Synthesis condition generation method, synthesis condition generation device, and program

Abstract

A synthesis condition generation method according to one embodiment of the present invention generates a synthesis condition in a trial run in which generation of a substance that conforms to the synthesis condition, measurement of an evaluation value for evaluating the substance, and generation of the synthesis condition are executed in the stated order. In the synthesis condition generation method, a computer executes: an evaluation value conversion procedure in which, when there exists a trial run n' in which the evaluation value yn' (n'∈{1, …, N}) of a substance could not be measured among the past N rounds of trial runs, an evaluation value yn' in the trial run n' is replaced with the worst evaluation value; a prediction procedure in which the evaluation value y of an untried synthesis condition x is predicted using a synthesis condition xn and an evaluation value yn in a trial run (n=1, …, N); and a generation procedure in which a synthesis condition that yields a better evaluation value than the best value of an evaluation value yn (n=1, …, N) is generated, as a synthesis condition xN+1 in a trial run N+1, using the result of having predicted the evaluation value y of the synthesis condition x of the untried trial run.

Description

[Technical field]

[0001] The present invention relates to a synthesis condition generating method, a synthesis condition generating device, and a program. [Background technology]

[0002] Conventionally, synthesis conditions that give good evaluation values ​​are searched for by repeating trials to measure the evaluation values ​​of materials produced according to given synthesis conditions (recipe). For example, as shown in Figure 1, (1) synthesis conditions x n (2) generate this synthesis condition x n (3) generating a sample (substance) according to the above, and evaluating the sample to obtain an evaluation value y n By repeating this process, synthesis conditions that give good evaluation values ​​are searched for.

[0003] As a method for generating a thin film (material, sample) according to given synthesis conditions, for example, there is a molecular beam epitaxy method (Non-Patent Document 1). In addition, there are thin film synthesis and bulk synthesis methods such as a sputtering method, a pulsed laser ablation method, a physical vapor deposition method, a chemical vapor deposition method, a spin coating method, an atomic layer deposition method, a floating zone method, a flux method, a Czochralski method, etc. In addition, as a method for generating synthesis conditions to be tried next under a problem setting in which an evaluation value for the synthesis conditions is always obtained, there is a Bayesian optimization method (Patent Document 1, Non-Patent Documents 2 and 3). Note that since the process of generating a material according to the synthesis conditions and measuring the evaluation value of the material with an evaluation index designed in advance according to the use of the material is complicated, there may be a situation in which the relationship between the synthesis conditions and the evaluation value cannot be explicitly described, but Bayesian optimization is known as a method that can search for synthesis conditions even in such a situation. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2018-147075 A [Non-patent literature]

[0005] [Non-Patent Document 1] Y. Krockenberger et al. Emerging superconductivity hidden beneath chargetransfer insulators, Scientific Reports 3:2235, 2013. [Non-Patent Document 2] J. Snoek, H. Larochelle, and RP Adams: Practical Bayesian Optimization of Machine Learning Algorithms, Advances in Neural Information Processing Systems 25, pp. 2951-2959, 2012. [Non-Patent Document 3] YK Wakabayashi et al. Machine-learning-assisted thin-film growth: Bayesian optimization in molecular beam epitaxy of SrRuO3 thin films, APL Materials 7, 101114, 2019. Summary of the Invention [Problem to be solved by the invention]

[0006] However, the conventional method for searching for synthesis conditions that give good evaluation values ​​has a problem that there may be synthesis conditions for which no evaluation value can be obtained. This is because stable substances are not necessarily produced under all synthesis conditions within the search range of synthesis conditions. In addition, it is possible that a substance with a different composition or crystal structure from the target substance is produced under certain synthesis conditions.

[0007] One embodiment of the present invention has been made in view of the above points, and has an object to generate synthesis conditions that allow an optimal substance to be obtained under given evaluation indexes. [Means for solving the problem]

[0008] In order to achieve the above object, a synthesis condition generation method according to one embodiment is a synthesis condition generation method for generating synthesis conditions in a trial in which a substance is generated according to synthesis conditions, an evaluation value for evaluating the substance is measured, and the synthesis conditions are generated in sequence, the synthesis condition generation method being a synthesis condition generation method for generating synthesis conditions in a trial in which a substance evaluation value y n' If there is a trial n' where (n'∈{1, ,N}) could not be measured, the evaluation value y n' with the worst evaluation value, and the synthesis condition x in trial n (n = 1, , N) n and the evaluation value y n A prediction procedure for predicting an evaluation value y of an untried synthesis condition x using the above, and a prediction procedure for predicting an evaluation value y of the untried synthesis condition x using the result of predicting the evaluation value y of the untried synthesis condition x. n The composite condition that has a better evaluation value than the best value of (n=1, , N) is the composite condition x N+1 and a generating procedure for generating the following: Effect of the Invention

[0009] It is possible to generate synthesis conditions that will yield an optimal substance under given evaluation criteria. [Brief description of the drawings]

[0010] [Figure 1] FIG. 13 is a diagram showing an example of a trial for obtaining a synthesis condition with a good evaluation value. [Diagram 2] 1 is a diagram illustrating an example of an overall configuration of a synthesis condition generation system according to an embodiment of the present invention. [Diagram 3] FIG. 2 is a diagram illustrating an example of a hardware configuration of a synthesis condition generating device according to the present embodiment. [Figure 4] FIG. 2 is a diagram illustrating an example of a functional configuration of a synthesis condition generating device according to the present embodiment. [Diagram 5] 11 is a flowchart illustrating an example of a synthesis condition generation process according to the embodiment. [Figure 6] FIG. 13 is a diagram showing evaluation values ​​in a simulation experiment. [Figure 7] FIG. 13 is a diagram showing the results of a simulation experiment. [Figure 8] FIG. 1 shows the results of an SrRuO3 experiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] An embodiment of the present invention will be described below. In the following embodiment, a synthesis condition generating system 1 capable of generating synthesis conditions for obtaining an optimal substance under a given evaluation index will be described.

[0012] <Overall configuration of synthesis condition generation system 1> An example of the overall configuration of the synthesis condition generating system 1 according to this embodiment is shown in Fig. 2. As shown in Fig. 2, the synthesis condition generating system 1 according to this embodiment includes a synthesis condition generating device 10, a material generating device 20, and a material evaluating device 30.

[0013] The synthesis condition generation device 10 generates synthesis conditions to be used in the current trial using the synthesis conditions (recipe) obtained up to the previous trial and their evaluation values, with n being an index indicating the number of trials. Synthesis conditions are conditions for obtaining a certain substance by proceeding with the reaction of one or more raw materials by a certain method (e.g., the amount of constituent elements, reaction temperature, timing of inputting raw materials, etc.), and are generally expressed as vectors. As an example, in a task for producing a SrRuO3 thin film, the synthesis conditions are expressed as a vector having three elements: (a) the amount of Ru supplied, (b) the temperature of the substrate to which the thin film is attached, and (c) the distance between the substrate and the ozone nozzle that supplies oxygen atoms O. Hereinafter, the nth trial is also referred to as "trial n", and the synthesis conditions used to generate a substance in trial n are referred to as "x n "

[0014] The material generator 20 generates a material (sample) by promoting a reaction of the raw materials according to the synthesis conditions generated by the synthesis condition generator 10. The material generator 20 is an apparatus for generating a material by a thin film synthesis / bulk synthesis method such as molecular beam epitaxy, sputtering, pulsed laser ablation, physical vapor deposition, chemical vapor deposition, spin coating, atomic layer deposition, floating zone deposition, flux method, Czochralski method, etc. As an example, in a task of generating a SrRuO3 thin film, a thin film generator using a molecular beam epitaxy method can be used as the material generator 20.

[0015] The material evaluation device 30 measures an evaluation value obtained by evaluating the material (sample) generated by the material generation device 20 using a predetermined evaluation index. As the evaluation index, an existing evaluation index according to the desired material and its shape (thin film, bulk, etc.) may be used. As an example, in a task of generating a SrRuO3 thin film, the residual resistivity ratio (RRR) may be used as an evaluation index for measuring the quality of the arrangement of elements in the thin film.

[0016] Hereinafter, the evaluation value will be defined as consisting of binary information indicating whether or not a substance (sample) is a stable substance or a desired substance, and a scalar value indicating the quality of that substance, and will take a scalar value if the substance is stable or the desired substance, and NaN if the substance is neither stable nor the desired substance. For example, if one wishes to obtain an insulator with high electrical resistance, the evaluation value will be the electrical resistance value if the desired sample is obtained, and NaN if not. Note that NaN stands for Not a Number.

[0017] In addition, the synthesis condition x n The evaluation value for the material generated by the material generating device 20 is expressed as "y n ", and the larger the evaluation value, the better the evaluation. Note that, in the case where the smaller the evaluation value, the better the evaluation, it is possible to similarly apply this embodiment by, for example, inverting the sign of the evaluation value.

[0018] Here, the synthesis condition generation system 1 according to the present embodiment calculates synthesis conditions obtained in trials so far and their evaluation values ​​{(x n ,y n )|n=1, . . . , N} (where N is the number of trials so far), the synthesis condition generation device 10 generates the synthesis condition x N+1 Generate this composite condition x N+1 The material generating device 20 generates a material (sample) according to the above, and the material evaluating device 30 evaluates the sample to obtain an evaluation value y N+1 The synthesis conditions that give good evaluation values ​​are searched for by repeating the above trials. In the following, the material production device 20 and the material evaluation device 30 are treated as black boxes, and the synthesis conditions obtained in the trials so far and their evaluation values ​​{(x n ,y n )|n=1,...,N}, the synthetic condition x that gives the better evaluation value in the N+1th trial N+1 A case will be described in which the synthesis condition generating device 10 generates the above. By repeating this trial, it becomes possible to generate synthesis conditions that will yield an optimal substance.

[0019] However, the overall configuration of the synthesis condition generation system 1 shown in Fig. 2 is an example, and is not limited thereto. For example, the synthesis condition generation device 10 and two or more devices among the material generation device 20 and the material evaluation device 30 may be realized by the same device.

[0020] <Hardware Configuration of the Synthesis Condition Generation Device 10> An example of the hardware configuration of the synthesis condition generating device 10 according to this embodiment is shown in Fig. 3. As shown in Fig. 3, the synthesis condition generating device 10 according to this embodiment has an input device 101, a display device 102, an external I / F 103, a communication I / F 104, a RAM (Random Access Memory) 105, a ROM (Read Only Memory) 106, an auxiliary storage device 107, and a processor 108. Each of these pieces of hardware is connected to each other via a bus 109 so as to be able to communicate with each other.

[0021] The input device 101 is, for example, a keyboard, a mouse, a touch panel, etc. The display device 102 is, for example, a display, a display panel, etc. Note that the synthesis condition generation device 10 does not necessarily have to have at least one of the input device 101 and the display device 102, for example.

[0022] The external I / F 103 is an interface with an external device such as a recording medium 103a. The synthesis condition generating device 10 can read and write data from and to the recording medium 103a via the external I / F 103. Examples of the recording medium 103a include a flexible disk, a CD (Compact Disc), a DVD (Digital Versatile Disk), a SD memory card (Secure Digital memory card), and a USB (Universal Serial Bus) memory card.

[0023] The communication I / F 104 is an interface for connecting the synthesis condition generation device 10 to a communication network. The RAM 105 is a volatile semiconductor memory (storage device) that temporarily stores programs and data. The ROM 106 is a non-volatile semiconductor memory (storage device) that can store programs and data even when the power is turned off. The auxiliary storage device 107 is a storage device (storage device) such as a hard disk drive (HDD) or a solid state drive (SSD). The processor 108 is an arithmetic device such as a central processing unit (CPU) or a graphics processing unit (GPU).

[0024] The synthesis condition generating device 10 according to the present embodiment can realize the synthesis condition generating process described below by having the hardware configuration shown in Fig. 3. Note that the hardware configuration shown in Fig. 3 is an example, and the hardware configuration of the synthesis condition generating device 10 is not limited thereto. For example, the synthesis condition generating device 10 may have a plurality of auxiliary storage devices 107 and a plurality of processors 108, or may have various hardware other than the hardware shown in the figure.

[0025] <Functional configuration of the synthesis condition generating device 10> FIG. 4 shows a functional configuration example of the synthesis condition generating device 10 according to the present embodiment. As shown in FIG. 4, the synthesis condition generating device 10 according to the present embodiment has an evaluation value input unit 201, an evaluation value prediction unit 202, a synthesis condition search unit 203, and a synthesis condition output unit 204. Each of these units is realized, for example, by a process in which one or more programs installed in the synthesis condition generating device 10 are executed by the processor 108 or the like. In addition, the synthesis condition generating device 10 according to the present embodiment has a storage unit 205. The storage unit 205 is realized, for example, by the auxiliary storage device 107 or the like. Note that the storage unit 205 may be realized, for example, by a storage device such as a database server connected to the synthesis condition generating device 10 via a communication network.

[0026] The evaluation value input unit 201 inputs the evaluation value y evaluated by the material evaluation device 30 in trial N (the previous trial). N Note that this evaluation value y N is stored in the storage unit 205.

[0027] Here, each evaluation value y n Since (n=1, ,N) can be NaN, y n Conversion processing is required when =NaN. Also, the composite condition x n The difference in scale of each element that makes up the evaluation value y n Depending on the scale of the evaluation value, the prediction accuracy of the evaluation value prediction unit 202 may deteriorate. Therefore, in order to ensure the prediction accuracy and to efficiently search for the synthesis condition, the synthesis condition x n and its evaluation value y n Therefore, the evaluation value input unit 201 normalizes the evaluation value y N When you input, for each n=1,...,N, the following conversion process is used to obtain the evaluation value y n y n ' and the composite condition x n andy n ' are normalized respectively.

[0028] Conversion process The evaluation value input unit 201 inputs each evaluation value y n (n=1, ,N) is converted using the following formula (1).

[0029]

number

[0030]

number

[0031]

number

[0032] Normalization process Below, the synthesis condition x n Let x be a vector consisting of D elements. n =[x n,1 ,x n,D ] Τ Here, T represents the transpose of a vector or matrix.

[0033] D elements x n,d (d=1, ,D) represents the conditions for obtaining a certain material by a certain method. For example, in the task of producing a SrRuO3 thin film, x n,1 is the supply of Ru, x n,2 is the temperature of the substrate on which the thin film is deposited, x n,3 and represent the distance between the ozone nozzle supplying oxygen atoms O and the substrate, respectively.

[0034] dth element x n,d The lower and upper bounds of the search range are

[0035]

number

[0036]

number

[0037] At this time, the evaluation value input unit 201 inputs the synthesis condition x n =[x n,1 ,x n,D ] Τ The normalized synthesis condition

[0038]

number

[0039]

number

[0040]

number

[0041]

number

[0042]

number

[0043] In the following description, the synthesis condition x n The normalized synthesis condition is " ~ x n ", the evaluation value after conversion process y n The normalized evaluation value obtained by normalizing ' is ~ y n In addition, when there is no risk of misunderstanding, the normalized composition condition is also simply called the "composition condition" and the normalized evaluation value is also simply called the "evaluation value". Furthermore, the variable representing the normalized composition condition is written as ~ x" and the variable representing the normalized evaluation value is " ~ y".

[0044] The evaluation value prediction unit 202 calculates the normalized synthesis condition and the normalized evaluation value D N :={( ~ x n , ~ y n ')|n=1,...,N}, synthesis conditions that have not been tried before ~ The evaluation value of x ~ Predict y, where: ~ y is the average μ( ~ x), variance σ 2 ( ~ x) is a normally distributed random variable

[0045]

number

[0046] In a Gaussian process, ~ y1',..., ~ yN The joint distribution of ' is given by the following covariance matrix K N We model the distribution as a normal distribution with

[0047]

number

[0048] If the kernel function k is the RBF kernel, then ~ x m and ~ x n When is a D-dimensional vector, for these, the kernel function k is expressed as follows:

[0049]

number

[0050] It should be noted that the kernel function k may be any other kernel function than the RBF kernel. For example, the synthesis condition ~In cases where the evaluation value for x changes sharply, the Matern kernel described in Non-Patent Document 2 may be adopted. In addition to this, various other kernel functions can be adopted as necessary.

[0051] At this time, the evaluation value prediction unit 202 uses the above kernel function k(·,·) and D N Using the above, μ( ~ x), σ 2 ( ~ Calculate x).

[0052]

number

[0053] The synthesis condition search unit 203 uses the distribution of the evaluation values ​​predicted by the formula (4) to find a better evaluation value. ~ Composite condition that gives y ~ Search for x. Here, according to equation (9), a certain synthesis condition ~ Given x, the corresponding μ( ~ x), σ 2 ( ~ x) is obtained. Then, a certain synthesis condition ~ The best evaluation value of x so far as a measure of its goodness

[0054]

number

[0055]

number

[0056] In the example shown in equation (10), the expected value of improvement in the evaluation value (Reference 1) is used, but other criteria can also be used to improve the efficiency of the search (References 2 and 3).

[0057] This allows for good synthesis conditions. ~ The problem of searching for x is to find a synthesis condition that makes Eq. (10) larger. ~ Problem of searching for x

[0058]

number

[0059] Here, the synthesis conditions are ~ When the dimension of the search space for x is large, the number of points for evaluating equation (10) increases as an exponential function of the dimension. Therefore, in order to reduce the number of times equation (10) is evaluated, equation (10) is first evaluated on a coarse grid, and then an area having the top few points with the highest expected value of improvement at its vertices is selected, and the grid resolution is increased within that area and evaluation is performed, thereby approximately maximizing equation (11). In other words, first the search space is searched and evaluated on a grid with coarse resolution, and then the area having the top few points with the highest evaluation results at its vertices is searched and evaluated on a grid with finer resolution. This results in a synthesis condition that approximately maximizes the expectation value of Eq. (10). ~ x N+1 is obtained.

[0060] The synthesis condition output unit 204 outputs the synthesis condition searched by the synthesis condition search unit 203. ~ x N+1 For each element of x, the denormalization process (reverting the normalization process) shown in the following formula (12) is performed. N+1 =[x N+1,1 ,x N+1,D ] Τ Output.

[0061]

number

[0062] The storage unit 205 stores the synthesis conditions obtained in the past trials and their evaluation values ​​{(x n ,y n )|n=1, . . . , N}. The storage unit 205 also stores the normalized synthesis conditions and the normalized evaluation values ​​{( ~ x n , ~ y n ')|n=1,···,N}.

[0063] <Synthetic condition generation process> Next, the synthesis condition generation process according to this embodiment will be described with reference to FIG. 5. In the following, the synthesis condition x that gives a better evaluation value in the (N+1)th trial after N trials have been performed is N+1 The case where the above formula is generated will be described.

[0064] First, the evaluation value input unit 201 inputs the evaluation value y N (Step S101). This evaluation value y N is the synthesis condition x in the previous trial N N Corresponding to (x N ,y N ) in the storage unit 205. Note that the synthesis condition x N Note that, , is already stored in the storage unit 205.

[0065] Next, the evaluation value input unit 201 inputs the evaluation value y n (n=1, ,N) is converted into y by the conversion process of equation (1). n ' respectively (step S102).

[0066] Next, the evaluation value input unit 201 inputs the synthesis condition x n (n=1, ,N) and the evaluation value y after the conversion process n '(n=1, . . . , N) are each normalized by a normalization process (step S103).

[0067] That is, the evaluation value input unit 201 calculates the synthesis condition x n Normalize the normalized composition condition ~ x n and calculate the evaluation value y after conversion processing by using formula (3). n Normalized evaluation value obtained by normalizing ' ~ y n ' is generated. These normalized synthesis conditions and normalized evaluation values ​​( ~ x n , ~ y n ') is stored in the storage unit 205.

[0068] The following steps S104 to S107 are carried out to find untried synthesis conditions in the search space. ~ x. As described above, it is preferable to first repeat the following steps S104 to S107 with a rough grid in the search space as an untried synthesis condition, and then select an area having a few points with the highest expected improvement values ​​as vertices, increase the resolution of the grid in that area, and repeat the following steps S104 to S107 with those grids as an untried synthesis condition. Steps S104 to S107 in a certain repetition are described below.

[0069] The evaluation value prediction unit 202 selects untried synthesis conditions from within the search space. ~ Select x (step S104).

[0070] Next, the evaluation value prediction unit 202 performs the synthesis under the synthesis condition selected in step S104. ~ The evaluation value of x is predicted (step S105). Specifically, the evaluation value prediction unit 202 predicts the evaluation value of x under the synthesis condition ~ The mean μ( ~x), variance σ 2 ( ~ x) is calculated using equation (9).

[0071] Next, the synthesis condition search unit 203 calculates the expected value of equation (10) (the expected value of improvement from the highest evaluation value obtained in the first N trials) using the distribution of evaluation values ​​predicted by equation (4) (step S106).

[0072] Then, the synthesis condition searching unit 203 evaluates the expected values ​​obtained in step S106 above (for example, sorting the expected values ​​obtained so far in ascending or descending order, identifying the maximum value, etc.) (step S107).

[0073] By repeating the above steps S104 to S107, the synthesis condition shown in formula (11) is satisfied. ~ x N+1 can be obtained approximately.

[0074] The synthesis condition output unit 204 outputs the synthesis condition finally obtained by repeating the above steps S104 to S107. ~ x N+1 To obtain the synthesis condition x N+1 (Step S108). Note that this synthesis condition x N+1 is stored in the storage unit 205.

[0075] Then, the synthesis condition output unit 204 outputs the synthesis condition x obtained in step S108. N+1 to the material production device 20 (step S109).

[0076] The synthesis conditions for the N+1th trial are as follows: N+1 Therefore, in the N+1th trial, this synthesis condition N+1 A sample (substance) is generated by the material generation device 20 according to the above, and the sample is evaluated by the material evaluation device 30 to obtain an evaluation value y N+1Such trials are repeated a predetermined number of times, and the synthesis conditions used to generate the sample for which the best evaluation value (for example, the maximum evaluation value) is measured are obtained as the synthesis conditions for obtaining the optimal substance.

[0077] <Experimental Results> Hereinafter, two types of experimental results using the synthesis condition generation system 1 according to this embodiment will be described. The first is an experiment using simulation data (hereinafter referred to as simulation experiment), in which a two-dimensional search space for synthesis conditions is given as simulation data. The second is an experiment to generate SrRuO3 by molecular beam epitaxy (hereinafter referred to as SrRuO3 experiment).

[0078] Simulation experiments In this experiment, the evaluation value expressed by the following formula (13) was used.

[0079]

number

[0080]

number

[0081] The above evaluation value is x≒[0.48,0.52] Τ It reaches a maximum of y ≒ 1.24 near this point.

[0082] The search range for the composite condition is x∈[-1,1]×[-1,1], where

[0083]

number

[0084] The results of this experiment are shown in Fig. 7. In Fig. 7, "Our method" represents the synthesis condition generation system 1 according to this embodiment. "PAD" is a conventional method that treats NaN as an evaluation value of 0. In Fig. 7, the horizontal axis represents the number of trials to obtain an evaluation value, and the vertical axis represents the highest evaluation value excluding NaN obtained up to that trial (i.e., the further the graph is in the upper left (fewer trials and higher evaluation values) the better).

[0085] As shown in Figure 7, our method achieved an evaluation value of over 1.20 after about 15 trials, whereas pad achieved an evaluation value of less than 1.20 even after more than 30 trials. This shows that our method achieved a better evaluation value with fewer trials than the conventional method.

[0086] SrRuO3 experiment The elements of the synthesis conditions and their search ranges are shown in Table 1 below.

[0087] [Table 1] The evaluation value was the residual resistance ratio y:=ρ(300K) / ρ(4K). However, depending on the value of the synthesis condition x used in thin film formation, y=NaN.

[0088] The results of this experiment are shown in Figure 8. In Figure 8, the horizontal axis shows the number of trials to obtain an evaluation value, and the vertical axis shows the highest evaluation value obtained up to that trial. As shown in Figure 8, it can be confirmed that the evaluation value continues to improve as the number of trials increases.

[0089] <Summary> As described above, the synthesis condition generation system 1 according to this embodiment can automatically search in a synthesis condition space where material generation may or may not be successful, and can find synthesis conditions with good evaluation values, and can therefore be used, for example, in the creation of new materials. In addition, the synthesis condition generation system 1 according to this embodiment can optimize (maximize or minimize) evaluation values ​​with fewer trials than conventional methods, and can be expected to have the effect of reducing the development period and development time for new materials and reducing costs.

[0090] The present invention is not limited to the above-described embodiments specifically disclosed, and various modifications, changes, combinations with known technologies, etc. are possible without departing from the scope of the claims.

[0091] [References] Reference 1: J. Mockus, V. Tiesis, and A. Zilinskas: The application of Bayesian methods for seeking the extremum. Towards Global Optimization 2, 1978. Reference 2: N. Srinivas, A. Krause, S. Kakade, and M. Seeger: Gaussian Process Optimization in the Bandit Setting: No Regret and Experimental Design, International Conference on Machine Learning, 2010. Reference 3: E. Contal, V. Perchet, and N. Vayatis: Gaussian Process Optimization with Mutual Information, International Conference on Machine Learning, 2014. [Explanation of symbols]

[0092] 1. Synthesis Condition Generation System 10 Synthesis condition generator 20 Material generation device 30 Material Evaluation Equipment 101 Input Device 102 Display device 103 External I / F 103a Recording media 104 Communication I / F 105 RAM 106 ROM 107 Auxiliary storage 108 processors 109 Bus 201 Evaluation value input section 202 Evaluation value prediction unit 203 Synthesis condition search section 204 Synthesis condition output part 205 Storage section

Claims

1. A synthesis condition generation method for generating synthesis conditions in a trial in which a substance is generated according to a synthesis condition, an evaluation value is measured for evaluating the substance, and the synthesis condition is generated, the synthesis condition generation method comprising the steps of: The evaluation value y of the substance in the N trials so far ｎ' If there is a trial n' where (n'∈{1, . . . , N}) could not be measured, the evaluation value y ｎ' a rating transformation procedure to replace the worst rating value; Synthesis condition x in trial n (n=1, . . . , N) ｎ and the evaluation value y ｎ A prediction procedure for predicting an evaluation value y of an untried synthesis condition x using Using the result of predicting the evaluation value y of the untried synthesis condition x, ｎ The synthesis condition that has a better evaluation value than the best value of (n=1, ..., N) is defined as the synthesis condition x Ｎ＋１ and a generation procedure for generating The computer executes The prediction step comprises: Selecting an untried synthesis condition x from grid points of a predetermined first resolution within a search space of synthesis conditions, and predicting an evaluation value y of the selected untried synthesis condition; After all of the untried synthesis conditions x have been selected from the grid points of the first resolution, a predetermined number of the untried synthesis conditions x are selected based on a predetermined condition, and within an area having the selected untried synthesis conditions x as vertices, untried synthesis conditions x are selected from grid points of a second resolution higher than the first resolution, and an evaluation value y of the selected untried synthesis conditions is predicted.

2. The prediction step comprises: Synthesis condition x in trial n (n=1, . . . , N) ｎ and the evaluation value y ｎ Using the above, the mean μ(x) and variance σ ２ (x) is calculated as a result of said prediction; The generating procedure includes: The mean μ(x) and variance σ ２ Using (x), the expected value that the evaluation value y of the untried synthesis condition x will be better than the best value is calculated, and the untried synthesis condition that maximizes the expected value is defined as the synthesis condition x. Ｎ＋１ The synthetic condition generating method according to claim 1 , wherein the synthetic condition generating method generates the synthetic condition as follows:

3. The prediction step comprises:

3. The synthesis condition generating method according to claim 2, further comprising the steps of: selecting a predetermined number of the untried synthesis conditions x in descending order of expected value after all the untried synthesis conditions x have been selected from the grid points of the first resolution.

4. The synthesis condition generation method includes: Synthesis condition x in trial N Ｎ and the evaluation value y Ｎ and a normalization procedure for normalizing The prediction step comprises: Normalized synthesis condition x ｎ and the evaluation value y ｎ The synthesis condition generating method according to claim 1 , further comprising: predicting an evaluation value y of the untried synthesis condition x using:

5. The evaluation value conversion step includes: All evaluation values ​​y in trial n (n=1, . . . , N) ｎ If (n=1, . . . , N) cannot be measured, y Ｎ The synthesis condition generating method according to claim 1 , wherein:

6. A synthesis condition generating device for generating synthesis conditions in a trial in which a substance is generated according to a synthesis condition, an evaluation value is measured for evaluating the substance, and the synthesis condition is generated, the synthesis condition generating device comprising: The evaluation value y of the substance in the N trials so far ｎ' If there is a trial n' where (n'∈{1, . . . , N}) could not be measured, the evaluation value y ｎ' an evaluation value conversion unit that replaces the worst evaluation value; Synthesis condition x in trial n (n=1, . . . , N) ｎ and the evaluation value y ｎ A prediction unit configured to predict an evaluation value y of an untried synthesis condition x using Using the result of predicting the evaluation value y of the untried synthesis condition x, ｎ The synthesis condition that has a better evaluation value than the best value of (n=1, ..., N) is defined as the synthesis condition x Ｎ＋１ A generating unit configured to generate having The prediction unit is Selecting an untried synthesis condition x from grid points of a predetermined first resolution within a search space of synthesis conditions, and predicting an evaluation value y of the selected untried synthesis condition; After all of the untried synthesis conditions x have been selected from the grid points of the first resolution, a predetermined number of the untried synthesis conditions x are selected based on a predetermined condition, and within an area having the selected untried synthesis conditions x as vertices, an untried synthesis condition x is selected from grid points of a second resolution higher than the first resolution, and an evaluation value y of the selected untried synthesis condition is predicted.

7. A program for causing a computer to execute the synthesis condition generating method according to any one of claims 1 to 5.

Images