Information processing device and information processing method

Abstract

[Problem] To improve the performance of an exploration of the structure of a substance. [Solution] This information processing device comprises at least one processor and at least one memory. The at least one processor acquires information on a target of exploration from another information processing device, acquires structures of a plurality of substances on the basis of the information on the target of exploration, updates the structures of the plurality of acquired substances, evaluates the structures of the plurality of updated substances by using a neural network, and transmits, to the other information processing device, information on the structure of a substance that has been explored, on the basis of the result of the evaluation.

Description

Information processing device and information processing method

[0001] The present disclosure relates to an information processing device and an information processing method.

[0002] Knowing the structure or physical properties, such as the arrangement and bonding state of atoms that make up a crystal, is desired in a wide range of fields. Currently, methods for searching the structure of compounds are being actively researched, but further improvements in speed and precision are desired in this field.

[0003] LB Vilhelmsen, et.al., “A genetic algorithm for first principles global structure optimization of supported nano structures,” J. Chem. Phys., Volume 141, Issue 4, July 28, 2014

[0004] One non-limiting problem that embodiments of the present disclosure seek to solve is the discovery of structures of desired substances.

[0005] According to one embodiment, an information processing device includes at least one processor and at least one memory, and the at least one processor: acquires information about a search target from another information processing device; acquires structures of a plurality of substances based on the information about the search target; updates the acquired structures of the plurality of substances; evaluates the updated structures of the plurality of substances using a neural network; and transmits information about the structures of the searched substances to the other information processing device based on the evaluation result.

[0006] 1 is a block diagram schematically showing an information processing apparatus according to an embodiment; 2 is a block diagram schematically showing an information processing system according to an embodiment; 3 is a block diagram schematically showing an information processing system according to an embodiment; 4 is a diagram schematically showing a processing flow according to an embodiment; 5 is a diagram schematically showing a sampling flow according to an embodiment; 6 is a diagram showing at least one implementation of an information processing apparatus according to an embodiment;

[0007] The problems to be solved by the embodiments of the present disclosure are not limited to the problems described above, and as further examples of some problems that are not limited to these, problems corresponding to the effects described in the embodiments can also be considered. In other words, a problem corresponding to at least one of the effects described in the description of the embodiments of the present disclosure can be considered to be a problem to be solved by the present disclosure.

[0008] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The following describes embodiments of the present invention with reference to the drawings. The drawings and the description of the embodiments are given by way of example only and are not intended to limit the present invention.

[0009] 1 is a block diagram illustrating a schematic diagram of an information processing device according to an embodiment. The information processing device 10 includes a processing circuit 100, a memory circuit 102, and an input / output I / F 104. The information processing device 10 searches for and outputs the structure of a compound based on input information.

[0010] The processing circuit 100 is a circuit (processor) that realizes processing in the information processing device 10. The processing circuit 100 executes various processes based on input information and / or data stored in the memory circuit 102.

[0011] The memory circuitry 102 stores data required for processing by the processing circuitry 100 and / or data required for starting up the information processing device 10 .

[0012] The input / output I / F 104 is an interface that connects the outside and the inside of the information processing device 10. A user interface may be provided as part of the input / output I / F 104.

[0013] In addition, the information processing device 10 may be provided with other components, such as a power supply unit and a control unit, that are necessary for the information processing device 10 to execute processing.

[0014] 2 is a diagram illustrating an example of an information processing system according to an embodiment. The information processing system 1 includes an information processing device 10 and a storage 30.

[0015] The storage 30 can store at least a portion of the data required for processing by the processing circuit 100. The processing circuit 100 can acquire the data stored in the storage 30 via the input / output I / F 104 and execute processing. The storage 30 may be configured to store data, for example, as a file server or part of a server provided on a cloud.

[0016] As shown in this diagram, the memory circuitry 102 does not store all of the data required for the processing of the processing circuitry 100 inside the information processing device 10, but rather the data processed by the processing circuitry 100 and / or data intermediately acquired during the processing of the processing circuitry 100 can be stored in a storage 30 provided outside the information processing device 10. In this way, the information processing system 1 can be configured such that part of the functions of the memory circuitry 102 are located outside the information processing device 10.

[0017] FIG. 3 is a diagram schematically illustrating an example of an information processing system according to an embodiment. The information processing system 1 includes at least an information processing device 10 and an information processing device 20 .

[0018] The information processing device 10 is, for example, an information processing device that operates as a server, and the information processing device 20 is, for example, an information processing device that operates as a client.

[0019] The information processing device 20 includes a processing circuit 200, a memory circuit 202, and an input / output I / F 204. The information processing device 20 sends information to the information processing device 10 along with a request to search for the structure of a compound. The information processing device 10 executes a search for the structure of a compound based on the information and request acquired from the information processing device 20.

[0020] The processing circuit 200 is a circuit (processor) that executes processing in the information processing device 20. The processing circuit 200, for example, receives a request from a user via the input / output I / F 204 and transmits this request to the information processing device 10 via the input / output I / F 204.

[0021] The memory circuit 202 stores data necessary for the operation of the information processing device 20. The memory circuit 202 can also store the results of calculations performed by the information processing device 10 obtained via the input / output I / F 204.

[0022] The input / output I / F 204 is an interface that connects the outside and the inside of the information processing device 20 .

[0023] The input / output I / F 104 and the input / output I / F 204 can be connected via, for example, a wired or wireless network. In this case, the input / output I / F 104 and the input / output I / F 204 are provided with appropriate network I / Fs.

[0024] 2, or may include one or more information processing devices 10 for one or more information processing devices 20. When including multiple information processing devices 10, multiple processors may cooperate to process requests from one or more information processing devices 20.

[0025] In this case, the information processing device 10 may be an information processing device provided on the cloud, or may be a device that provides services to the information processing device 20 in the form of SaaS (Software as a Service).

[0026] A stable structure has lower energy than neighboring structures. When multiple elements are mixed, the compound can be said to have a stable (crystalline) structure if it has a structure with lower energy. By referring to a graph showing the state, you can find out what composition ratios result in a stable structure. For example, a structure on the convex hull can be said to be a stable structure.

[0027] One non-limiting object of the present disclosure is to search for a compound structure by updating the convex hull. One non-limiting problem that the embodiments of the present disclosure attempt to solve is to search for a substance structure that is more energetically stable or has desired physical properties.

[0028] In the present disclosure, the term "substance" refers to various materials and is a concept that includes compounds or simple substances. For example, the substance may be a simple crystal, a carbon nanotube, a fullerene, a nanoparticle, or the like. The structure of a substance may be searched for, for example, the surface structure, molecular structure, or the like of the substance. Hereinafter, when simply referred to as "structure," it may be interpreted as the atomic structure, including the atomic arrangement, atomic arrangement, or the like of the substance.

[0029] 4 is a diagram schematically illustrating the flow of a substance structure search process by the information processing device 10. In the following, the process may be described as the process of the information processing device 10 or the process of the information processing device 20, but this may be interpreted as the process of the processing circuit 100 or the process circuit 200, respectively, to the extent that there is no contradiction.

[0030] The information processing device 10 samples the structure based on given conditions (S10).

[0031] The information processing device 10 updates (relaxes) the sampled structure (S20).

[0032] The information processing device 10 evaluates the updated structure (S30).

[0033] The information processing device 10 stores the structure together with the evaluation as history (S40).

[0034] In the second and subsequent iterations (S10 to S40), the information processing device 10 performs sampling based on the evaluation from the history. When a predetermined calculation completion condition is met, the information processing device 10 outputs a structure according to the evaluation value (S50).

[0035] Each of the above steps will now be described in more detail.

[0036] (Initialization of History) When the information processing device 10 has previously executed a search using the same combination of elements, the information processing device 10 may use the results of this past search or information related to the progress of the search as history.

[0037] When the information processing device 10 performs a search using a combination of elements that has never been performed before, or a new search that does not rely on past performance of combinations that have been performed before, the information processing device 10 can use a structure in which elements are randomly arranged as an initial structure. For example, the information processing device 10 can use, as an initial history, multiple structures in which elements obtained as conditions are randomly arranged within cells in a DFT (Density Functional Theory) calculation.

[0038] As a non-limiting example, the information processing device 10 can arrange elements by dividing a cell into multiple sub-cells, randomly placing elements in one sub-cell, and copying this sub-cell to fill the cell.

[0039] (S10: Sampling of Structure) In sampling of the structure, the information processing device 10 may use, as a non-limiting example, an evolutionary algorithm. The information processing device 10 samples the structure using, for example, a genetic algorithm.

[0040] A genetic algorithm, which is one type of evolutionary algorithm, selects a sample from multiple, typically a huge number of, candidate samples for the parent generation to form the parent generation. Among the samples belonging to this parent generation, crossover between samples and mutation within the samples are performed to generate candidate samples for the child generation.

[0041] In this disclosure, the term structural sampling, or simply sampling, refers to the process of selecting a parent generation sample and generating a child generation sample through a process that includes at least crossover (and may involve mutation).

[0042] 5 is a diagram showing an example of sampling according to an embodiment. The information processing device 10 generates a child generation from history information, and executes subsequent processing using this child generation as a sampled structure.

[0043] The information processing device 10 prepares for generating a child generation by performing elite selection using the history as a parent generation (S100). The elite selection is performed based on evaluations associated with samples in the history.

[0044] The information processing device 10 can make the selection using, for example, the energy value for each structure (sample) obtained as an evaluation in the previous iteration. For example, if it is desired to obtain a crystal structure composed of two elements, the composition on a two-dimensional convex hull can be selected as an elite sample, with the ratio of the two elements on the horizontal axis and the energy [eV / atom] in each structure in various arrangements at the element ratio on the vertical axis.

[0045] In addition, the information processing device 10 can make a selection based on a multi-objective optimization technique using indices related to stability or desired physical characteristics of multiple physical property values, etc. In this case, as a non-limiting example, the information processing device 10 can also make a selection using other evaluation values ​​acquired in the evaluation process of S30, which will be described later.

[0046] As a non-limiting example, the information processing device 10 can select elites based on a partially ordered sequence using a predetermined index as a sort key. A partially ordered sequence generally refers to a relationship in which reflectivity, antisymmetry, and transitivity hold.

[0047] In one embodiment, for structures p and q, p<=q is said to satisfy a partial order law when E(p)<=E(q) and Comp(p)=Comp(q). Here, E(p) is the value of the energy or sorting index of structure p, and Comp(p) is the composition ratio of p (e.g., the composition ratio of two elements). In other words, for different samples p and q having the same composition ratio, if the index value indicates E(p)<=E(q), then p<=q is said to be satisfied.

[0048] The information processing device 10 calculates an index value E for each sample structure stored in the history. The information processing device 10 performs non-dominated sorting on E and extracts samples in ascending order of rank. For example, when there are multiple structures having the same element composition ratio, the information processing device 10 ranks these structures in ascending order of index and selects them as samples in ascending order of rank.

[0049] The information processing device 10 can further select samples in ascending order of rank, and when the rank exceeds the number of samples to be selected, select a sample using the congestion degree D. The congestion degree D of sample p can be, for example, D(p) = (Manhattan distance from two samples near p), D(p) = (average of Manhattan distances from multiple samples near p), or D(p) = (Manhattan distance from multiple samples near p) + (normalized energy gap of p). "Nearby" refers to other samples with a similar index, for example, energy value.

[0050] By making this selection, the information processing device 10 can perform elite selection, but the sample selected in this manner is always an end point of a convex hull in a multidimensional space (for example, a space formed by energy values ​​relative to the composition ratios of multiple elements). Therefore, the selected sample is likely to be a sample that shows a structure that has not been improved over a long iteration period.

[0051] The information processing device 10 can filter the samples before non-dominated sorting in order to appropriately select or not select such samples. The filtering can be expressed by, for example, the following equation.

[0052] g(t) is the current generation of t, and g * (t) is the generation of t that satisfies the following:

[0053] "history" represents the entire structure (history) searched so far. "r(t)" is a multidimensional vector representing the composition of structure "t", and "α", "ε", and "δ" are hyperparameters. "||" represents the Manhattan norm of the multidimensional vector. For example, the information processing device 10 excludes "t" that satisfies formula (1) from candidates for elite selection.

[0054] By performing non-dominated sorting and crowding sorting on the candidates selected in this manner, the information processing device 10 can perform elite selection. This selection makes it possible to remove from candidates individuals whose structures are present in large numbers in the vicinity, that is, individuals belonging to areas where many individuals having similar genes have already been evaluated, and to preferentially select individuals from other areas, that is, areas where little searching has been performed in the history, as samples for the next generation.

[0055] After selecting the sample, the information processing device 10 generates a child generation sample by using the selected sample itself, crossing over a plurality of selected samples, and mutating the selected sample (S102).

[0056] When the information processing device 10 uses the selected sample itself, it uses the structure in the history of the parent generation as one of the samples of the structure of the child generation. For example, the elite and one sample selected in the figure are used as the structure belonging to the child generation.

[0057] When crossing samples, the information processing device 10 replaces at least a part of the structure of a selected parent generation with a part of the structure of another selected parent generation. For example, in the figure, part of the structure of sample 1 is replaced with part of the structure of sample 3 to create a sample of the child generation indicated by 1+3.

[0058] Crossover can be performed, for example, on a cell-by-cell basis in DFT calculations, etc. By replacing the element structure in at least a partial region of a structure shown in a cell in a certain sample with the element structure in a partial region of another sample, a crossed sample can be obtained.

[0059] Furthermore, crossover is not limited to the above, and may be performed, for example, by exchanging part of a structure represented by a graph with part of a structure of another graph, or by exchanging part of a structure described in SMILES with part of a structure described in another SMILES.

[0060] When mutating a sample, the information processing device 10 changes the type of at least some of the elements in the structure of the sample, the arrangement of at least some of the elements, or the connections between at least some of the elements to something different. For example, the structure of sample 4 in the figure is mutated to create sample 4'.

[0061] For example, the information processing device 10 can mutate a structure in one sample by swapping two different types of elements. For example, the information processing device 10 can mutate a structure in one sample by moving the positions of four rectangular elements so that they form a parallelogram. As other examples, the information processing device 10 can mutate a structure by placing (adding) random elements in random positions, deleting elements, or randomly changing the type of elements.

[0062] Additionally, although not shown in the figure, a child generation sample may be generated by performing both crossover and mutation on two samples. For example, a child generation sample (1+2)' may be generated from sample 1 and sample 2.

[0063] In this way, the information processing device 10 performs sampling of the structure of a substance using a genetic algorithm. Of course, the information processing device 10 can similarly perform sampling using other evolutionary algorithm techniques.

[0064] The information processing device 10 can also perform sampling using a method based on a surrogate model such as Bayesian optimization, a sampling-based method such as Markov Chain Monte Carlo (MCMC), or a method based on the symmetry of a space group.

[0065] (S20: Update (Relaxation), S30: Evaluation) After sampling, the information processing device 10 updates the sample for each sampled structure so that the structure is established as a stable structure or a structure having desirable physical properties.

[0066] The structure obtained by the above crossover and mutation may not be stable as a crystal structure, for example, because the atoms are too close to each other. In order to update the generated unstable structure to a stable structure, the information processing device 10 performs relaxation processing.

[0067] The information processing device 10 can locally optimize the sampled structure using, for example, NNP (Neural Network Potential).

[0068] The information processing device 10 applies NNP to the sampled structure to calculate the energy. The information processing device 10 performs structural optimization to obtain a locally stable structure so that the energy value becomes small. The information processing device 10 can perform local structural optimization by, for example, calculating the position derivative with respect to the energy.

[0069] Through this process, the information processing device 10 updates (relaxes) the sample obtained by crossover and / or mutation with a stable structure near the sample. In other words, the information processing device 10 updates (relaxes) the sample to a more stable structure.

[0070] More specifically, the information processing device 10 updates the sample to a sample indicating a point where the gradient of the NNP's potential becomes 0 by applying the BFGS (Broyden-Fletcher-Goldfarb-Shanno Algorithm) or the like using the target sample as an initial value. The term "neighborhood" may refer to a region including the point where the gradient becomes 0 and the sample (before updating). The shape of such a region varies depending on the shape of the NNP's potential surface and the position of the sample.

[0071] The information processing device 10 can employ any appropriate optimization method. For example, the information processing device 10 can locally employ the BFGS method to obtain a stable structure from the energy values ​​of nearby structures. In addition, the information processing device 10 can use a hill-climbing method, a quasi-Newton method, or the like to relax the structure and update the sampled structure.

[0072] This process makes it possible to update structures obtained from the genetic algorithm that are likely to be inappropriate to more stable structures and then perform the search.

[0073] In addition, the information processing device 10 can also evaluate the sampled structure in S30 during the process of S20. That is, the evaluation value obtained for optimization in the process of S20 can be used as the evaluation value in S30.

[0074] In the process of S10, if an index other than the evaluation used in the local optimization of S20 is used as the evaluation, the calculation process of that index can be performed on the updated sample after the process of S20 is completed. Even in this case, if the evaluation value used in the process of S20 is part of the evaluation in S30, the evaluation value obtained during the process of S20 can be used.

[0075] The information processing device 10 can evaluate the sampled structure by, for example, calculating the energy of each relaxed structure and finding the difference between the calculated energy and the energy of the true convex hull.

[0076] Since searches are generally performed in situations where the true convex hull is unknown, instead, the information processing device 10 can calculate and evaluate the distance between the convex hull of the current energy and the energy of each calculated structure, for example, the distance to the structure belonging to the current convex hull.

[0077] As another example, the information processing device 10 can use a known energy value, for example, a publicly available database, as a convex hull and use the distance between this convex hull and the convex hull of the sampled structure as an evaluation.

[0078] As another example, the information processing device 10 can use the area of ​​the convex hull or the length of the line segments as an overall evaluation, or can use the distance between a point indicating a certain configuration and the convex hull as an evaluation.

[0079] (S40: Store, S50: Output) The information processing device 10 associates the evaluation result with the evaluated structure and stores them (S40). The stored structure is registered as history and can be used as a parent generation sample in the next iteration.

[0080] The information processing device 10 repeats the processes from S10 to S40 until a termination condition is met. The termination condition can be set in the same manner as in general optimization, such as until the number of properly evaluated structures reaches a predetermined number, until the time required for the search has passed a predetermined time, until a predetermined physical property value of the searched structure reaches a predetermined value, or until a predetermined number of iterations have been performed. After the termination condition is met, the information processing device 10 outputs information about the finally obtained sample (S50).

[0081] The information processing device 10 can output, for example, structures for samples that make up the convex hull, and can also output structures whose distance from the convex hull is within a predetermined value, such as 0.05 eV / atom. Such structures may be synthesized within the predetermined value, as errors may occur in calculations such as NNP and DFT. In such cases, it is meaningful to output the structures as search result candidates. Furthermore, when searching based on physical property values, it is possible to return stable or metastable structures whose physical property values ​​fall within a desired range.

[0082] By optimizing and outputting the convex hull that can be obtained for evaluation or sampling, the information processing device 10 can perform a search based on the Pareto front in multivariate optimization or a search based on pseudo-Pareto front optimization and output the results.

[0083] The information processing device 10 can output, for example, the type of element, the position of atoms, and the cell structure, which are parameters that determine the crystal structure, as information related to the structure. The information processing device 10 can also output energy or physical property values.

[0084] The information processing device 10 may also output a search history, which makes it possible to present to the user information on how the search was performed, for example, whether the search was performed efficiently.

[0085] The information processing device 10 can also provide a viewer to output information for visualizing the structure. This output can be presented, for example, via the input / output I / Fs 104 and 204.

[0086] The information processing device 10 can also output intermediate progress as needed even during the execution of the search. The information processing device 10 can output not only the structure but also, for example, evaluation value information as a graph or the state of the convex hull as a graph.

[0087] According to this embodiment, the information processing device 10 can automatically search for the structure of a substance composed of the elements simply by inputting conditions such as the type of element to be used. Therefore, for example, the information processing system 1 shown in Fig. 3 can provide a service as SaaS, in which high-speed calculations are performed on the server side and the search results are output to the client simply by inputting information on the type of element (e.g., a substance containing Li and In, or a crystal containing Ti and O) from the client side.

[0088] Calculations based on classical theory, such as first-principles calculations, molecular dynamics calculations, or density functional theory calculations, that obtain values ​​that can serve as evaluation values ​​for energy and the like, require a long time for calculations on each individual structure.

[0089] When using a genetic algorithm as an example that is not limited to sampling, the information processing device 10 may execute this very long process multiple times to evaluate multiple samples of the offspring generation, and it may take several days to several weeks, for example, to obtain evaluation values ​​for one generation and acquire a stable structure.

[0090] By using the NNP method, it is possible to quickly and accurately calculate the evaluation value and relax the structure. As a result, according to one embodiment, it is possible to realize a global and robust search for the structure of a material using an evolutionary algorithm such as a high-speed genetic algorithm.

[0091] When a genetic algorithm is used as an example of sampling, it is possible to simultaneously optimize multiple samples. As a result of being able to increase the number of samples that can be sampled, it is possible to realize high-speed and global structure search. By using NNP, it is possible to quickly perform evaluation, which has been a bottleneck in searches using multiple samples simultaneously. In one embodiment of the present disclosure, by using NNP in this way, it is possible to quickly perform structure search of substances while ensuring robustness.

[0092] Using multiple samples simultaneously makes it possible to achieve more accurate processing. In addition, the speedup of evaluation using NNP makes it possible to expand the scope of search more than ever before by using multiple general-purpose accelerators such as GPUs, without using a supercomputer.

[0093] Furthermore, information that can be input as conditions is not limited to element combinations. For example, the information processing device 10 may also accept environmental information, such as temperature and pressure, as conditions. In this case, the information processing device 10 can reflect this environmental information, such as temperature and pressure, when acquiring physical property values ​​in NNP, thereby enabling more accurate local search for stable structures under certain conditions. As a result, the information processing device 10 can execute the search results as a whole process more based on the conditions.

[0094] The information to be input as the conditions can be input by the user via the client-side information processing device 20, for example.

[0095] The conditions may be set as a combination of elements and their composition ratios. The composition ratios may be expressed as ranges. In this case, the information processing device 10 may, as a non-limiting example, execute processing based on the composition ratios at the timing of sampling (e.g., initial sample generation, crossover, mutation). When the number of elements is large, it is possible to limit the search range.

[0096] The information processing device 10 can, for example, determine the range of the composition ratio of Li to In and perform sampling. This sampling makes it possible to search for the structure of a substance in different states, such as a charged state and a discharged state.

[0097] The information processing device 10 can, for example, determine the range of the abundance ratio of a certain element between multiple layers that make up a semiconductor and perform sampling. This sampling can search for structures such as the atomic arrangement between fixed semiconductor layers. Even a small amount of converted atoms can significantly change the physical properties of a semiconductor, and the search can cover such changes in physical properties as well.

[0098] The information processing device 10 can specify a range in a curve formed by an already acquired convex hull as an example of a search range and perform sampling. This sampling allows for a search within the specified range to be performed with high granularity.

[0099] The range of physical property values ​​such as energy and density may be set as a condition. In this case, the information processing device 10 can perform local optimization with higher accuracy when using NNP. Furthermore, it is possible to obtain a structure that is closer to the conditions in evaluation using multi-objective optimization.

[0100] As a condition, the possibility of synthesis for the acquired structure may be output. This possibility of synthesis may be determined based on an existing process.

[0101] The evaluation function to be used may be set as a condition.

[0102] As described above, the information processing device 10 can output the new structure itself. In addition, the information processing device 10 can also output at least one of the following: the state of the convex hull of the energy or other variables used in the evaluation, a visualization of the parent from which each structure was obtained (including changes over time), the optimization history, or the physical property values ​​of the structure (crystal group, cell structure).

[0103] According to the above-described embodiment, since the calculation of the physical property values ​​to be evaluated can be realized at high speed using NNP, the number of individuals per generation of the evolutionary algorithm can be increased. In other words, it is possible to further increase the degree of parallelism when using an accelerator with multiple cores, a cluster, or grid computing. As a result, it is possible to improve the calculation speed and improve the accuracy by widening the search range.

[0104] 3, an information processing device 10 may perform a search using an evolutionary algorithm, and an information processing device 20 may perform an evaluation. As another example, an information processing device 10 may perform a search using an evolutionary algorithm, and another information processing device 10 may perform an evaluation. In this way, each process can be executed by an appropriate information processing device as appropriate.

[0105] Fig. 6 is a diagram showing how to evaluate the results output in the above embodiment. The information processing device 10 in Fig. 1 to Fig. 3 or the information processing device 20 in Fig. 3 can evaluate the results of a completed search (S60).

[0106] The information processing device 10 or the information processing device 20 may, for example, treat the sample structure output as a search result as a candidate for the structure to be obtained, and in this case, by evaluating the multiple candidate structures obtained by the search, it is possible to extract the most desirable structure from among the candidates.

[0107] For example, in addition to the evaluation in S30, physical properties such as energy can be calculated using classical DFT calculations. By evaluating in this way, it is possible to calculate energy values ​​using a different calculation method for candidates quickly searched using NNP, thereby increasing the accuracy of obtaining the desired structure.

[0108] Furthermore, when the information processing device 10 or the information processing device 20 uses a known convex hull as the object of evaluation, the information processing device 10 or the information processing device 20 may calculate values ​​such as energy for a structure belonging to the known convex hull by a classical method or a method using NNP. By acquiring the energy value of the known structure in this way, it becomes possible to confirm whether the evaluation in the information processing device 10 has been performed appropriately.

[0109] The information processing device 10 acquires at least information about the search target from the information processing device 20 .

[0110] This information about the search target refers to information acquired from the information processing device 20 used by the user. This information about the search target may include information such as the types of elements constituting the substance and the composition ratios, as described in the embodiments. It is not limited to this, and may include any information indicating the substance to be searched for or any information indicating the search conditions (temperature, pressure, range of convex hull, range of physical property values, etc.). The information processing device 10 may also acquire information about the search target from the information processing device 20 via another device.

[0111] The information processing device 10 acquires structures of a plurality of substances based on the acquired information on the search target, and updates the acquired structures of the plurality of substances.

[0112] Obtaining the structures of a plurality of substances may be a concept that includes at least one of the following.

[0113] That is, the information processing device 10 may acquire the structures of multiple substances by sampling.

[0114] The information processing device 10 may also acquire structures of multiple substances by random arrangement.

[0115] The information processing device 10 may also acquire structures of multiple substances based on history information. Even when performing elite selection using history, the information processing device 10 may select at least one of the candidates as a substance structure generated by random placement. In this way, a global search can be incorporated into a local search.

[0116] Furthermore, information on the structure of a substance may be a concept including at least one of the following:

[0117] That is, information on the structure of a substance may be information including the type of atom (element) and the position information of the atom (element).

[0118] Furthermore, the information on the structure of a substance may be information including information on a cell structure.

[0119] The information processing device 10 may be configured to acquire the structure of at least one substance based on the acquired information on the search target, and update the acquired structure of at least one substance.

[0120] The information processing device 10 can evaluate the updated structures of the plurality of substances using a neural network (e.g., NNP) and, based on the evaluation results, transmit information on the structures of the searched substances to the information processing device 20. Furthermore, the information processing device 10 may transmit information on the structures of the searched substances to the information processing device 20 via another device.

[0121] NNP can be considered a type of neural network.

[0122] This does not exclude the use of other neural network models other than the neural network that constructs a graph based on the so-called atomic arrangement, and the evaluation of the structure by these other neural network models. In other words, the NNP in the present disclosure does not exclude the use of a neural network model that acquires information about the structure of a substance as a broader concept.

[0123] The information processing device 10 can update the structures of the multiple substances that have been obtained using a neural network.

[0124] This neural network may be the NNP neural network described above, or a different model. The information processing device 10 may be configured to evaluate the updated information on the structure of at least one substance using a neural network.

[0125] The information processing device 10 may also be configured to transmit information on the structures of one or more searched substances to the information processing device 20 .

[0126] As described above, there may be multiple information processing devices 20.

[0127] Similarly, the information processing device 10 may be configured as a system including multiple devices.

[0128] The information processing system 1 described in the present disclosure may include one or more information processing devices 10 and one or more information processing devices 20. Similarly, the information processing system 1 may further include one or more storages 30.

[0129] In the information processing system 1, these components are typically connected via a network line using wireless or wired means such as the Internet, but the connection method is not limited to these and any means that can appropriately connect these components can be used.

[0130] As an example of this configuration, the information processing system 1 can also provide SaaS in which the information processing device 10 is a server and the information processing device 20 is a client.

[0131] Some or all of the devices (information processing device 10 or information processing device 20) in the above-described embodiments may be configured as hardware, or may be configured as software (program) information processing executed by a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit), etc. When configured as software information processing, software that realizes at least some of the functions of each device in the above-described embodiments may be stored on a non-transitory storage medium (non-transitory computer-readable medium) such as a CD-ROM (Compact Disc-Read Only Memory) or a USB (Universal Serial Bus) memory, and the software information processing may be executed by loading the software into a computer. The software may also be downloaded via a communication network. Furthermore, all or part of the software processing may be implemented in a circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array), thereby allowing the software information processing to be executed by hardware.

[0132] The storage medium that stores the software may be a removable medium such as an optical disk, or a fixed medium such as a hard disk or memory. The storage medium may be located inside the computer (such as a main memory or auxiliary memory), or may be located outside the computer.

[0133] 7 is a block diagram showing an example of the hardware configuration of each device (information processing device 10 or information processing device 20) in the above-described embodiment. Each device may be realized as a computer 7 including, for example, a processor 71, a main storage device 72 (memory), an auxiliary storage device 73 (memory), a network interface 74, and a device interface 75, which are connected via a bus 76.

[0134] Although the computer 7 in FIG. 7 includes one of each component, it may include multiple of the same component. Also, while FIG. 7 shows one computer 7, the software may be installed on multiple computers, and each of the multiple computers may execute the same or different parts of the software. In this case, a distributed computing system may be used in which the computers communicate with each other via a network interface 74 or the like to execute the processes. In other words, each device (information processing device 10 or information processing device 20) in the above-described embodiment may be configured as a system in which one or more computers execute instructions stored in one or more storage devices to realize its functions. Furthermore, the system may be configured such that information sent from a terminal is processed by one or more computers located on a cloud, and the processing results are sent to the terminal.

[0135] The various calculations of each device (information processing device 10 or information processing device 20) in the above-described embodiments may be executed in parallel using one or more processors, or using multiple computers via a network. Furthermore, the various calculations may be distributed to multiple processing cores within a processor and executed in parallel. Furthermore, some or all of the processes, means, etc. disclosed herein may be realized by at least one of a processor and a storage device provided on a cloud that can communicate with computer 7 via a network. Thus, each device in the above-described embodiments may be implemented in the form of parallel computing using one or more computers.

[0136] The processor 71 may be an electronic circuit (processing circuit, processing circuitry, CPU, GPU, FPGA, ASIC, etc.) that at least controls or performs calculations on a computer. The processor 71 may also be a general-purpose processor, a dedicated processing circuit designed to perform a specific calculation, or a semiconductor device that includes both a general-purpose processor and a dedicated processing circuit. The processor 71 may also include an optical circuit or a calculation function based on quantum computing.

[0137] The processor 71 may perform arithmetic processing based on data or software input from each device, etc., configured inside the computer 7, and may output arithmetic results or control signals to each device, etc. The processor 71 may control each component constituting the computer 7 by executing the OS (Operating System) of the computer 7, applications, etc.

[0138] Each device (information processing device 10 or information processing device 20) in the above-described embodiment may be realized by one or more processors 71. Here, the processor 71 may refer to one or more electronic circuits arranged on one chip, or may refer to one or more electronic circuits arranged on two or more chips or two or more devices. When multiple electronic circuits are used, the electronic circuits may communicate with each other via wire or wirelessly.

[0139] The main memory device 72 may store instructions executed by the processor 71, various data, etc., and information stored in the main memory device 72 may be read by the processor 71. The auxiliary memory device 73 is a memory device other than the main memory device 72. Note that these memory devices refer to any electronic component capable of storing electronic information and may be semiconductor memory. The semiconductor memory may be either volatile memory or non-volatile memory. The memory device for saving various data, etc. in each device (information processing device 10 or information processing device 20) in the above-described embodiments may be realized by the main memory device 72 or the auxiliary memory device 73, or may be realized by an internal memory built into the processor 71. For example, the memory circuit 102 or the memory circuit 202 in the above-described embodiments may be realized by the main memory device 72 or the auxiliary memory device 73.

[0140] When each device (information processing device 10 or information processing device 20) in the above-described embodiment is configured with at least one storage device (memory) and at least one processor connected (coupled) to this at least one storage device, at least one processor may be connected to one storage device. Also, at least one storage device may be connected to one processor. Also, a configuration in which at least one processor among multiple processors is connected to at least one storage device among multiple storage devices may be included. Also, this configuration may be realized by storage devices and processors included in multiple computers. Furthermore, a configuration in which a storage device is integrated with a processor (for example, a cache memory including an L1 cache and an L2 cache) may be included.

[0141] The network interface 74 is an interface for connecting to the communication network 8 wirelessly or via a wire. The network interface 74 may be an appropriate interface, such as one that conforms to an existing communication standard. The network interface 74 may exchange information with an external device 9A connected via the communication network 8. The communication network 8 may be any one of a WAN (Wide Area Network), a LAN (Local Area Network), a PAN (Personal Area Network), etc., or a combination thereof, as long as information is exchanged between the computer 7 and the external device 9A. An example of a WAN is the Internet, an example of a LAN is IEEE 802.11 or Ethernet (registered trademark), and an example of a PAN is Bluetooth (registered trademark) or NFC (Near Field Communication), etc.

[0142] The device interface 75 is an interface such as USB that directly connects to the external device 9B.

[0143] The external device 9A is a device connected to the computer 7 via a network. The external device 9B is a device directly connected to the computer 7.

[0144] For example, the external device 9A or the external device 9B may be an input device. The input device may be a device such as a camera, a microphone, a motion capture device, various sensors, a keyboard, a mouse, or a touch panel, and provides acquired information to the computer 7. Alternatively, the external device 9A or the external device 9B may be a device equipped with an input unit, a memory, and a processor, such as a personal computer, a tablet terminal, or a smartphone.

[0145] Furthermore, the external device 9A or the external device 9B may be, for example, an output device. The output device may be, for example, a display device such as an LCD (Liquid Crystal Display) or an organic EL (Electro Luminescence) panel, or a speaker that outputs sound or the like. Alternatively, the external device 9A or the external device 9B may be a device including an output unit, a memory, and a processor, such as a personal computer, a tablet terminal, or a smartphone.

[0146] Furthermore, the external device 9A or the external device 9B may be a storage device (memory). For example, the external device 9A may be a network storage or the like, and the external device 9B may be a storage such as an HDD.

[0147] Furthermore, the external device 9A or the external device 9B may be a device having some of the functions of the components of each device (information processing device 10 or information processing device 20) in the above-described embodiments. In other words, the computer 7 may transmit some or all of the processing results to the external device 9A or the external device 9B, or may receive some or all of the processing results from the external device 9A or the external device 9B.

[0148] In this specification (including the claims), when the expression "at least one of a, b, and c" or "at least one of a, b, or c" (including similar expressions) is used, it includes any of a, b, c, a-b, a-c, b-c, or a-b-c. It may also include multiple instances of any element, such as a-a, a-b-b, a-a-b-b-c-c, etc. It also includes the addition of elements other than the listed elements (a, b, and c), such as having d, as in a-b-c-d.

[0149] In this specification (including claims), when expressions such as "using / using data as input / based on / according to / in response to" (including similar expressions) are used, unless otherwise specified, this includes cases where the data itself is used, or where data that has been processed in some way (e.g., data with noise added, normalized data, features extracted from data, intermediate representations of data, etc.) is used. Furthermore, when a statement is made that a result is obtained "using data as input / based on / according to / in response to" (including similar expressions), this includes cases where the result is obtained based solely on the data, or where the result is influenced by other data, factors, conditions, and / or states other than the data, unless otherwise specified. Furthermore, when a statement is made that "data is output" (including similar expressions), this includes cases where the data itself is used as output, or where data that has been processed in some way (e.g., data with noise added, normalized data, features extracted from data, intermediate representations of data, etc.) is used as output, unless otherwise specified.

[0150] When the terms "connected" and "coupled" are used in this specification (including the claims), they are intended as open-ended terms that encompass any of direct connection / coupling, indirect connection / coupling, electrically connection / coupling, communicatively connection / coupling, functionally connection / coupling, and physically connection / coupling. These terms should be interpreted appropriately according to the context in which they are used, but any connection / coupling form that is not intentionally or naturally excluded should be interpreted as being included in these terms without limitation.

[0151] In this specification (including the claims), the expression "A configured to B" may include the physical structure of element A having a configuration capable of performing operation B, and the permanent or temporary setting / configuration of element A being configured / set to actually perform operation B. For example, if element A is a general-purpose processor, it is sufficient that the processor has a hardware configuration capable of performing operation B, and is configured to actually perform operation B by setting a permanent or temporary program (instruction). Also, if element A is a dedicated processor or dedicated arithmetic circuit, it is sufficient that the circuit structure of the processor is implemented to actually perform operation B, regardless of whether control instructions and data are actually attached to it.

[0152] When used in this specification (including the claims), terms implying containing or possessing (e.g., "comprising" or "including" and "having"), they are intended to be open-ended terms that include containing or possessing things other than the object designated by the object of the term. When the object of such terms implies no quantity or a singular number (e.g., expressions using the articles "a" or "an"), the expression should be construed as not being limited to a specific number.

[0153] In this specification (including the claims), although expressions such as "one or more" or "at least one" are used in some places and expressions that do not specify a quantity or that imply a singular number (expressions using the articles a or an) are used in other places, the latter expressions are not intended to mean "one." In general, expressions that do not specify a quantity or that imply a singular number (expressions using the articles a or an) should be interpreted as not necessarily being limited to a specific number.

[0154] In this specification, when a particular advantage / result is described as being obtained from a particular configuration of an embodiment, it should be understood that the same advantage / result can also be obtained from one or more other embodiments having the same configuration, unless otherwise stated. However, it should be understood that the presence or absence of the effect generally depends on various factors, conditions, and / or circumstances, and that the effect is not necessarily obtained by the configuration. The effect is merely obtained by the configuration described in the embodiment when various factors, conditions, and / or circumstances are satisfied, and the effect does not necessarily occur in a claimed invention that defines the same or a similar configuration.

[0155] When terms such as "maximize" and "maximization" are used in this specification (including the claims), they include finding a global maximum, finding an approximation of a global maximum, finding a local maximum, and finding an approximation of a local maximum, and should be interpreted accordingly according to the context in which the term is used. They also include finding approximations of these maxima probabilistically or heuristically. Similarly, when terms such as "minimize" and "minimization" are used, they include finding a global minimum, finding an approximation of a global minimum, finding a local minimum, and finding an approximation of a local minimum, and should be interpreted accordingly according to the context in which the term is used. They also include finding approximations of these minima probabilistically or heuristically. Similarly, when terms such as "optimize" and "optimization" are used, they include finding a global optimum, finding an approximation of a global optimum, finding a local optimum, and finding an approximation of a local optimum, and should be interpreted accordingly according to the context in which the term is used. It also includes finding approximations of these optimum values ​​probabilistically or heuristically.

[0156] In this specification (including claims), when multiple pieces of hardware perform a predetermined process, the pieces of hardware may cooperate to perform the predetermined process, or some of the hardware may perform all of the predetermined process. Furthermore, some of the hardware may perform part of the predetermined process, and other hardware may perform the rest of the predetermined process. In this specification (including claims), when expressions such as "one or more pieces of hardware perform a first process, and the one or more pieces of hardware perform a second process" (including similar expressions) are used, the hardware performing the first process and the hardware performing the second process may be the same or different. In other words, it is sufficient that the hardware performing the first process and the hardware performing the second process are included in the one or more pieces of hardware. Note that hardware may also include an electronic circuit or a device including an electronic circuit.

[0157] In this specification (including the claims), when multiple storage devices (memories) store data, each of the multiple storage devices may store only a portion of the data, or may store the entire data. Also, a configuration in which only some of the multiple storage devices store data may be included.

[0158] Although the embodiments of the present disclosure have been described in detail above, the present disclosure is not limited to the individual embodiments described above. Various additions, modifications, substitutions, and partial deletions are possible within the scope of the conceptual idea and spirit of the present disclosure, which is derived from the content defined in the claims and their equivalents. For example, when numerical values ​​or formulas are used in the above-described embodiments, they are shown for illustrative purposes and do not limit the scope of the present disclosure. Furthermore, the order of each operation shown in the embodiments is also illustrative and does not limit the scope of the present disclosure.

[0159] The above embodiments can be summarized or expanded as follows.

[0160] (1) An information processing device comprising at least one processor and at least one memory, wherein the at least one processor acquires information about a search target from another information processing device, acquires structures of a plurality of substances based on the information about the search target, updates the acquired structures of the plurality of substances, evaluates the updated structures of the plurality of substances using a neural network, and transmits information about the structures of the searched substances to the other information processing device based on the evaluation result.

[0161] (2) The information processing device according to (1), wherein the at least one processor updates the structures of the plurality of substances obtained using the neural network.

[0162] (3) The information processing device according to (2), wherein the at least one processor performs the update by converting the acquired structure of the substance into a nearby stable structure.

[0163] (4) The information processing device according to (1), wherein the at least one processor evaluates the updated structures of the plurality of substances based on a multi-objective optimization technique.

[0164] (5) The information processing device according to (4), wherein the at least one processor acquires structures of the plurality of substances based on a Pareto front.

[0165] (6) The information processing device according to (5), wherein the at least one processor acquires the structures of the plurality of substances by using an evolutionary algorithm.

[0166] (7) The information processing device according to any one of (1) to (6), wherein the at least one processor repeatedly executes the acquisition, the update, and the evaluation until a predetermined condition is satisfied.

[0167] (8) The information processing device according to (7), wherein the at least one processor executes the obtaining, the updating, and the evaluating in parallel.

[0168] (9) The information processing device according to any one of (1) to (8), wherein the information about the search target includes information about at least one of the type of element or the composition ratio of the element.

[0169] (10) The information processing device according to (9), wherein the information about the search target further includes information about at least any one of temperature, pressure, a range of a convex hull, or a range of a physical property value.

[0170] (11) The information processing device according to any one of (1) to (10), wherein the at least one memory transmits information relating to the convex hull of the searched substance to the other information processing device.

[0171] (12) The information processing device according to any one of (1) to (11), wherein the neural network is a neural network potential (NNP).

[0172] (13) The information processing device according to (12), wherein the at least one processor evaluates the updated structures of the plurality of substances by energy.

[0173] (14) The information processing device according to (12) or (13), wherein the at least one memory evaluates the structure of the searched substance using DFT (Density Functional Theory) calculations.

[0174] (15) An information processing system comprising: the information processing device according to any one of (1) to (14); and at least the other information processing device.

[0175] (16) The information processing system according to (15), wherein the neural network is a neural network potential (NNP).

[0176] (17) The information processing system according to (16), wherein the structure of the searched substance is evaluated by at least one of the information processing device and the other information processing device using DFT (Density Functional Theory) calculations.

[0177] (18) An information processing method, comprising: by at least one processor, acquiring information on a search target from another information processing device; acquiring structures of a plurality of substances based on the information on the search target; updating the acquired structures of the plurality of substances; evaluating the updated structures of the plurality of substances using a neural network; and transmitting information on the structures of the searched substances to the other information processing device based on the evaluation results.

[0178] (19) The information processing method according to (18), wherein at least a part of the information processing device according to (1) to (17) is executed by the at least one processor.

[0179] 1: Information processing system, 10: Information processing device, 100: Processing circuit, 102: Memory circuit, 104: Input / output I / F, 20: Information processing device, 200: Processing circuit, 202: Memory circuit, 204: Input / output I / F, 30: Storage

Claims

1. An information processing device comprising at least one processor and at least one memory, wherein the at least one processor acquires information relating to a search target from another information processing device, acquires structures of a plurality of substances based on the information about the search target, updates the acquired structures of the plurality of substances, evaluates the updated structures of the plurality of substances using a neural network, and transmits information on the structures of the searched substances to the other information processing device based on a result of the evaluation.

2. The information processing device according to claim 1, wherein the at least one processor uses the neural network to update the structures of the plurality of substances obtained.

3. The information processing device according to claim 2, wherein the at least one processor performs the updating by converting the acquired structures of the plurality of substances into nearby stable structures.

4. The information processing device according to claim 1, wherein the at least one processor evaluates the updated structures of the plurality of substances based on a multi-objective optimization technique.

5. The information processing device according to claim 4, wherein the at least one processor acquires structures of the plurality of substances based on a Pareto front.

6. The information processing device according to claim 5, wherein the at least one processor obtains structures of the plurality of substances by using an evolutionary algorithm.

7. The information processing device according to claim 1, wherein the at least one processor repeatedly executes the acquisition, the update, and the evaluation until a predetermined condition is satisfied.

8. The information processing device according to claim 7, wherein the at least one processor executes the obtaining, the updating, and the evaluating in parallel.

9. The information processing device according to claim 1, wherein the information on the search target includes at least information on either the type of element or the composition ratio of the element.

10. The information processing device according to claim 9, wherein the information on the search target further includes information on at least any one of temperature, pressure, a range of a convex hull, or a range of a physical property value.

11. The information processing device according to claim 1, wherein said at least one memory transmits information regarding the convex hull of the searched materials to said other information processing device.

12. The information processing device according to any one of claims 1 to 11, wherein the neural network is a Neural Network Potential (NNP).

13. The information processing device according to claim 12, wherein the at least one processor evaluates the updated structures of the plurality of materials by energy.

14. The information processing device according to claim 12, wherein the at least one memory evaluates the structure of the searched substance using DFT (Density Functional Theory) calculations.

15. An information processing system comprising: an information processing device according to any one of claims 1 to 11; and at least the other information processing device.

16. The information processing system according to claim 15, wherein the neural network is NNP (Neural Network Potential).

17. The information processing system according to claim 16, wherein the structure of the substance searched for is evaluated by at least one of the information processing device and the other information processing device using DFT (Density Functional Theory) calculations.

18. An information processing method, comprising: at least one processor: acquiring information regarding a search target from another information processing device; acquiring structures of a plurality of substances based on the information regarding the search target; updating the acquired structures of the plurality of substances; evaluating the updated structures of the plurality of substances using a neural network; and transmitting information on the structures of the searched substances to the other information processing device based on a result of the evaluation.

Images