic parameter determination method according to this invention are compared. The amount of calculation in the prior art is described in expression 2. In this invention, molecules are sequentially selected, and calculation is iterated on each of the molecules until the amount of change of calculated value of solvation energy becomes below energy convergence threshold. The number of iteration is represented by I(n), which is different by a molecule. However, there are such molecules as not to be used in atomic parameter calculation. Therefore, the amount of calculation can be calculated by “expression 7” using δ function which is defined 1 for molecules used in atomic parameter determination, and 0 for molecules not used. $\begin{array}{cc}{T}_{\mathrm{invention}}=\sum _{n=1}^N\delta \left(n\right)\left(1+I\left(n\right)\right)f\left(A\left(n\right)\right)& \left(\mathrm{Expression}7\right)\end{array}$

Comparison between expression 2 and expression 7 shows smaller amount of calculation in a method of embodiment of this invention as compared with the prior art in view of three points. Firstly, calculation is executed on all molecules in the prior. art, while, in embodiment of this invention, calculation is executed on a part of the molecules by presence of the δ function, by which the amount of calculation is reduced. Secondly, the first term in the parenthesis at the right-hand side of the second line of expression 2 is “I” in the prior art, while in embodiment of this invention, the first term in the parenthesis at the right-hand side of expression 7 is “1”, by which the amount of calculation is reduced. Thirdly, the second term in the parenthesis is “I·p(n)” in the prior art, while in Embodiment of this invention, the first term in the parenthesis is “I(n)”. The number of iteration, I, required for solving a matrix-equation as a multi-variable non-linear problem, is considered to be larger than the number of iteration “I(n)” required for solving a one-variable equation as a one-variable non-linear problem, therefore the amount of calculation in Embodiment of this invention is considered less.

Embodiment 2

Explanation will be given on a determination example of 6 atomic parameters using 10 molecules with reference to FIG. 16A to FIG. 18.

FIGS. 16A to 16D show the atomic type 1601 used in the application of the present embodiment, the determination process 1602 of atomic parameter value, and the atomic type 1603 corresponding to each molecule. As the atomic type 1601, 3 kinds, CH₃, CH₂, and NH₂ were defined. However, atomic parameters of CH₃, and CH₂, are expressed by a linear function with averaged bond distance 1 as a variable, and atomic parameter of NH₂ is expressed by a linear function with atomic charge q as a variable. Each of the linear functions has 2 variables, therefore, the number of variables to be determined becomes 6. Atomic parameter value at the initial status is undetermined. Explanation will be given below on determination process of atomic parameter.

First of all, in FIG. 16A, the molecules 1604 (molecules numbered 1, 9, 4, 10 and 3) having the number of undetermined atomic parameter of 1, were selected. Atomic parameters to reproduce experimental value of salvation energy were determined on each of the molecules selected.

Then, in FIG. 16B, values were assigned for undetermined atomic parameters. As for atomic type CH₃, the atomic parameter r_CH3 1605 was determined as r_CH3=−2.0323*1+4.5674 based on the atomic parameters obtained on molecules numbered 1 and 9. Similarly, the atomic parameter r_CH3 1606 was determined as r_CH2=3.8382*1−2.9689. As for atomic type NH₂, atomic parameter was still kept “undetermined” at this point, because 2 variables necessary in expressing atomic parameter by a linear function were not determined. As for atomic types CH₃ and CH₂, values were assigned to the undetermined atomic parameter 1607 by substitution of averaged bond distance 1 to the linear function determined. As a result, the number 1608 of undetermined atomic parameters became 0 as for the molecule 7 and the molecule 5, and 1 as for the molecule 2, the molecule 6 and the molecule 8.

Then, in FIG. 16C, the molecule 1609 (molecules numbered 2, 6, and 8) having the number of undetermined atomic parameters of 1, were selected. And the atomic parameters to reproduce experimental value of salvation energy were determined on each of the molecules selected.

Finally, in FIG. 16D, based on the 4 atomic parameters 1610 determined on atomic type NH₂ and averaged bond distance, the atomic parameter r_NH2 1611 was determined as r_NH2=0.2943*q+1.8561. As for each of the molecules numbered 3, 2, 6 and 8, because atomic parameters determined on each of the molecules and atomic parameters obtained by substitution of atomic charge to the linear function are different, calculation errors were calculated, using atomic parameter values obtained finally from the latter.

FIG. 17 shows comparison of the amount of calculation and calculation error between in prior art explained in the above column “conventional technique” and in the present embodiment. The abscissa axis in the drawing is the amount of calculation defined as the calculation time 1701 required for determination of atomic parameter using one computer, and the longitudinal axis is the calculation error 1702 defined as average value of absolute calculation error of each of the molecules. The plot of lozenge marks 1703 shows relation between amount of calculation and calculation error in the prior art, and the plot of rectangular marks 1704 shows relation between amount of calculation and calculation error in the present embodiment. In the present embodiment, molecules are sequentially selected and the calculation is completed at a time when all of the atomic parameters are determined. The amount of calculation required was 57 minutes, and a calculation error of about 0.1 kcal/mol was obtained. Therefore, a method of the present embodiment shows about 9 times less amount of calculation required to attain an calculation error of 0.1 kcal/mol, under calculation condition of the present embodiment.

FIG. 18 shows comparison of salvation energy prediction accuracy. On 6 molecules, atomic parameters after iteration of 12 times by the prior art, and atomic parameters obtained by a method of the present embodiment were applied respectively. The molecule 1801 selected has a structure similar to that of 10 molecules used in atomic parameter determination. In the drawing, the experimental value 1802, the calculation error 1803 in the prior art, the calculation error 1804 in the present embodiment and the averaged absolute error 1805 in both methods are shown. Because a method of the present embodiment always determines atomic parameter one by one, it may be considered as approximation to the atomic parameter determination method according to the prior art. However, the averaged absolute error 1805 of the present embodiment is a little smaller as compared with the conventional method. Therefore, the present embodiment that reduces the amount of calculation required for atomic parameter determination can be said to maintain also prediction accuracy of salvation energy. A method for expressing atomic parameter as a function of atomic charge or averaged bond distance is considered rather effective in calculation of solvation energy in high accuracy.

Embodiment 3

Generalization of the atomic parameter determination method relevant to this invention may be possible by substitution of a molecule with “an analysis object”, atoms composing a molecule with “elements composing an analysis object”, an atomic parameter with “an elemental parameter”, experimental value of solvation energy with simply “experimental value” and calculation value of solvation energy with simply “calculation value”. Explanation will be given below on a generalized parameter determination method in comparison, with the prior art with reference to FIG. 19 and FIG. 20.

FIG. 19 is a flowchart of the parameter determination method of the present embodiment. First of all, the convergence threshold τ 19011, the information necessary in calculation of analysis object and experimental value 19012, and the elemental type and element attribute value 19013 of each element composing an analysis object are input from the input apparatus 1901. Subsequently, in the initialization step 1902, the initialization step 19021 of elemental parameter set by each of atomic types and the sorting up step 19022 of all analysis objects by number of undetermined elemental parameters are executed. In calculation order list generation step 1903, the analysis objects selection step 19031 from analysis objects having the number of undetermined atomic parameters is 1, from a molecule containing undetermined atomic parameter and orders of arrangement and calculation of the analysis objects selected are determined to execute the calculation order generation step 19032 are executed. Then in the one-variable equation solving step 1904, solving of one-valuable equation is iterated on an analysis object on the top of calculation order list, until the amount of change of calculated value becomes below 1, and the undetermined elemental parameter determination step 19041 and calculation order list update step 19042 are executed. In calculation order list termination judging step 19043, in the case where an analysis object is still present on the top of calculation order list, procedure is returned to the step 19041; in the case where an analysis object is not present any more, procedure is forwarded to the elemental parameters update step 1905. Next, in the elemental parameters update step 1905, as for an elemental type corresponding to determined elemental parameter, the elemental parameter expression formula update step 19051, which is a function of value of elemental attribute, the sorting up step 19052 of undetermined analysis objects by undetermined elemental parameters, the sorting down step 19053 of analysis objects with the identical number of undetermined elemental parameters, by number of elemental parameters and the undetermined analysis object update step 19054 are executed. Next, in the elemental parameter calculation termination judging step 1906, whether an undetermined analysis object is present or not, is judged. In the case where the undetermined analysis object is present, calculation order list generation step 1903, the one-variable equation solving step 1904 and the elemental parameters update step 1905 are iterated. In the case where the undetermined analysis object is not present, calculation error of each analysis object is calculated using the resultant elemental parameter, in the calculation error computing step 1907, and the determined elemental parameters 19081 is output by the output apparatus 1908.

FIG. 20 is details of processes in calculation order list generation step, the one-variable equation solving step, and the elemental parameters update step, in the parameter determination method of the present embodiment. As an example, explanation will be given on the first and the second rounds of iterations of each of the steps, in the case of the number of analysis objects of 9, and the number of elemental type of 8. In the first round process 2002 to be executed in calculation order list generation step 2001, the analysis object 20022 (analysis object 8, analysis object 3 and analysis object 4) having the number of undetermined elemental parameters of 1, are selected from analysis objects arranged in sort up of the number of undetermined elemental parameters 20021 in advance. Calculation order of these analysis objects is analysis object 8, analysis object 3 and analysis object 4, namely the same as the arrangement order of the analysis objects, and this order is stored in calculation order list 20023. In the first round process 2004 to be executed in the one-variable equation solving step 2003, undetermined elemental parameter is determined based on calculation order list. Because the first calculation is on analysis object 8, calculated value is obtained first, using elemental parameter r_a,8, and value of r_a,8 is determined so as to be able to reproduce experimental value E₈. In this process, value that can reproduce experimental value is searched while changing r_a,8, therefore, iterated solving of one-variable equation is required. Similar process is executed on the 2 selected analysis objects left. In the first round process 2006 to be executed in the atomic parameters update step 2005, expression formula of elemental parameter as a function of value of elemental attribute, is updated on a, b and h that are the elemental type corresponding to determined elemental parameter 20061. For example, the elemental parameter 200062 of an elemental type a, can be expressed as r_a(q)=c₀+c₁p₁; where c₀ and c₁ are coefficients. Then, undetermined value of elemental parameter classified to elemental type a is obtained by substitution of value elemental attribute p₁ into r_a(q)=c₀+c₁p₁. However, as for the element 20063, wherein elemental attribute value is apart from that of element used for the parameter determination by certain threshold or more, there may be an option of not setting elemental parameter. The “value assigned” element 2010 is shown by a filled circle, and the value not assigned, namely the “undetermined element” 2011 is shown by an open circle. Then, 6 analysis objects, 5, 6, 7, 2, 1 and 9, which are undetermined analysis object at this time, are sorted up by the number 20064 of undetermined elemental parameter. In addition, in the case where the number of undetermined parameters is the same, the analysis objects are sort down by the number 20065 of elemental parameters. Here, analysis objects having the number of undetermined parameters of equal to or larger than 1 become new undetermined molecules. Analysis object 2 has the number of undetermined parameters of 0 at this moment, and thus is not used in elemental parameter determination. In the second round process 2007 to be executed in calculation order list generation step 2001, the analysis object 20072 (analysis object 5, and analysis object 6) having the number of undetermined elemental parameters 20071 of 1, are selected from analysis objects arranged in advance. Calculation order of these analysis objects is analysis object 5, and analysis object 6, and this order is stored in calculation order list 20073. In the second round process 2008 to be executed in the one-variable equation solving step 2003, undetermined elemental parameter is determined based on calculation order list. Calculated values are obtained first using elemental parameter r_a,5, under fixed value condition, and undetermined elemental parameter r_d,5, to determined r_d,5 so as to reproduce experimental value E₅. In this process, value that can reproduce experimental value is searched while changing only r_d,5, therefore, process to be executed is iterated solving of one-variable equation, even for an analysis object composed of 2 elemental parameters. Similar process is executed on the 1 selected analysis object left. In the second round process 2009 to be executed in the atomic parameters update step 2005, expression formula of elemental parameter, as a function of value of elemental attribute, is updated on a, b, h, d and e that are the elemental type 20091 corresponding to determined elemental parameter. For example, the elemental parameter 20092 of an elemental type a, is updated from r_a(q)=c₀+c₁p₁ to r_a(q)=c′₀+c′₁p₁. After updating expression formula and determination of value of elemental parameter of an element classified to the corresponding elemental type, re-arrangement operation of undetermined analysis objects is iterated in the same manner as the first round of iteration. This process is iterated until an undetermined analysis object is absent.

Finally, the amount of calculations obtained by a conventional parameter determination method and the parameter determination method according to the present embodiment are compared. The amount of calculation in the prior art is described in expression 2. In the present embodiment, analysis objects are sequentially selected, and calculation is iterated on each of the analysis objects until the amount of change of calculated value becomes below energy convergence threshold. The number of iteration is represented by I(n), which is different by an analysis object, however, there are analysis objects not used in elemental parameter determination. Therefore, the amount of calculation is calculated by “expression 8” using δ function which is defined 1 for analysis objects used in elemental parameter determination, and 0 for analysis objects not used. $\begin{array}{cc}{T}_{\mathrm{invention}}=\sum _{n=1}^N\delta \left(n\right)\left(1+I\left(n\right)\right)f\left(A\left(n\right)\right)& \left(\mathrm{Expression}8\right)\end{array}$

Comparison between expression 2 and expression 8 shows smaller amount of calculation in a method of present embodiment as compared with the prior art in view of three points. Firstly, calculation is executed on all analysis objects in the prior art, while in the present embodiment, calculation is executed on a part of the analysis objects by presence of the 5 function, by which the amount of calculation is reduced. Secondly, the first term in the parenthesis at the right-hand side of the second line of expression 2 is “I” in the prior art, while in the present embodiment, the first term in the parenthesis at the right-hand side of expression 8 is “1”, by which the amount of calculation is reduced. Thirdly, the second term in the parenthesis is “I·p(n)” in the prior art, while in the present Embodiment, the first term in the parenthesis is “I(n)”. The number of iteration, I, required for solving of a matrix-equation as a multi-variable non-linear problem, is considered to be larger than the the number of iteration “I(n)” required for solving of a one-variable equation as a one-variable non-linear problem, therefore the amount of calculation in the present Embodiment is considered less.

Embodiment 4

Explanation will be given below on the atomic parameter determination method under the fixed acceptable error condition relevant to the present embodiment of this invention, with reference to FIG. 21 and FIG. 22.

FIG. 21 is a flowchart of the atomic parameter determination method under the fixed acceptable error condition. First of all, the acceptable energy error ε 21011, the energy convergence threshold τ 21012, the three-dimensional coordinate information and experimental value 21013 of salvation energy for a molecule and the atomic type, atomic charge, and averaged bond distance 21014 for each atom composing a molecule are input from the input apparatus 2101. Subsequently, in the initialization step 2102, the initialization step 21021 of atomic parameter set by each atomic type, and the sorting up step 21022 of all molecules by the number of undetermined atomic parameters are executed. Then, in calculation order list generation step 2103, the molecule selection step 21031 of a molecule having the number of undetermined atomic parameters of equal to or less than 1, from a molecule containing undetermined atomic parameter and the calculation order generation step 21032, by determination of sort order of selected molecules as calculation order, are executed. Then in the one-variable equation solving step 2104, solving of one-valuable equation is iterated, in the case where a molecule is judged to be included in the judgment step 21041 whether a molecule on the top of calculation order list contains undetermined atomic parameters or not, until the amount of change of calculated value of solvation energy becomes below τ, and the undetermined atomic parameter determination step 21042 is executed. Next, in spite of the results of the judgment step 21041, the step 21043 calculating the difference between calculated value and experimental value of the molecule by using determined atomic parameter, and the step 21044 comparing absolute value of calculation error with acceptable error are executed, and in the case where the calculation error is over the acceptable energy error, the procedure is returned to the step 21041 via the setting afresh step 21045 of undetermined atomic parameter, while in the case where the calculation error is equal to or smaller than the acceptable energy error, calculation order list update step 21046 is executed. In calculation order list termination judging step 21047, in the case where a molecule is still present on the top of calculation order list, procedure is returned to the step 21041; however, in the case where a molecule is not present any more, procedure is forwarded to the atomic parameters update step 2105. Next, in the atomic parameters update step 2105, as for an atomic type corresponding to determined atomic parameter, the atomic parameter expression formula update step 21051, the equation being a function of atomic charge and averaged bond distance, the step 21052 for tentatively determining values of atomic parameters for some or all of undetermined atoms, the sorting up step 21053 of undetermined molecules by the number of undetermined atomic parameters, the sorting down step 21054 of molecules with the identical number of undetermined atomic parameters by the number of atomic parameters and the undetermined molecule update step 21055 are executed. Next, in the atomic parameter calculation termination judging step 2106, whether an undetermined molecule is present or not, is judged. In the case where the undetermined molecule is present, calculation order list generation step 2103, the one-variable equation solving step 2104 and the atomic parameters update step 2105 are iterated. In the case where the undetermined molecule is not present, the value 21071 of determined atomic parameters is output by the output apparatus 2107.

FIG. 22 is details of processes in calculation order list generation step, the one-variable equation solving step, and the atomic parameters update step, in the atomic parameter determination method. As an example, explanation will be given on the first and the second rounds of iterations of each of the steps, in the case of the number of molecules of 9, and the number of atomic types of 8. In the first round process 2202 to be executed in calculation order list generation step 2201, the molecules 22022 (molecule 8, molecule 3 and molecule 4) having the number of undetermined atomic parameters of 1 or less are selected from molecules sorted up by the number 22021 of undetermined atomic parameters in advance. Calculation order of these molecules is molecule 8, molecule 3 and molecule 4, namely the same as the arrangement order of the molecules, and this order is stored in calculation order list 22023. In the first round process 2204 to be executed in the one-variable equation solving step 2203, undetermined atomic parameter is determined based on calculation order list. Because the first calculation is on molecule 8, salvation energy is calculated first, using atomic parameter r_a,8, and value of r_a,8 is determined so as to reproduce experimental value E₈. In this process, value that can reproduce experimental value is searched while changing r_a,8, therefore, iterated solving of one-variable equation is required. Similar process is executed on the 2 selected molecules left. In the first round process 2206 to be executed in the atomic parameters update step 2205, expression formula of atomic parameter as a function of atomic charge and averaged bond distance, is updated on a, b and h that are the atomic type 22061 corresponding to determined atomic parameter. Expression formula of atomic parameter may be a function not having value of an atomic attribute as a variable. For example, the atomic parameter 22062 of atomic type a can be expressed as r_a(q, 1)=c₀=r_a,8. Undetermined atomic parameter value classified to atomic type a is tentatively determined as r_a(q, 1)=c₀ However, as for the atom 22063, wherein atomic charge or averaged bond distance is apart from atomic charge or averaged bond distance of atom used in parameter determination, by certain threshold or more, there may be an option of not tentatively determining atomic parameter value. The atomic parameter determined atom 2210 is shown by a filled circle, the tentatively determined atom 2211 is shown by filled triangle, and the value not assigned, namely undetermined atom 2212 is shown by an open circle. Then, 6 molecules 5, 6, 7, 2, 1 and 9, which are undetermined molecules at this time, are sorted up by the number 22064 of undetermined atomic parameters. In addition, in the case where the number of undetermined parameters is the same, the molecules are sorted down by the number 22065 of atomic parameters. Here, all of the 6 molecules become newly undetermined molecules. In the second round process 2207 to be executed in calculation order list generation step 2201, the molecule 22072 (molecule 2, molecule 5, and molecule 6) having the number 22071 of undetermined atomic parameters of 1 or less, are selected from molecules arranged in advance. Calculation order of these molecules is molecule 2, molecule 5, and molecule 6, and this order is stored in calculation order list 22073. In the Second round process 2208 to be executed in the one-variable equation solving step 2203, undetermined atomic parameter is determined based on calculation order list. Explanation will be given particularly on molecule 2 having no undetermined parameters. Firstly, salvation energy is calculated, using atomic parameter given. Then, whether absolute value of calculation error, defined as difference between calculated value and experimental value, is below acceptable energy error or not, is confirmed. In the case where absolute value of calculation error is equal to or larger than acceptable energy error, only one of the 2 atomic parameters r_a,2 and r_h,2 contained in molecule 2 is set newly as an undetermined atomic parameter. Atomic parameter to be selected is one having larger solvation energy gradient relative to change in both atomic parameters. In the case where the selected parameter is r_a,2, value of r_a,2 is determined so as to reproduce experimental value E₂. In the Second round process 2209 to be executed in the atomic parameters update step 2205, expression formula of atomic parameter, as a function of atomic charge and averaged bond distance, is updated on a, b, h, d and e that are the atomic type 22091 corresponding to determined atomic parameter. For example, the atomic parameter 22092 of atomic type “a” was r_a(q, 1)=c₀=r_a,8, however, because use of newly determined value of r_a,2 is also required for atomic parameter of an atomic type “a”, based on the calculation result of molecule 2, use of more complicated function form is required. For example, it may be updated to r_a(q, 1)=c′₀+c′₁q. After updating expression formula and determination of atomic parameter value of an atom classified to the corresponding atomic type, re-arrangement operation of undetermined molecules is iterated in the same manner as the first round of iteration. This process is iterated until an undetermined molecule is absent.

Explanation will specifically be given on a mounting example of process in relation to a part different from process in embodiment 1, among the steps of a flowchart of the atomic parameter determination method under the fixed acceptable error condition, shown by FIG. 21.

FIG. 23 is details of calculation order list update step. At the time of the start 2301, calculation of a molecule on the top of calculation order list is over. In the new setting judgment step 2302 of undetermined parameter, whether a process for newly setting one of the tentatively determined atomic parameters as an undetermined atomic parameter is terminated or not, is judged in a determination process of undetermined parameter of the molecule. In the case where the determination results is “yes”, whole delete process 2303 of a calculation order list is executed; while the results is “no”, the delete process 2304 of molecule on the top from a calculation order list and sequentially move forward the residual molecules is executed. Then, in calculation order list termination judging step 2305, whether the calculation list is empty or not, is judged; in the case 2306 of empty, process goes forward to the atomic parameters update step; in the case 2307 of not empty 2307, process returns to the undetermined atomic parameter determination step.

Explanation will be given on the atomic parameter, r_t(q, 1), update step 21051 of atomic type corresponding to determined atomic parameter, using FIGS. 24A and 24B, and FIGS. 25A and 25B.

FIG. 24A shows an example of atomic parameter update in the case that the atomic parameter value 2401 is expressed by a function of value 2402 of an atomic attribute such as atomic charge as a variable. A filled circle in the drawing shows the “value assigned” atom 2403, and an open circle shows the “value not assigned” atom 2404. First of all, by using the “value assigned” atom 2403, the linear function 2405 that correlates atomic parameter value and value of an atomic attribute is defined. Then, the value 2406 of an atomic attribute of a “value not assigned” atom is substituted to a linear function to tentatively determine atomic parameter value.

FIG. 24B shows an example of execution result of the step for preparation of calculation order list, the step for solving a one-variable equation, and the step for atomic parameter-updating, using tentatively determined atomic parameter value. There are the atom 2407 having value of tentatively determined atomic parameter set as determined value as it is and the atom 2408 having re-determined atomic parameter, after tentatively determined atomic parameter value being set as undetermined atomic parameter. Description of atomic parameter values of these atoms by a linear function of value of an atomic attribute varies already determined atomic parameter values and also salvation energy of molecules with already determined atomic parameter. However, definition of the three regions 2409 based on value of an atomic attribute and definition of the linear function 2410 relative to each region can express by a function using newly determined atomic parameters, without changing salvation energy of molecules with already determined atomic parameter.

FIG. 25A shows an example of atomic parameter update in the case that the atomic parameter value 2501 is expressed by a function of values of two atomic attributes such as the atomic charge 2502 and the averaged bond distance 2503 as variables. A filled circle in the drawing shows the “value assigned” atom 2504, and an open circle shows the “value not assigned” atom 2505. First of all, by using the “value assigned” atom 2504, the linear function correlating atomic parameter value with value of an atomic attribute 2506 is defined. Then, atomic charge and the averaged bond distance 2507 of the “value not assigned” atom are substituted in a linear function to tentatively determine atomic parameter value.

FIG. 25B shows an example of execution result of the step for preparing calculation order list, the step for solving a one-variable equation, and the step for updating atomic parameter, using tentatively determined atomic parameter value. There exist the atom 2508, which the tentatively determined value became the determined value as it is, and the atom 2509, which the tentatively determined value was set as undetermined atomic parameter, and atomic parameter was re-determined. Description of values of atomic parameters of these atoms by a linear function of values of atomic attributes varies already determined atomic parameter values, and also salvation energy of molecules with already determined atomic parameter. However, definition of the two regions 2510 on a plane formed by atomic charge and averaged bond distance, and definition of the linear function 2511 defined in each region based on value of an atomic attribute can express by a function using newly determined atomic parameters, without changing salvation energy of molecules with already determined atomic parameter.

Finally, the amount of calculations in a conventional atomic parameter determination method and the atomic parameter determination method according to the present embodiment are compared. The amount of calculation in the prior art is described in expression 2. In the present embodiment, molecules are sequentially selected, and calculation is iterated on each of the molecules until the amount of change of calculated value of salvation energy becomes below energy convergence threshold. The number of iteration is represented by I(n), which is different by a molecule, however, there are also molecules having an undetermined atomic parameter number of 0. Therefore, the amount of calculation is calculated by “expression 9” using δ function, which is defined 1 for molecules requiring determination process of undetermined atomic parameter, and 0 for molecules not requiring the process. $\begin{array}{cc}{T}_{\mathrm{invention}}=\sum _{n=1}^N\left(1+I\left(n\right)·\delta \left(n\right)\right)f\left(A\left(n\right)\right)& \left(\mathrm{Expression}9\right)\end{array}$

Comparison between expression 2 and expression 9 shows smaller amount of calculation in a method of present embodiment as compared with the prior art in view of two points. Firstly, the first term in the parenthesis at the right-hand side of the second line of expression 2 is “I” in the prior art, while in the present Embodiment, the first term in the parenthesis is “1”, by which the amount of calculation is reduced. Secondly, the second term in the parenthesis is “I·p(n)” in the prior art, while in the present Embodiment, the first term in the parenthesis is “I(n)·δ(n)”, which reduces the amount of calculation by 2 reasons. One is that the number of iteration, I, required for solving of a matrix-equation as a multi-variable non-linear problem, is considered to be larger than the number of iteration “I(n)” required for solving of a one-variable equation as a one-variable non-linear problem. The second is that minimal value of the number of atomic type p(n) included in molecule n is 1, but δ function takes a value of 0 or 1.

Furthermore, difference of the amount of calculation between in the prior art and in a method of the present embodiment increases when number of molecules N is large. Explanation will be given below on reason therefor.

In a method of the present embodiment, atomic parameter is expressed by a function having atomic charge and averaged bond distance as variables. Form of the function becomes complicated during the processes of selection of a molecule and determination of undetermined parameters. However, range of atomic charge and averaged bond distance that can be taken by an atom is finite. Therefore, after sophistication of function form to a certain degree, it is considered that atomic parameter value can tentatively be determined in high accuracy for every combination of atomic charge and averaged bond distance. Therefore, in the case where the number of molecules is large, as for a molecule selected at the final phase of parameter determination process, probability of tentative determination of atomic parameter value in high accuracy becomes high for all atoms composing the molecule. This corresponds to the fact that, in corresponding the summation order on the molecule n in expression 9, with the calculation order of molecules used in atomic parameter determination, probability of δ(n)=1 is high when n is small, and probability of δ(n)=0 is high with increase of n. Small molecule number N is considered not to increase probability of δ(n)=0 so much, even at the final phase of parameter determination process, but large molecule number N is considered to increase probability of δ(n)=0.

In addition, a method of the present Embodiment has effect of reducing calculation number of a large molecule. Discussion was made above on small and large of constant term part, which is multiplied to f(A(n)), in comparison of expression 2 with expression 9, which show the amount of calculation required for atomic parameter determination, however, because f(A(n)) is a power function of atomic number A(n) of the molecule n, reduction of calculation number of a molecule with large A(n) is more important. Selection order of molecules in the present embodiment is, in general, “a small molecule early and a large molecule later”; a firstly selected molecule is a molecule having an atomic parameter number of 1; a molecule having a non-hydrogen atom number of 1 such as methane, ammonia or the like corresponds to this type. Although there is an exception that a benzene molecule has an atomic parameter number of 1, due to having high symmetry, in spite of containing 6 carbon atoms, usually a small molecule is preferentially selected, and a large molecule is selected at the final phase of parameter determination process. As discussed above, probability of δ(n)=0 is high at the final phase of parameter determination process; namely probability that calculation iteration is required is high, to determine undetermined atomic parameter.

Summary of the above discussion will be given in FIG. 26. It is assumed that the number of molecules N used in the atomic parameter determination is assumed sufficiently large, and in addition, molecules are arranged in calculation order in the present embodiment, relative to summation symbol of expression 2 and expression 9. In spite of the early calculation stage 2601, namely when n is small, and the last calculation stage 2602, namely when n is large, the amount 2603 of calculation for the molecule n in the prior art is constant. Because it is an iterative calculation method, in each time, for solvation energy and solvation energy gradient relative to change in atomic parameter value, for all the atoms, change in calculation order of molecules does not effect the amount of calculation. On the other hand, because the amount 2604 of calculation for the molecule n in embodiment of this invention has high probability of execution of determination process of undetermined parameter at the early phase of calculation, value of δ(n)=1 is taken in expression 9. However, because in the early calculation stage 2601, probability is high that a selected molecule is small, and thus the amount f(A(n)) 2605 of calculation required for calculation of solvation energy in one time is small as result. The amount 2604 of calculation for the molecule n at the last calculation stage, value of δ(n)=0 is taken in expression 9, because possibility is high that atomic parameters are tentatively determined for all atoms composing a molecule. Because in the last calculation stage 2602, probability is high that of a selected molecule is large, the amount f(A(n)) 2605 of calculation of solvent energy in one time, is large as result.

Embodiment 5

The present embodiment shows atomic parameter determination result under the fixed acceptable error condition using 63 molecules. Contents of molecules used in atomic parameter determination include 18 cations, 11 anions and 34 neutral molecules. The acceptable error was set 1.4 kcal/mol for the cations and the anions, and 0.20 kcal/mol for the neutral molecules. As a result, 57 atomic parameters were obtained.

FIG. 27 is comparison of the amount of calculation in Embodiment of this invention and in prior art, in the atomic parameter determination method using 63 molecules. In the present embodiment, the averaged number 2701 of iteration of calculation of solvation energy of each of the molecules in sequential determination process of 57 atomic parameters was 3.32 times. In addition, the amount 2702 of calculation, defined by calculation time required for atomic parameter determination using one computer, was about 504 minutes. Calculation time in using a gradient matrix could not actually be measured, however, the lower bound 2703 of the amount of calculation was estimated as 3183 minutes by the following calculation. Firstly, in expression 6, there is actually measured value of the amount f(A(n)) of calculation, required for salvation energy calculation in one time, on each of molecules n. In addition, the number of atomic type contained in each of the molecules was determined using definition of atomic type, by Tomasi et al, (V. Barone, M. Cossi and J. Tomasi, “A new definition of cavities for the computation of salvation free energies by the polarizable continuum model”, Journal of Chemical Physics, Vol. 107, 3210, (1997)]. Here, unknown value left is only the number of iteration I. Minimization of objective function defined by a matrix-equation is a multi-variable non-linear problem, and many methods for reducing the number of iteration required for solving are present. However, the number of iteration I is considered more than 3.32 times that is averaged value of the number of iteration required for solving a one-variable equation. Accordingly, the number of iteration was estimated to be the minimal integer I=4, equal to or larger than 3.32. From the result, the atomic parameter determination method of the present embodiment can be said to provide the amount of calculation about 6.3 times less as a minimum, as compared with the prior art. In the prior art, it is required that atomic type is re-defined to iterate whole process of atomic parameter determination, when desired accuracy is not obtained, therefore, when this number of iteration is taken as k, difference in the amount of calculation becomes 6.3 k times.

Embodiment 6

Generalization of the atomic parameter determination method by the fixed acceptable error condition, of the present embodiment, is possible by substitution of a molecule with “an analysis object”, atoms composing a molecule with “elements composing an analysis object”, an atomic parameter with “an elemental parameter”, experimental value of salvation energy with simply “experimental value”, and calculation value of solvation energy with simply “calculation value”. Explanation will be given below on the parameter determination method by the fixed generalized acceptable error condition, with reference to FIG. 28 and FIG. 29.

FIG. 28 is a flowchart of simulation parameter determination method under the fixed acceptable error condition. First of all, the acceptable error ε 28011, the convergence threshold τ 28012, the information necessary in calculation of an analysis object and experimental value 28013 and the elemental type and element attribute value 28014 for each element composing an analysis object are input from the input apparatus 2801. Subsequently, in the initialization step 2802, set by each of atomic types, the elemental parameter initialization step 28021 and the sorting up step 28022 of all analysis objects by the number of undetermined elemental parameters are executed. In calculation order list generation step 2803, the selection step 28031 of analysis objects having undetermined elemental parameter number of equal to or less than 1, from analysis objects containing undetermined elemental parameter, and orders of arrangement and calculation of the analysis objects selected are determined to execute calculation order list generation step 28032 are executed. Then in the one-variable equation solving step 2804, in the case where “inclusion” is judged, in the judgment step 28041 on whether an analysis object on the top of a calculation order list contains undetermined elemental parameter or not, the solving of a one-valuable equation is iterated on an analysis object on the top of calculation order list, until the amount of change of calculated value becomes below τ, and the undetermined elemental parameter determination step 28042 is executed. Then in spite of the result of the judgment step 28041, the calculation error computing step 28043, as difference between calculation value of the analysis object using elemental parameter, and experimental value, and the comparing step 28404 of calculation error with acceptable error are executed, and in the case where the calculation error is over the acceptable error, procedure is returned to the step 28041 via the setting afresh step 28045 of undetermined elemental parameter for one of tentatively determined atomic parameters; in the case where the calculation error is equal to or less than the acceptable error, calculation order list update step 28046 is executed. In calculation order list termination judging step 28047, in the case where an analysis object is still present on the top of calculation order list, procedure is returned to the step 28041; in the case where an analysis object is not present any more, procedure is forwarded to the elemental parameters update step 2805. Next, in the elemental parameters update step 2805, as for an elemental type corresponding to determined elemental parameter, the elemental parameter expression formula update step 28051, the equation being a function of value of elemental attribute, the tentatively determining step 28052 of some or all of values of elemental parameters of an undetermined element, the sorting up step 28053 of undetermined analysis objects by number of undetermined elemental parameters, the sorting down step 28054 of analysis objects with the identical number of undetermined elemental parameters by number of elemental parameters and the undetermined analysis object update step 28055 are executed. In the elemental parameter calculation termination judging step 2806, whether an undetermined analysis object is still present or not, is judged. In the case where the undetermined analysis object is present, calculation order list generation step 2803, the one-variable equation solving step 2804 and the elemental parameters update step 2805 are iterated; and in the case where the undetermined analysis object is not present, the determined elemental parameters 28071 is output by the output apparatus 2807.

FIG. 29 is details of processes in calculation order list generation step, the one-variable equation solving step, and the elemental parameters update step, in an elemental parameter determination method. As an example, explanation will be given on the first and the second rounds of iterations of each of the steps, in the case of the number of analysis objects of 9, and the number of elemental type of 8. In the first round process 2902 to be executed in calculation order list generation step 2901, the analysis object 29022 (analysis object 8, analysis object 3 and analysis object 4) having the number of undetermined elemental parameters of 1 or less are selected from analysis objects sorted up by the number 29021 of undetermined elemental parameters in advance. Calculation order of these analysis objects is analysis object 8, analysis object 3 and analysis object 4, namely the same as the arrangement order of the analysis objects, and this order is stored in calculation order list 29023. In the first round process 2904 to be executed in the one-variable equation solving step 2903, undetermined elemental parameter is determined based on calculation order list. Because the first calculation is on analysis object 8, calculated value is obtained, using elemental parameter r_a,8, and value of r_a,8 is determined so as to reproduce experimental value E₈. In this process, value that can reproduce experimental value is searched while changing r_a,8, therefore, iterated solving of a one-variable equation is required. Similar process is executed on the 2 selected analysis objects left. In the first round process 2906 to be executed in the elemental parameters update step 2905, expression formula of elemental parameter as a function of values of elemental attributes, is updated on a, b and h that are the elemental type 29061 corresponding to determined elemental parameter. Expression formula of elemental parameter may be a function not having value of elemental attribute as a variable. For example, the elemental parameter 29062 of an elemental type a, can be expressed as r_a(p₁,p₂)=c₀=r_a,8. Then, undetermined elemental parameter value classified to elemental type a is obtained as r_a(p₁,p₂)=c₀. However, on the element 29063, wherein elemental attribute value is apart from that of other element belonging to the identical elemental type by certain threshold or more, there may be an option of not tentatively determine elemental parameter. The element 2910 having determined value of elemental parameter is shown by a filled circle, tentatively determined element 2911 is shown by a filled triangle, and the value not assigned, namely the undetermined element 2912 is shown by an open circle. Then, at this time, 6 analysis objects 5, 6, 7, 2, 1 and 9 are re-arranged in sort up of the number 29064 of undetermined elemental parameter. In addition, in the case where the number of undetermined parameters is the same, the analysis objects are re-sorted down by the number 29065 of elemental parameters. Here, all of these 6 analysis objects become undetermined analysis objects newly. In the second round process 2907 to be executed in calculation order list generation step 2901, the analysis object, having the number 29072 (analysis object 2, analysis object 5 and analysis object 6) of undetermined elemental parameters 29071 of 1 or less, are selected from analysis objects arranged in advance. Calculation order of these analysis objects is analysis object 2, analysis object 5 and analysis object 6, and this order is stored in calculation order list 29073. In the second round process 2908 to be executed in the one-variable equation solving step 2903, undetermined elemental parameter is determined based on calculation order list. Explanation will particularly be given on analysis object 2 not having undetermined parameter. First of all, calculation value is obtained using an elemental parameter given. Then, whether absolute value of calculation error, defined as difference between calculated value and experimental value, is below acceptable error or not, is confirmed. In the case where absolute value of calculation error is equal to or larger than acceptable error, any one of the 2 elemental parameters, r_a,2, and r_h,2, contained in analysis object 2 is set newly as undetermined elemental parameter. The elemental parameter to be selected should be one having larger gradient of calculation value relative to change in both elemental parameters. In the case where the selected parameter is r_a,2, value of r_a,2 is determined so as to reproduce experimental value E₂. In the second round process 2909 to be executed in the atomic parameters update step 2905, expression formula of elemental parameter, as a function of value of elemental attribute, is updated on a, b, h, d and e that are the elemental type 29091 corresponding to determined elemental parameter. For example, the elemental parameter 29092 of an elemental type a, is updated from r_a(p₁,p₂)=c₀=r_a,8, however, use of more complicated function form is required, because use of newly determined value of r_a,2 is required as an elemental parameter of elemental type a, based on calculation result of analysis object 2. For example, updating to r_a(p₁,p₂)=c′₀+c′₁p₁ may be allowed. After updating expression formula and determination of value of elemental parameter of an element classified to the corresponding elemental type, re-arrangement operation of undetermined analysis objects is iterated in the same manner as the first round of iteration. This process is iterated until an undetermined analysis object is absent.

Finally, the amount of calculations in the parameter determination method by the prior art and the parameter determination method according to the present embodiment are compared. The amount of calculation in the prior art is described in expression 2. In the present embodiment, analysis objects are sequentially selected, and calculation is iterated on each of the analysis objects until the amount of change of calculated value becomes below convergence threshold. The number of iteration is expressed by I(n), which is different by an analysis object, however, there are analysis objects having an undetermined elemental parameter number of 0. Therefore, the amount of calculation is calculated by “expression 10” using δ function, which is defined 1 for analysis objects requiring undetermined elemental parameter determination process, and 0 for analysis objects not required the process. $\begin{array}{cc}{T}_{\mathrm{invention}}=\sum _{n=1}^N\left(1+I\left(n\right)·\delta \left(n\right)\right)f\left(A\left(n\right)\right)& \left(\mathrm{Expression}10\right)\end{array}$

Comparison between expression 2 and expression 10 shows smaller amount of calculation in a method of present embodiment as compared with the prior art in view of 2 points. Firstly, the first term in the parenthesis at the right-hand side of the second line of expression 2 is “I” in the prior art, while in the present Embodiment, the first term in the parenthesis is “1”, by which the amount of calculation is reduced. Secondly, the second term in the parenthesis is “I·p(n)” in the prior art, while in the present Embodiment, the first term is “I(n)·ε(n)”, therefore, the amount of calculation is less by the following two reasons. One is that the number of iteration, I, required for solving a matrix-equation as a multi-variable non-linear problem, is considered to be larger than the number “I(n)” of iteration required for solving a one-variable equation as a one-variable non-linear problem. Second is that minimum value of the number p(n) of elemental type contained in analysis object n is 1; on the other hand, 6 function takes a value of either 0 or 1.

Embodiment 7

Explanation will be given on an example of mounting example of user interface accompanying with the atomic parameter determination method, using FIG. 30 to FIG. 33.

FIG. 30 is a flowchart in the atomic parameter determination method including the information display step. Explanation was already given in detail, in embodiment 1 and embodiment 4, on sections not relating to information display, therefore such explanation is simplified here. In the atomic parameter determination according to the present embodiment, input step 3001 and the initialization step 3002 are executed first. Then, in calculation order list generation step 3003, a calculation order list is displayed on a screen by calculation order list display step 30032 via the inner process 30031. A user may add change in the calculation order displayed on a screen, and finally settles the calculation order. In the calculation progress display step 30033, outline of progress of atomic parameter calculation and a calculation progress by a molecule are displayed. Then, the equation solving step 3004 is executed. In the atomic parameters update step 3005, via the inner process 30051, in the atomic parameter display step 30052, atomic parameter is displayed using space formed by atomic parameter, atomic charge and averaged bond distance. Finally, the output step 3007 is executed via the termination judging step 3006. In the above flowchart, sections relating to information display are 3 parts; calculation order list display step 30032, the calculation progress display step 30033 and the atomic parameter display step 30052.

FIG. 31 is an example of screen display in calculation order list display step. Calculation order list tab 31011 is selected among 3 display objects contained in the display object displaying tab 3101. Display view of calculation order list is divided into 3; namely, the calculation order display region 3102, the distribution diagram display region 3103 and the molecular structure map display region 3104. At the early display stage, calculation order list 31021 is displayed in the calculation order display region 3102. These tables can be selected line by line, and in this drawing molecule 5 is in a selected status. Calculation order of selected molecules can be adjusted by the up-arrow button 31022 or the down-arrow button 31023. In addition, by the delete button 31024, “no-calculation” choice on the selected molecule is also possible. When a user pushes the confirm button 31025, calculation order is settled. Supporting data for a user to judge change in calculation order includes a distribution chart and a molecular structure chart. In the distribution diagram display region 3103, an example of atomic distribution chart is shown where undetermined atomic parameters of molecule 5 are classified into atomic type d, which belongs. Atoms are plotted on a plane formed by atomic charge and averaged bond distance. The plot has a different shape or a pattern by reflecting a status of “value assigned”, “value not assigned” or the like of atomic parameter corresponding to each atom, or a selection status in calculation order list 31021. In the present embodiment, atoms classified to atomic type d are only 2, and both are plotted at positions apart on a plane; therefore, it is visually understood that even if one atomic parameter is determined, the other atomic parameter cannot necessarily be tentatively determined in high accuracy based on thus determined atomic parameter. Based on this information, a user may choose to descend calculation priority order of molecule 5. In the molecular structure map display region 3104, a molecular conformation and a sphere having radius corresponding to atomic parameter value of each molecule forming the molecule are displayed in an overlapped way. The sphere is displayed also here with a different pattern by reflecting a status of “value assigned”, “value not assigned” or the like of atomic parameter corresponding to each atom, or a selection status in calculation order list 31021. Molecule 5 is a molecule composed of 3 non-hydrogen atoms, and two of them are understood to be equivalent in view of symmetry. Atomic parameter value for the atom X has tentatively been determined already, and value has been assigned. On the other hand, it is visually understood that atomic parameter value for the atom Y will be determined from now. Atoms in a molecule displayed in a plot in the distribution diagram display region 3103 or in the molecular structure map display region 3104 may also be selected. When a plot is selected a structure of a molecule to which the corresponding atom belongs is displayed in the molecular structure map display region 3104. In addition, when an atom is selected, a distribution chart of an atom related to corresponding atomic type is displayed in the distribution diagram display region 3103. These functions are capable of furnishing information of other molecules directly relating to parameter determination of molecule 5 noticed.

FIG. 32 is an example of screen display in the calculation progress display step. The progress tab 32011 is selected among 3 display objects contained in the display object displaying tab 3201. Display view of a calculation progress is divided into the display region of summary of an atomic parameter calculation progress 3202 and the display region of an atomic parameter calculation progress for each molecule 3203. In the display region of summary of an atomic parameter calculation progress 3202, summing up result of a calculation status of a molecule such as “determined”, “in progress” or “undetermined”, and calculation elapsed time is displayed. In addition, in the display region of an atomic parameter calculation progress for each molecule 3203, a calculation status of each molecule, molecular ID, atomic parameter number and undetermined atomic parameter number are displayed. From these sets of information, a progress degree of parameter determination procedure can be understood.

FIG. 33 is an example of screen display in the atomic parameter display step. The atomic parameter tab 33011 is selected among 3 display objects contained in the display object displaying tab 3301. Display view of atomic parameter is divided into the display region of the atomic parameter determination status of each atomic type 3302, and the display region of atomic parameter value 3303. At the early display stage, atomic type, total atom number belonging to the corresponding atomic type, atom number with atomic parameter determined, tentatively determined atom number and undetermined atom number are displayed in a Table form in the display region of the atomic parameter determination status of each atomic type 3302. These tables can be selected by each line, and in this drawing, atomic type a is in a selected status. In the display region of atomic parameter value 3303, atomic parameter value in selected atomic type is displayed using space formed by atomic parameter, atomic charge and averaged bond distance. By graphic display of atomic parameter value in three-dimensional space, dependence of atomic parameter on atomic charge and averaged bond distance can intuitively be understood.

Description will be given below on effect of the above-described embodiments.

By a system to preferentially select a molecule containing atomic parameter that can be determined in high certainty, and to control calculation order of solvation energy relating to selected molecule by calculation order list, according to the present embodiment, atomic parameter determination is possible by smaller amount of calculation as compared with the prior art.

In embodiment 1, determination possibility of atomic parameter used in calculation of solvation energy of a molecule, by smaller amount of calculation as compared with the prior art was shown. In embodiment 2, by demonstration calculation using 10 molecules, it was shown that the atomic parameter determination method of the present embodiment is capable of determining atomic parameter with equivalent accuracy, by smaller amount of calculation as compared with a method based on the prior art. In embodiment 4, it was shown that atomic parameter used in calculation of salvation energy of a molecule can be determined by smaller amount of calculation as compared with the prior art, and that calculation error can also be controlled by the atomic parameter determination method with acceptable error fixed. In embodiment 5, it was shown, by demonstration calculation using 63 molecules, that the amount of calculation required for the atomic parameter determination method under the fixed acceptable error condition is smaller than estimation level of the amount of calculation required for a method based on the prior art.

In addition, by a system to preferential select an analysis object containing parameter that can be determined in high certainty, and to control calculation relating to analysis objects selected, by calculation order list, according to the present embodiment, atomic parameter determination is possible by smaller amount of calculation as compared with the prior art.

The embodiment of this invention is capable of dividing calculation object into elements, and defining value of attribute of each of the elements, and thus applicable to the case where evaluation of calculated value or gradient relative to parameter change of calculated value is the rate determining step of calculation.

Furthermore, embodiment of this invention is capable of reflecting user knowledge to parameter determination, because determination process of calculation order of analysis objects, as important elements of the parameter determination method, is visualized, and function that a user can change calculation order is provided. In embodiment 7, a display example of supporting data for a user to judge change in calculation order was shown.

It should be noted, description of reference numerals is as follows.

• 101: Calculation object molecule
• 102: Water-based solvent
• 103: Space occupied by a molecule
• 104: Space occupied by dielectrics
• 105: Sphere arranged on an atom
• 106: Radius of a sphere arranged on an atom
• 201: Input apparatus
• 2011: Input of energy convergence threshold τ
• 2012: Input of a structure and experimental value of salvation energy of a molecule
• 2013: Input of atomic type for each atom composing a molecule
• 202: The atomic parameter initialization step
• 203: The matrix generation step
• 204: The matrix-equation solving step
• 205: The atomic parameters update step
• 206: The atomic parameter calculation termination-judging step
• 207: Output apparatus
• 2071: Output of converged atomic parameter value
• 301: The matrix generation step
• 302: Process to be executed in the matrix generation step
• 3021: Solvation energies of all molecules,
• 3022: Solvation energy gradient of all molecules relative to change in each atomic parameter
• 3023: Matrix composed of molecular number as the number of line, and (the number of atomic type +1) as the number of columnt
• 3024: Matrix element
• 303: The matrix-equation solving step
• 304: Process to be executed in the matrix-equation solving step
• 3041: Matrix
• 3042: Solution vector having the number of element of (the number of atomic type +1)
• 3043: Experimental value vector of solvation energy, having the same element number as the number of molecule
• 3044: Objective function
• 305: The atomic parameters update step
• 306: Process to be executed in the atomic parameters update step
• 3061: New atomic parameter vector
• 3062: Atomic parameter vector before update
• 3063: Vector deleted only top element from the solution vector obtained in the matrix equation-solving step 303
• 401: Atomic type
• 402: The resultant atomic parameter value
• 403: Atomic type that each atom contains
• 501: Calculation amount
• 502: Calculation error
• 503: Plot of lozenge marks
• 601: Input apparatus
• 6011: Input of convergence threshold τ
• 6012: Input of necessary information and experimental value for calculation of analysis object
• 6013: Input of an elemental type for each element composing an analysis object
• 602: The elemental parameter initialization step
• 603: The matrix generation step
• 604: The matrix-equation solving step
• 605: The elemental parameters update step
• 606: The elemental parameter calculation termination-judging step
• 607: Output apparatus
• 6071: Output of converged elemental parameter
• 701: The matrix generation step
• 702: Process to be executed in the matrix generation step
• 7021: Calculated value of all analysis objects
• 7022: Gradients of calculated value of all analysis objects to change of each elemental parameter
• 7023: Matrix composed of analysis object number as the number of line, and (elemental type number +1) as the number of columnt
• 7024: Matrix element
• 703: The matrix-equation solving step
• 704: Process to be executed in the matrix-equation solving step
• 7041: Matrix
• 7042: Solution vector having the number of element of (elemental type number +1)
• 7043: Experimental value vector having the same elemental number as analysis object number
• 7044: Objective function
• 705: The elemental parameters update step
• 706: Process to be executed in the elemental parameters update step
• 7061: New elemental parameter vector
• 7062: Elemental parameter vector before update
• 7063: Vector deleted only top element from the solution vector obtained in the matrix-equation solving step 303
• 801: Input apparatus
• 8011: Input of energy convergence threshold τ
• 8012: Input of three-dimensional coordinate information and experimental value of salvation energy for molecule
• 8013: Input of atomic type, atomic charge, and averaged bond distance for each atom
• 802: The initialization step
• 8021: The initialization step of atomic parameter
• 8022: The sorting up step of all the molecules by the number of undetermined atomic parameters
• 803: Calculation order list generation step
• 8031: The molecule selection step
• 8032: The calculation order determination step
• 804: The one-variable equation solving step
• 8041: The undetermined atomic parameter determination step
• 8042: Calculation order list update step
• 8043: Calculation order list termination judging step
• 805: The atomic parameters update step
• 8051: The atomic parameter expression formula update step
• 8052: The sorting up step of undetermined molecules by the number of undetermined atomic parameters
• 8053: The sorting down step of molecules with the identical number of undetermined atomic parameters by the number of atomic parameters
• 8054: The undetermined molecule update step
• 806: The atomic parameter calculation termination judging step
• 807: The calculation error computing step
• 808: Output apparatus
• 8081: Output of determined atomic parameter value
• 901: Calculation order list generation step
• 902: First round process to be executed in calculation order list generation step
• 9021: The number of undetermined atomic parameters
• 9022: Molecule having the number of undetermined atomic parameter of 1
• 9023: Calculation order list
• 903: The one-variable equation solving step
• 904: First round process to be executed in the one-variable equation solving step
• 905: The atomic parameters update step
• 906: First round process to be executed in the atomic parameters update step
• 9061: Atomic type corresponding to determined atomic parameter
• 9062: Atomic parameter of the atomic type a
• 9063: Atom, wherein atomic charge is apart from that of other atom belonging to the identical atomic type by certain threshold or more
• 9064: The number of undetermined atomic parameters
• 9065: The number of atomic parameters
• 907: Second round process to be executed in calculation order list generation step
• 9071: The number of undetermined atomic parameters
• 9072: Molecule having the number of undetermined atomic parameters of 1
• 9073: Calculation order list
• 908: Second round process to be executed in the one-variable equation solving step
• 909: Second round process to be executed in the atomic parameters update step
• 9091: Atomic type corresponding to determined atomic parameter
• 9092: Atomic parameter of atomic type a
• 910: “Value assigned” atom
• 911: “Value not assigned” (undetermined) atom
• 1001: Three-dimensional coordinate information of the molecule
• 1002: Structure of para-xylene molecule
• 1003: Experimental value of salvation energy
• 1004: Experimental value of solvation energy of a para-xylene molecule
• 1005: The number of atomic parameters
• 1006: The number of atomic parameters of a para-xylene molecule
• 1101: Atomic type
• 1102: Atomic charge
• 1103: Averaged bond distance
• 1201: Atomic information
• 1202: Atomic parameter value
• 1203: Status of atomic parameter
• 1204: Status of atomic parameter (value assigned)
• 1205: Status of atomic parameter (value not assigned)
• 1206: Molecular information
• 1207: The number of “value assigned” atomic parameters contained in a molecule
• 1208: The number of “value not assigned” atomic parameters contained in a molecule
• 1209: Calculation status of the molecule
• 1210: Initial value of molecular information
• 1211: Initial value of number of “value assigned” atomic parameters contained in the molecule
• 1212: Initial value of Number of “value not assigned” atomic parameters contained in the molecule
• 1213: Initial value of calculation status of the molecular
• 1301: Calculation order list
• 1302: The input file generation step
• 1303: Input file for solvation energy calculation
• 1304: The salvation energy calculation job input step
• 1305: Computer
• 1306: Solvation energy calculation program
• 1307: The calculation result outputting step
• 1308: Output file for salvation energy calculation result
• 1309: The output result processing step
• 1401: Atomic parameter value
• 1402: Value of an atomic attribute
• 1403: Atom having “value assigned” atomic parameter
• 1404: Atom having “value not assigned” atomic parameter
• 1405: A linear function correlating atomic parameter value with value of an atomic attribute
• 1406: Value of an atomic attribute of a “value not assigned” atom
• 1407: “Value not assigned” atom having value of an atomic attribute far apart from value of an atomic attribute of a “value assigned” atom
• 1501: Atomic parameter value
• 1502: Atomic charge
• 1503: Averaged bond distance
• 1504: Atom having “value assigned” atomic parameter
• 1505: Atom having “value not assigned” atomic parameter
• 1506: A linear function correlating atomic parameter value with value of an atomic attribute
• 1507: Atomic charge and averaged bond distance of a “value not assigned” atom
• 1508: “Value not assigned” atom having value of an atomic attribute far apart from value of an atomic attribute of a “value assigned” atom
• 1601: Atomic type
• 1602: Determination process of atomic parameter value
• 1603: Atomic type that each atom contains
• 1604: Molecule having the number of undetermined atomic parameter of 1
• 1605: Atomic parameter r_CH3
• 1606: Atomic parameter r_CH3
• 1607: Undetermined atomic parameter
• 1608: The number of undetermined atomic parameters
• 1609: Molecule having the number of undetermined atomic parameters of 1
• 1610: Four atomic parameters determined with respect to atomic type NH₂
• 1611: Atomic parameter r_NH2
• 1701: Amount of calculation
• 1702: Calculation error
• 1703: Plot of lozenge marks
• 1704: Plot of rectangular marks
• 1801: Molecule
• 1802: Experimental value of solvation energy
• 1803: Calculation error in the prior art
• 1804: Calculation error in Embodiment of this invention
• 1805: Averaged absolute error
• 1901: Input apparatus
• 19011: Input of energy convergence threshold τ
• 19012: Input of necessary information and experimental value for calculation of analysis object
• 19013: Input of elemental type and element attribute value of each element
• 1902: The initialization step
• 19021: The elemental parameter initialization step
• 19022: The sorting up step of all analysis objects by the number of undetermined elemental parameters
• 1903: Calculation order list generation step
• 19031: The analysis objects selection step
• 19032: The calculation order determination step
• 1904: The one-variable equation solving step
• 19041: The undetermined elemental parameter determination step
• 19042: Calculation order list update step
• 19043: Calculation order list termination judging step
• 1905: The elemental parameters update step
• 19051: The elemental parameter expression formula update step
• 19052: The sorting up step of undetermined analysis objects by the number of undetermined elemental parameters
• 19053: The sorting down step of analysis objects with the identical number of undetermined elemental parameters by the number of elemental parameters
• 19054: The undetermined analysis objects update step
• 1906: The elemental parameter calculation termination judging step
• 1907: The calculation error computing step
• 1908: Output apparatus
• 19081: Output of determined elemental parameters
• 2001: Calculation order list generation step
• 2002: First round process to be executed in calculation order list generation step
• 20021: The number of undetermined elemental parameters
• 20022: Analysis object having the number of undetermined elemental parameters of 1
• 20023: Calculation order list
• 2003: The one-variable equation solving step
• 2004: First round process to be executed in the one-variable equation solving step
• 2005: The elemental parameters update step
• 2006: First round process to be executed in the atomic parameters update step
• 20061: Elemental type corresponding to determined elemental parameter
• 20062: Elemental parameter of an elemental type a
• 20063: Element, wherein elemental attribute value is apart from that of element used for the parameter determination by certain threshold or more
• 20064: The number of undetermined elemental parameters
• 20065: The number of elemental parameters
• 2007: Second round process to be executed in calculation order list generation step
• 20071: The Number of undetermined elemental parameters
• 20072: Analysis object having the number of undetermined elemental parameters of 1
• 20073: Calculation order list
• 2008: Second round process to be executed in the one-variable equation solving step
• 2009: Second round process to be executed in the atomic parameters update step
• 20091: Elemental type corresponding to determined elemental parameter
• 20092: Elemental parameter of elemental type a
• 2010: “Value assigned” element
• 2011: “Value not assigned” (undetermined) element
• 2101: Input apparatus
• 21011: Input of acceptable energy error ε
• 21012: Input of energy convergence threshold τ
• 21013: Input of three-dimensional coordinate information and experimental value of salvation energy for molecule
• 21014: Input of atomic type, atomic charge and averaged bond distance for each atom
• 2102: The initialization step
• 21021: The atomic parameter initialization step
• 21022: The sorting up step of all molecules by the number of undetermined atomic parameters
• 2103: The calculation order list generation step
• 21031: The molecule selection step
• 21032: The calculation order determination step
• 2104: The one-variable equation solving step
• 21041: The judgment step whether the molecule on the top of calculation order list contains undetermined atomic parameters or not
• 21042: The undetermined atomic parameter determination step
• 21043: The calculation error computing step
• 21044: The step for comparing the calculation error with the acceptable error
• 21045: The setting afresh step of undetermined atomic parameter
• 21046: Calculation order list update step
• 21047: Calculation order list termination judging step
• 2105: The atomic parameters update step
• 21051: The atomic parameter expression formula update step
• 21052: The step for tentatively determining some or all of undetermined atomic parameters.
• 21053: The sorting up step of undetermined molecules by the number of undetermined atomic parameters
• 21054: The sorting down step of molecules with the identical number of undetermined atomic parameters by the number of atomic parameters
• 21055: The undetermined molecule update step
• 2106: The atomic parameter calculation termination judging step
• 2107: Output apparatus
• 21071: Output of value of determined atomic parameters
• 2201: The calculation order list generation step
• 2202: First round process to be executed in the calculation order list generation step
• 22021: The number of undetermined atomic parameters
• 22022: Molecule having the number of undetermined atomic parameters of 1 or less
• 22023: Calculation order list
• 2203: The one-variable equation solving step
• 2204: First round process to be executed in the one-variable equation solving step
• 2205: The atomic parameters update step
• 2206: First round process to be executed in the atomic parameters update step
• 22061: Atomic type corresponding to determined atomic parameter
• 22062: Atomic parameter of atomic type a
• 22063: Atom, wherein atomic charge is apart from that of other atom belonging to the identical atomic type by certain threshold or more
• 22064: The number of undetermined atomic parameters
• 22065: The number of atomic parameters
• 2207: Second round process to be executed in calculation order list generation step
• 22071: The number of undetermined atomic parameters
• 22072: Molecule having the number of undetermined atomic parameters of 1 or less
• 22073: Calculation order list
• 2208: Second round process to be executed in the one-variable equation solving step
• 2209: Second round process to be executed in the atomic parameters update step
• 22091: Atomic type corresponding to determined atomic parameter
• 22092: Atomic parameter of atomic type a
• 2210: “Value assigned” (determined) atom
• 2211: “Value assigned” (tentatively determined) atom
• 2212: “Value not assigned” (undetermined) atom
• 2301: Start
• 2302: The new setting judgment step of undetermined parameter
• 2303: Delete process of all calculation order list
• 2304: Delete process of molecule on the top from a calculation order list
• 2305: Calculation order list termination judging step
• 2306: Termination (proceed to the atomic parameters update step).
• 2307: Termination (proceed to the undetermined atomic parameter determination step).
• 2401: Atomic parameter value
• 2402: One value of an atomic attribute
• 2403: “Value assigned” atom
• 2404: “Value not assigned” atom
• 2405: A linear function correlating atomic parameter value with value of an atomic attribute
• 2406: Value of an atomic attribute of a “value not assigned” atom
• 2407: Atom having value of tentatively determined atomic parameter set as undetermined atomic parameter as it is
• 2408: Atom having re-determined atomic parameter, after tentatively determined atomic parameter value being set as undetermined atomic parameter
• 2409: Three regions based on value of an atomic attribute
• 2410: A linear function defined in each region
• 2501: Atomic parameter value.
• 2502: Atomic charge.
• 2503: Averaged bond distance.
• 2504: “Value assigned” atom
• 2505: “Value not assigned” atom
• 2506: A linear function correlating atomic parameter value with value of an atomic attribute
• 2507: Atomic charge and averaged bond distance of a “value not assigned” atom
• 2508: Atom, which the tentatively determined value became the determined value as it is
• 2509: Atom, which the tentatively determined value was set as undetermined atomic parameter, and atomic parameter was re-determined
• 2510: Two regions on a plane formed by atomic charge and averaged bond distance
• 2511: A linear function defined in each region
• 2601: The early calculation stage
• 2602: The last calculation stage
• 2603: The amount of calculation for molecule n in the prior art
• 2604: The amount of calculation for molecule n in this invention
• 2605: The amount of calculation required for calculation of solvent energy in one time
• 2701: Averaged number of iteration
• 2702: The amount of calculation in using Embodiment of this invention
• 2703: The lower bound of the amount of calculation using a gradient-matrix
• 2801: Input apparatus
• 28011: Input of acceptable error ε
• 28012: Input of convergence threshold τ
• 28013: Input of necessary information and experimental value for calculation of analysis object
• 28014: Input of elemental type and value of element attribute for each element
• 2802: The initialization step
• 28021: The elemental parameter initialization step
• 28022: The sorting up step of all analysis objects by the number of undetermined elemental parameters
• 2803: Calculation order list generation step
• 28031: The analysis objects selection step
• 28032: The calculation order determination step
• 2804: The one-variable equation solving step
• 28041: The judgment step of whether an analysis object on the top of a calculation order list contains undetermined elemental parameter or not
• 28042: The undetermined elemental parameter determination step
• 28043: The calculation error computing step
• 28044: The step for comparing the calculation error with the acceptable error
• 28045: The setting afresh step of undetermined elemental parameter
• 28046: Calculation order list update step
• 28047: Calculation order list termination judging step
• 2805: The elemental parameters update step
• 28051: The elemental parameter expression formula update step
• 28052: The tentatively determining step of some or all of values of elemental parameters of undetermined element
• 28053: The sorting up step of undetermined analysis objects by number of undetermined elemental parameters
• 28054: The sorting down step of analysis objects with the identical number of undetermined elemental parameters by number of elemental parameters
• 28055: The undetermined analysis object update step
• 2806: The elemental parameter calculation termination judging step
• 2807: Output apparatus
• 28071: Output of determined elemental parameters
• 2901: Calculation order list generation step
• 2902: First round process to be executed in calculation order list generation step
• 29021: The number of undetermined elemental parameters
• 29022: Analysis object having the number of undetermined elemental parameters of 1 or less
• 29023: Calculation order list
• 2903: The one-variable equation solving step
• 2904: First round process to be executed in the one-variable equation solving step
• 2905: The elemental parameters update step
• 2906: First round process to be executed in the elemental parameters update step
• 29061: Elemental type corresponding to determined elemental parameter
• 29062: Elemental parameter of the elemental type a
• 29063: Element, wherein elemental attribute value is apart from that of other element belonging to the identical elemental type by certain threshold or more
• 29064: The number of undetermined elemental parameters
• 29065: The number of elemental parameters
• 2907: Second round process to be executed in calculation order list generation step
• 29071: The number of undetermined elemental parameters
• 29072: Analysis object having the number of undetermined elemental parameters of 1 or less
• 29073: Calculation order list
• 2908: Second round process to be executed in the one-variable equation solving step
• 2909: Second round process to be executed in the atomic parameters update step
• 29091: Elemental type corresponding to determined elemental parameter
• 29092: Elemental parameter of an elemental type a
• 2910: “Value assigned” (determined) element
• 2911: “Value assigned” (tentatively assigned) element
• 2912: “Value not assigned” (undetermined) element
• 3001: The input step
• 3002: The initialization step
• 3003: The calculation order list generation step
• 30031: Inner process
• 30032: The calculation order list display step
• 30033: The calculation progress display step
• 3004: The equation solving step
• 3005: The atomic parameters update step
• 30051: Inner process
• 30052: The atomic parameter display step
• 3006: The termination judging step
• 3007: The output step
• 3101: Display object displaying tab
• 31011: Calculation order list tab
• 31021: Calculation order list
• 31022: Up-arrow button
• 31023: Down-arrow button
• 31024: Delete button
• 31025: Confirm button
• 3102: Calculation order display region
• 3103: Distribution diagram display region
• 3104: Molecular structure map display region
• 3201: Display object displaying tab
• 32011: Progress tab
• 3202: Display region of summary of the atomic parameter calculation progress
• 3203: Display region of the atomic parameter calculation progress for each molecule
• 3301: Display object displaying tab
• 33011: Atomic parameter tab
• 3302: Display region of the atomic parameter determination status
• 3303: Display region of atomic parameter value

It should be further understood by those skilled in the art that although the foregoing description has been made on embodiments of the invention, the invention is not limited thereto and various changes and modifications may be made without departing from the spirit of the invention and the scope of the appended claims.

Claims

1. A simulation parameter determination method, which is a method for elemental parameter determination necessary to reproduce, by simulation calculation, experimental value of characteristic value or property value measured on analysis objects of any of a plurality kinds of molecules, materials, parts or structures, each analysis object being composed of a plurality of one or a plurality of elements, each element being classified into one of a plurality kinds of elemental type s, or each element having one or a plurality of values of elemental attributes, wherein (1) in the input step, (1-1) acceptable error ε that is maximum value acceptable to difference between calculated value obtained by simulation, and experimental value; (1-2) convergence threshold τ that is standard value to judge convergence of calculated value; (1-3) information necessary for calculation of each analysis object, and experimental value for each analysis object; and (1-4) elemental type for each element composing the analysis object, and value of elemental attribute of each element composing the analysis object are input into a memory unit, (2) in the pre-processing step, (2-1) value of elemental parameter by each elemental type is initialized; and (2-2) a data set containing undetermined elemental parameter number on all of the analysis objects is present, and arrangement order of the data set is sorted up by the number of undetermined atomic parameters, (3) in calculation order list generation step, (3-1) an analysis object, having the number of undetermined elemental parameters of 1 or less, is selected from analysis objects containing undetermined elemental parameters; and (3-2) a calculation order list is generated based on re-arranged order at the above (2-2), of the selected analysis objects, as calculation order, (4) in the solving step of a one-variable equation, (4-1) only in the case where an analysis object on the top of calculation order list contains undetermined elemental parameter, solving of a one-variable equation for determination of undetermined parameters is iterated until the amount of change of calculated value becomes below convergence threshold, to determine undetermined elemental parameter, and in the case where an analysis object on the top of calculation order list does not contain undetermined elemental parameter, process is forwarded to the following step (4-2); (4-2) calculation error, that is difference between calculated value of said analysis object using determined elemental parameter and experimental value, is calculated; (4-3) absolute value of said calculation error, and said acceptable error are compared; (4-4) in the case where said calculation error is over the acceptable error, one of the tentatively determined elemental parameters is set newly as an undetermined elemental parameter to return to the step (4-1), and in the case where said calculation error is not over the acceptable error, process is forwarded to the following step (4-5); and (4-5) in the case where said calculation error is equal to or below the acceptable error, a data set on the top position of calculation order list is deleted; in the case where an analysis object is present on the top of calculation order list, process returns to the step (4-1); and in the case where calculation order list is empty, process is forwarded to the step (5), (5) in the elemental parameters update step, (5-1) elemental parameter of corresponding elemental type is updated, as a function of value of elemental attribute, using an elemental parameter determined in the above one-variable equation solving step; (5-2) some or all undetermined elemental parameters, which is classified to said elemental type, are tentatively determined, using a function describing value of elemental parameter of the updated elemental type; (5-3) a plurality of data sets of analysis objects including undetermined elemental parameters are sorted up by the number of undetermined atomic parameters; (5-4) subsequent to the above (5-3), analysis objects with the same number of undetermined atomic parameters are sorted down by the number of elemental parameters; and (5-5) whether all of or at least some of each of parameters of the data sets including undetermined elemental parameters is determined or not, and (6) whether undetermined analysis objects are present or not, is judged, and in the case where undetermined analysis objects are present, process returns to the above step (3), and in the case where undetermined analysis objects are not present, process terminates.

2. A simulation parameter determination method, which is a method for elemental parameter determination necessary to reproduce, by simulation calculation, experimental value of characteristic value or property value measured on analysis objects of any of a plurality kinds of molecules, materials, parts or structures, each analysis object being composed of a plurality of one or a plurality of elements, each element being classified into one of a plurality kinds of elemental type s, or each element having one or a plurality of values of elemental attributes, wherein (1) in the input step, (1-1) convergence threshold τ that is standard value to judge convergence of calculated value; (1-2) three-dimensional coordinate information of elements composing the analysis object, necessary for calculation of experimental value for each analysis object, or experimental value for each analysis object; and (1-3) elemental type for each element composing the analysis object, and value of elemental attribute of each element composing the analysis object are input into a memory unit, (2) in the pre-processing step, (2-1) value of elemental parameter by each elemental type is initialized; and (2-2) a data set containing undetermined elemental parameter number on all of the analysis objects is present, and arrangement order of the data set is sorted up by the number of undetermined atomic parameters, (3) in calculation order list generation step, (3-1) an analysis object, having the number of undetermined elemental parameters of 1, is selected from analysis objects containing undetermined elemental parameters; and (3-2) a calculation order list is generated based on re-arranged order at the above (2-2), of the selected analysis objects, as calculation order, (4) in the solving step of a one-variable equation, (4-1) on an analysis object on the top of calculation order list, solving of a one-variable equation for determination of undetermined parameters is iterated until the amount of change of calculated value becomes below convergence threshold, to determine undetermined elemental parameter; and (4-2) a data set on the top position of calculation order list is deleted; and in the case where an analysis object is present on the top of calculation order list, process returns to the step (4-1); and in the case where calculation order list is empty, process is forwarded to the step (5), (5) in the elemental parameters update step, (5-1) elemental parameter of corresponding elemental type is updated, as a function of value of elemental attribute, using an elemental parameter determined in the above one-variable equation solving step; (5-2) a plurality of data sets of analysis objects including undetermined elemental parameters are sorted up by the number of undetermined elemental parameters; (5-3) subsequent to the above (5-2), analysis objects with the same number of undetermined elemental parameters are sorted down by the number of elemental parameters; and (5-4) whether all of or al least some of each of parameters of the data sets of analysis objects including undetermined elemental parameters is determined or not, and (6) whether undetermined analysis objects are present or not, is judged, and in the case where undetermined analysis objects are present, process returns to the above step (3), and in the case where undetermined analysis objects are not present, process terminates.

3. A simulation parameter determination method, which is a method for atomic parameter determination necessary to reproduce, by simulation calculation, experimental value of each solvation energy measured on a plurality kinds of molecules, each molecule being composed of a plurality of atoms, each atom being classified into one of a plurality kinds of atomic types, or each atom having one or a plurality of values of atomic attributes, wherein (1) in the input step, (1-1) acceptable energy error e that is maximum value acceptable to difference between calculated value of salvation energy obtained by simulation calculation, and experimental value of solvation energy; (1-2) energy convergence threshold τ that is standard value to judge convergence of calculated value of solvation energy; (1-3) three-dimensional coordinate information of atoms composing the molecule, necessary for calculation of experimental value salvation energy for each molecule, or experimental value of solvation energy for each molecule; and (1-4) atomic type for each atom composing the molecule, and value of an atomic attribute of each atom composing the molecule are input into a memory unit, (2) in the pre-processing step, (2-1) atomic parameter value by each atomic type is initialized; and (2-2) a data set containing undetermined atomic parameter number on all of the molecules is present, and arrangement order of the data set is sorted up by the number of undetermined atomic parameters, (3) in calculation order list generation step, (3-1) a molecule, having the number of undetermined atomic parameters of 1 or less, is selected from molecules containing undetermined atomic parameters; and (3-2) a calculation order list is generated based on re-arranged order at the above (2-2), of the selected molecules, as calculation order, (4) in the solving step of a one-variable equation, (4-1) only in the case where an molecule on the top of calculation order list contains undetermined atomic parameter, solving of a one-variable equation for determination of undetermined parameters is iterated until the amount of change of calculated value of solvation energy becomes below energy convergence threshold, to determine undetermined parameter, and in the case where a molecule on the top of calculation order list does not contain undetermined atomic parameter, process is forwarded to the following step (4-2); (4-2) calculation error, that is difference between calculated value of said molecule using determined atomic parameter and experimental value, is calculated; (4-3) absolute value of said calculation error, and said acceptable energy error are compared; (4-4) in the case where said energy calculation error is over the acceptable energy error, one of the tentatively determined atomic parameters is set newly as an undetermined atomic parameter to return to the step (4-1), and in the case where said energy calculation error is not over the acceptable energy error, process is forwarded to the following step (4-5); and (4-5) in the case where said calculation error is equal to or below the acceptable error, a data set on the top position of calculation order list is deleted; in the case where a molecule is present on the top of calculation order list, process returns to the step (4-1); and in the case where calculation order list is empty, process is forwarded to the step (5), (5) in the atomic parameters update step, (5-1) atomic parameter of corresponding atomic type is updated, as a function of value of an atomic attribute, using an atomic parameter determined in the above one-variable equation solving step; (5-2) some or all of undetermined atomic parameters, which is classified to said atomic type, are tentatively determined, using a function describing atomic parameter value of the updated atomic type; (5-3) a plurality of data sets of molecules including undetermined atomic parameters are sorted up by the number of undetermined atomic parameters; (5-4) subsequent to the above (5-3), molecules with the same number of undetermined atomic parameters are sorted down by the number of atomic parameters; and (5-5) whether all of or al least some of each of parameters of the data sets of molecules including undetermined atomic parameters is determined or not, and (6) whether undetermined molecules are present or not, is judged, and in the case where undetermined molecules are present, process returns to the above step (3), and in the case where undetermined molecules are not present, process terminates.

4. A simulation parameter determination method, which is a method for atomic parameter determination necessary to reproduce, by simulation calculation, experimental value of each solvation energy measured on a plurality kinds of molecules, each molecule being composed of a plurality of atoms, each atom being classified into one of a plurality kinds of atomic types, or each atom having one or a plurality of values of atomic attributes, wherein (1) in the input step, (1-1) energy convergence threshold τ that is standard value to judge convergence of calculated value of solvation energy; (1-2) three-dimensional coordinate information of atoms composing the molecule, necessary for calculation of experimental value salvation energy for each molecule, or experimental value of solvation energy for each molecule; and (1-3) atomic type for each atom composing the molecule, and value of an atomic attribute of each atom composing the molecule are input into a memory unit, (2) in the pre-processing step, (2-1) atomic parameter value by each atomic type is initialized; and (2-2) a data set containing undetermined atomic parameter number on all of the molecules is present, and arrangement order of the data set is sorted up by the number of undetermined atomic parameters, (3) in calculation order list generation step, (3-1) a molecule, having the number of undetermined atomic parameters of 1, is selected from molecules containing undetermined atomic parameters; and (3-2) a calculation order list is generated based on re-arranged order at the above (2-2), of the selected molecules, as calculation order, (4) in the solving step of a one-variable equation, (4-1) solving of a one-variable equation for determination of undetermined parameters is iterated until the amount of change of calculated value of salvation energy becomes below convergence threshold, to determine undetermined parameter, on a molecule on the top of calculation order list; and (4-2) a data set on the top position of calculation order list is deleted; in the case where a molecule is present on the top of calculation order list, process returns to the step (4-1); and in the case where calculation order list is empty, process is forwarded to the step (5), (5) in the atomic parameters update step, (5-1) atomic parameter of corresponding atomic type is updated, as a function of value of an atomic attribute, using an atomic parameter determined in the above one-variable equation solving step; (5-2) a plurality of data sets of molecules including undetermined atomic parameters are sorted up by the number of undetermined atomic parameters; (5-3) subsequent to the above (5-2), molecules with the same number of undetermined atomic parameters are sorted down by the number of atomic parameters; and (5-4) whether all of or al least some of each of parameters of the data sets of molecules including undetermined atomic parameters is determined or not, and (6) whether undetermined molecules are present or not, is judged, and in the case where undetermined molecules are present, process returns to the above step (3), and in the case where undetermined molecules are not present, process terminates.

5. The simulation parameter determination method according to claim 3, wherein user interface is provided with function that, in said calculation order list generation step, when calculation order is displayed, and one molecule in said calculation order is selected, an atomic distribution map at a plane formed by atomic charge and averaged bond distance, and a molecular structure map that expresses the atomic parameter determination status of said selected molecule, by a three-dimensional structure of the molecular, and spheres drawn on each atom composing the molecule, in an overlapped way, are displayed, as for an atomic type to which undetermined atomic parameter of said selected molecule is classified, and a user can change calculation order based on these sets of display information.

6. The simulation parameter determination method according to claim 4, wherein user interface is provided with function that, in said calculation order list generation step, when calculation order is displayed, and one molecule in said calculation order is selected, an atomic distribution map at a plane formed by atomic charge and averaged bond distance, and a molecular structure map that expresses the atomic parameter determination status of said selected molecule, by a three-dimensional structure of the molecular, and spheres drawn on each atom composing the molecule, in an overlapped way, are displayed, as for an atomic type to which undetermined atomic parameter of said selected molecule is classified, and a user can change calculation order based on these sets of display information.

7. The simulation parameter determination method according to claim 3, wherein user interface is provided with function that, in said calculation order list generation step, after generation of calculation order list, or after a user executes re-arrangement or deletion of molecules in calculation order list, summary result of a calculation status of the molecule, such as “value assigned”, “in progress”, or “undetermined”, and calculation elapsed time are displayed, as summary of a progress of atomic parameter calculation, or a calculation status, the number of atomic parameters and the number of undetermined atomic parameters of each atom are displayed, as summary of a progress of atomic parameter of each atom.

8. The simulation parameter determination method according to claim 4, wherein user interface is provided with function that, in said calculation order list generation step, after generation of calculation order list, or after a user executes re-arrangement or deletion of molecules in calculation order list, summary result of a calculation status of the molecule, such as “value assigned”, “in progress”, or “undetermined”, and calculation elapsed time are displayed, as summary of a progress of atomic parameter calculation, or a calculation status, the number of atomic parameters and the number of undetermined atomic parameters of each atom are displayed, as summary of a progress of atomic parameter of each atom.

9. The simulation parameter determination method according to claim 3, wherein user interface is provided with function that, in said atomic parameters update step, the number of all atoms, the number of atoms having determined atomic parameter, the number of tentatively determined atoms and the number of undetermined atoms are displayed by each atomic type, as the atomic parameter determination status of each atomic type, and when one atomic type is selected, an atomic parameter of an atom belonging to said atomic type is displayed in space formed by atomic charge, averaged bond distance and atomic parameter.

10. The simulation parameter determination method according to claim 4, wherein user interface is provided with function that, in said atomic parameters update step, the number of all atoms, the number of atoms having determined atomic parameter, the number of tentatively determined atoms and the number of undetermined atoms are displayed by each atomic type, as the atomic parameter determination status of each atomic type, and when one atomic type is selected, an atomic parameter of an atom belonging to said atomic type is displayed in space formed by atomic charge, averaged bond distance and atomic parameter.