Information processing device and information processing method

Abstract

[Problem] Achieving high performance in a search for the structure of a substance. [Solution] An information processing device comprises at least one processor and at least one memory. The at least one processor acquires information about a search target from another information processing device, acquires a plurality of structures of a substance based on the information about the search target, updates the acquired plurality of structures of the substance, evaluates the updated plurality of structures of the substance using a neural network, and transmits information about a searched structure of a substance to the other information processing device based on a result of the evaluation.

Description

Area

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

[0002] In a wide range of fields, it is desirable to know the structure, such as the arrangement and bonding state or physical properties of atoms that form a crystal or similar structure. Currently, methods for finding the structure of a compound are being actively researched, and this is an area where higher speed and accuracy are desired. State of the art Non-patent literature

[0003] [Non-patent literature 1] LB Vilhelmsen et al., “A genetic algorithm for first principles global structure optimization of supported nano structures,” J. Chem. Phys., Vol. 141, Issue 4, July 28, 2014 Summary Problem to be solved by the invention

[0004] One of the non-restrictive problems that embodiments of the present disclosure are intended to solve is the search for a desired structure of a substance. Means to solve the problem

[0005] According to one embodiment, an information processing device comprises at least one processor and at least one memory. The at least one processor captures information about a search target from another information processing device, It captures multiple structures of a substance based on information about the search target. updates the recorded majority of the substance's structures, evaluates the updated majority of the substance's structures using a neural network, and transmits information about a sought-after structure of a substance to the other information processing device based on a result of the evaluation. Brief description of the drawings Fig. Figure 1 shows a block diagram schematically representing an information processing device according to one embodiment. Fig. Figure 2 shows a block diagram that schematically represents an information processing system according to one embodiment. Fig. Figure 3 shows a block diagram that schematically represents an information processing system according to one embodiment. Fig. Figure 4 shows a diagram that schematically illustrates the sequence of a process according to one embodiment. Fig. Figure 5 shows a diagram that schematically illustrates a sampling procedure according to one embodiment. Fig. Figure 6 shows a diagram that schematically depicts a sequence of processes according to one embodiment. Fig. Figure 7 shows a diagram illustrating at least one implementation of the information processing device according to an embodiment. Detailed description

[0006] A problem to be solved by embodiments of the present disclosure is not limited to the problem described above and may also be a problem corresponding to effects described in the embodiments as some non-limiting examples of the problem. In other words, the problem corresponding to at least one of the effects described in the explanation of the embodiments in the present disclosure may be the problem to be solved in the present disclosure.

[0007] Embodiments of the present invention are explained below with reference to the drawings. The drawings and the description of the embodiments are given as examples and are not intended to limit the present invention.

[0008] Fig. Figure 1 is a block diagram schematically representing an information processing device according to one embodiment. An information processing device 10 comprises a processing circuit 100, a storage circuit 102, and an input / output interface 104. The information processing device 10 searches for the structure of a connection based on input information and outputs it.

[0009] The processing circuit 100 is a circuit (processor) that implements processes 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.

[0010] The storage circuit 102 stores data required for the processes by the processing circuit 100, and / or data required to activate the information processing device 10.

[0011] The input and output interface 104 is an interface that connects the exterior and interior of the information processing device 10. A user interface may be provided as part of the input and output interface 104.

[0012] In addition to the above, the information processing device may include 10 configurations that are necessary for the information processing device 10 to perform the processes, such as a power supply unit and a control unit, as required.

[0013] Fig. Figure 2 is a diagram that schematically represents an example of an information processing system according to one embodiment. The information processing system 1 comprises the information processing device 10 and a memory 30.

[0014] Memory 30 can store at least some of the data required for the processes by processing circuit 100. Processing circuit 100 can access the data stored in memory 30 via input / output interface 104 and execute processes. Memory 30 can be configured to store data, such as a file server or part of a server deployed in the cloud.

[0015] As shown in this drawing, the storage circuit 102 does not store all the data required for the processes by the processing circuit 100 within the information processing device 10. Instead, the storage 30 provided outside the information processing device 10 can store the data processed by the processing circuit 100 and / or the data acquired in the meantime, and so on, within the processes by the processing circuit 100. As explained above, the information processing system 1 can be configured such that part of the function of the storage circuit 102 exists outside the information processing device 10.

[0016] Fig. Figure 3 is a diagram that schematically represents an example of an information processing system according to one embodiment. An information processing system 1 comprises at least one information processing device 10 and one information processing device 20.

[0017] Information processing device 10 is an information processing device that, for example, operates as a server, and information processing device 20 is an information processing device that, for example, operates as a client.

[0018] The information processing device 20 comprises a processing circuit 200, a storage circuit 202, and an input / output interface 204. The information processing device 20 transmits information to the information processing device 10 and requests the information processing device 10 to search for the structure of a connection. The information processing device 10 performs a search for the structure of the connection based on the information received from the information processing device 20 and the request.

[0019] The processing circuit 200 is a circuit (processor) that executes processes in the information processing device 20. For example, the processing circuit 200 accepts a request from a user via the input and output interface 204 and transmits the request via the input and output interface 204 to the information processing device 10.

[0020] The memory circuit 202 stores data necessary for the operation of the information processing device 20. Furthermore, the memory circuit 202 can also store an operation result of the information processing device 10, which was acquired via the input and output interface 204.

[0021] The input and output interface 204 is an interface that connects the outside and inside of the information processing device 20.

[0022] The input / output interface 104 and the input / output interface 204 can, for example, be connected via a wired or wireless network. In this case, the input / output interface 104 and the input / output interface 204 include a suitable network interface.

[0023] Information processing system 1 can additionally access memory 30 in Fig. 2 and can include one or more information processing devices 10 for one or more information processing devices 20. In the case where the multiple information processing devices 10 are included, a multiple of processors can cooperate to process the requirements of the one or more information processing devices 20.

[0024] In this embodiment, the information processing device 10 can be a cloud-based information processing device and can be a device that provides a service to the information processing device 20 in the form of SaaS (Software as a Service).

[0025] A stable structure has an energy that is lower than that of neighboring structures. A compound that exhibits a lower-energy structure when mixed with a plurality of chemical elements can be said to have a stable (crystal) structure. By referring to a graph representing its state, the composition rate that forms a stable structure can be determined. For example, a structure on a convex hull can be said to have a stable structure.

[0026] In the present disclosure, the search for the structure of a compound by updating the convex hull is a non-restrictive goal. One of the non-restrictive problems that the embodiment of the present disclosure is intended to solve is to search, within the framework of the structure search, for a structure of a substance that has a more energy-stable structure or for a structure that contains a desired physical property value.

[0027] In the present disclosure, the substance refers to various types of materials or the like and is a concept encompassing a compound or a simple substance. For example, the substance may be a crystal of a simple substance, a carbon nanotube, fullerene, nanoparticles, or the like. The structure search for the substance may, for example, refer to the structure of a substance's surface, a molecular structure, or the like. In the case of the simple description of a "structure" below, this can be understood as an atomic structure encompassing an atomic arrangement, an atomic array, or the like of this substance.

[0028] Fig. Figure 4 is a diagram that schematically depicts the sequence of a substance structure search process by the information processing device 10. Hereinafter, this can be described as the process by the information processing device 10 or the information processing device 20, and can be understood, within a consistent range, as the process by the processing circuit 100 or the processing circuit 200.

[0029] The information processing device 10 takes a sample of a structure based on a given condition (S10).

[0030] The information processing device 10 updates (relaxes) the sample-based structure (S20).

[0031] The information processing device 10 evaluates the updated structure (S30).

[0032] The information processing device 10 stores the structure as a history with a rating (S40).

[0033] The information processing device 10 performs a sampling based on the evaluation from the history in iteration (S10 to S40) in or after a second round. Furthermore, the information processing device 10 outputs a structure according to an evaluation value if a predetermined operation completion condition is met (S50).

[0034] Each process in the above processes will be explained in more detail. (Initialization of the process)

[0035] If the information processing device 10 has performed the search by combining the same chemical elements in the past, it can use the result in the search in the past or the information about an intermediate progress as the history.

[0036] In the case of executing a new search according to a combination of chemical elements that has not been executed previously, or executing a new search not according to the previous result over a combination that has been executed previously, the information processing device 10 can consider a structure in which chemical elements are randomly arranged as an initial structure. For example, the information processing device 10 can use as an initial history a plurality of structures in which chemical elements, captured as a condition, are randomly arranged in a cell in a density functional theory (DFT) computation.

[0037] The information processing device 10 can arrange chemical elements by, as a non-restrictive example, dividing a cell into a plurality of subcells, randomly arranging chemical elements in a subcell, copying the subcells, and filling the cell with them. (S10: Sampling of the structure)

[0038] Information processing device 10 can, as a non-restrictive example, use an evolutionary algorithm for sampling the structure. Information processing device 10 performs a sampling of the structure, for example, using a genetic algorithm.

[0039] The genetic algorithm, one of the evolutionary algorithms, selects samples from a plurality of samples, generally a vast number of candidates from the parental generation, and forms a parental generation. Within the samples belonging to the parental generation, crossbreeding and mutations are performed to generate candidate samples for a child generation.

[0040] The wording of the sampling of the structure or simple sampling in the present disclosure refers to the process of generating samples for the offspring generation via the selection of samples in the parent generation and the process which includes at least crossing (mutation may be carried out).

[0041] Fig. Figure 5 is a diagram illustrating an example of sampling according to one embodiment. The information processing device 10 generates the child generation from historical information and executes a subsequent process using the child generation as a sample-based structure.

[0042] The information processing device 10 prepares the generation of the next generation through elite selection using the history as the parent generation (S100). The elite selection is carried out based on the evaluation assigned during the history of the respective sample.

[0043] The information processing device 10 can perform the selection, for example, using an energy value for each structure (sample) that was captured as the rating in the previous iteration. For example, in the case where it is desired to capture a crystal structure composed of two chemical elements, a two-dimensional composition on the convex hull can be selected as an elite sample, with the ratio of the two chemical elements as a horizontal axis and the energy [eV / atom] in the structure in each of the different arrangements for the chemical element ratio as a vertical axis.

[0044] Additionally, the information processing device 10 can also perform the selection based on a multi-target optimization procedure, using not only energy but also an index relating to stability, such as a plurality of physical property values ​​or a desired physical property. In this case, as a non-restrictive example, the information processing device 10 can also perform the selection using other evaluation values ​​that were recorded in the evaluation process at S30, which will be explained later.

[0045] The information processing device 10 can perform elite selection as a non-restrictive example based on a partially ordered sequence using a predetermined index as a sorting key. A partial order is generally a relation that satisfies reflexivity, antisymmetry, and transitivity.

[0046] In one embodiment, p ≥ q with respect to structures p and q is assumed to mean that if E(p) ≤ E(q) and Comp(p) = Comp(q), then “≤” satisfies a partial ordering law. Here, E(p) denotes the energy of structure p or the value of an index of the ordering, and Comp(p) denotes a composition rate of p (for example, a composition ratio of the two chemical elements). Specifically, in other words, the case in which the value of the index E(p) ≤ E(q) with respect to different samples p and q indicates the same composition rate is described as p ≤ q.

[0047] Information processing device 10 calculates a value E, which is an index for each of the structures in the stored samples. Information processing device 10 performs a non-dominant sort on E to extract samples in ascending order of rank. For example, if there are multiple structures with the same chemical element composition rate, information processing device 10 assigns ranks to the structures in ascending order of index and selects the structure as a sample in ascending order of rank.

[0048] The information processing device 10 can furthermore perform selection in ascending order of rank and perform a sample selection using a crowding distance D at a rank that exceeds the number of samples to be selected. The crowding distance D of sample p can be, for example, D(p) = (Manhattan distance of two samples in a p-neighborhood), D(p) = (average of Manhattan distances of a plurality of samples in p-neighborhood), or D(p) = (Manhattan distances of a plurality of samples in p-neighborhood) + (normalized energy gap of p), or the like. The neighborhood specifies other samples that are close in index, for example, by energy value.

[0049] This selection enables the information processing device 10 to perform the elite selection, and the sample thus selected is always an endpoint of the convex hull in a multidimensional space (for example, a space formed with an energy value relative to the composition rate of a plurality of chemical elements). Therefore, it is more likely that the selected sample is one exhibiting a structure that has not improved during a long iteration.

[0050] The information processing device 10 can perform filtering on the sample prior to non-dominant sorting to appropriately select or exclude the sample. The filtering can be expressed, for example, by the following expressions. [Math 1] normalized(E(t))×αg(t)−g*(t)≤ε [Math 2] normalized(E(t))=Emax−E(t)Emax−Emin

[0051] g(t) is a current generation of t, and g*(t) is a generation of t that satisfies the following. [Math 3] argmin{E(s)|s∈history,|r(t)−r(s)|≤δ}

[0052] `history` represents the entire structure (history) that has been searched for up to that point. `r(t)` is a multidimensional vector representing a composition of a structure `t`, and `a`, `ε`, and `δ` are hyperparameters. Furthermore, `|·|` specifies a Manhattan norm of the multidimensional vector. For example, information processing device 10 removes `t` that satisfies expression (1) from the elite selection candidates.

[0053] By performing non-dominant sorting and crowding-distance sorting on the selected candidates, the information processing device 10 can perform elite selection. This selection makes it possible to remove individuals belonging to a region where many structures exist even in the neighborhood—that is, where many individuals with similar genes have already been evaluated—from the candidates, and to preferentially select individuals from a region where the search has not yet been performed as extensively as in the other region, as samples for the next generation.

[0054] After selecting the samples, the information processing device 10 generates the samples for the child generation by the selected samples themselves, crossing a plurality of selected samples and mutations in the selected samples (S102).

[0055] In the case of using the selected samples themselves, the information processing device 10 uses the structure in the course of the parent generation as it is, as a sample of the structure of the child generation. For example, sample 1, which was selected as the elite in the drawing, is considered the structure belonging to the child generation.

[0056] When crossing the samples, the information processing device 10 replaces at least one selected part of the structure of the parent generation with another selected part of the structure of the parent generation. For example, a part of the structure of sample 1 in the drawing is replaced by a part of the structure of sample 3 in a sample in the child generation designated 1+3.

[0057] A crossing can be performed in units of cells, for example in DFT calculations or the like. In a structure specified by a cell in a sample, a crossed sample can be obtained by replacing the structure of a chemical element in at least part of a region of that sample with the structure of the chemical element in part of a region of another sample.

[0058] Furthermore, the intersection is not limited to the above, but can be carried out, for example, by exchanging a part of the structure represented by one graph for a part of the structure of another graph, or by exchanging a part of a structure described in SMILES for another part of the structure described in SMILES.

[0059] During sample mutating, the information processing device 10 changes the type of at least one part of the chemical elements in the structure of a sample, the arrangement of at least one part of the chemical elements, or the combination between at least one part of the chemical elements into another. For example, the structure of sample 4 in the drawing is mutated into a sample designated 4'.

[0060] The information processing device 10 can substitute two different types of chemical elements in the structure, for example in a sample, to effect a mutation. For instance, the information processing device 10 can effect a mutation by moving the positions of four rectangular chemical elements in a sample so that they form a parallelogram. Other examples include the information processing device 10 randomly arranging (adding) or deleting chemical elements at random positions, or randomly changing the types of chemical elements to effect a mutation.

[0061] Additionally, although not shown, both the crossing and the mutation can be performed on two samples to generate a sample in the offspring generation. For example, sample (1+2)' in the offspring generation can be generated from sample 1 and sample 2.

[0062] As explained above, the information processing device 10 performs the sampling of the substance's structure using the genetic algorithm. Of course, the information processing device 10 can perform the sampling in a similar manner using a procedure of another evolutionary algorithm.

[0063] Furthermore, the information processing device 10 can also perform sampling by a method based on a substitute model such as Bayesian optimization, a sampling-based method such as a Markov Chain Monte Carlo (MCMC) method, or a method based on a symmetry of space groups. (S20: Update (Relaxation), S30: Evaluation)

[0064] After sampling, the information processing device 10 updates the sample for each of the sample-based structures so that the structure is established as a stable structure or a structure with a desired physical property value.

[0065] The structure subjected to the above crossing and mutation may not be stable as a crystal structure, for example, if the distance between atoms is too short. To update the generated unstable structure to a stable one, the information processing device 10 performs a relaxation process.

[0066] The information processing device 10 can optimize the sample-based structure locally, for example using NNP (Neural Network Potential).

[0067] The information processing device 10 applies NNP to the sample-based structure to calculate the energy. The information processing device 10 then performs the optimization of the structure to make it locally stable, thus minimizing its energy value. For example, the information processing device 10 can perform the local structure optimization by finding a positional differentiation with respect to energy.

[0068] Through this process, the information processing device 10 updates (relaxes) the sample captured by crossing and / or mutation with a stable structure in the sample's neighborhood. In other words, the information processing device 10 updates (relaxes) the sample to a more stable structure.

[0069] More precisely, the information processing device 10 updates the sample to a sample specifying a point where the gradient of the potential of the NNP becomes 0 by applying a BFGS (Broyden-Fletcher-Goldfarb-Shanno algorithm) or the like, using the target sample as an initial value. The neighborhood can specify a region that includes the point where the gradient becomes 0 and the sample (before the update). The shape of the region varies depending on the shape of a potential curvature surface of the NNP and the position of the sample.

[0070] The information processing device 10 can use any suitable method as the optimization procedure. The information processing device 10 can detect the stable structure from the energy value of the structures in the neighborhood, for example, by locally applying the BFGS method. Additionally, the information processing device 10 relaxes the structure using a hill climbing method, a quasi-Newton method, or the like to update the sample-based structure.

[0071] This process makes it possible to update the structure, which is likely inappropriate, to a more stable structure in the structure captured from the genetic algorithm and to perform a search.

[0072] It should be noted that the information processing device 10 can also evaluate the sample-based structure at S30 in the process at S20. Specifically, the evaluation value recorded for optimization in the process at S20 can be considered an evaluation value at S30.

[0073] If a different index is used than the one used in the local optimization at S20, and the same index is used in the process at S10, the index can be calculated on the sample updated after the process at S20 was completed. Similarly, if the rating used in the process at S20 is part of the rating at S30, the rating recorded during the intermediate step at S20 can be used.

[0074] The information processing device 10 can evaluate the sample-based structure, for example by calculating the energy of each of the relaxed structures and finding the difference between the calculated energy and the energy of a true convex hull.

[0075] The search is generally performed in a situation where the true convex hull is unknown, and therefore the information processing device 10 can calculate the distance from the convex hull with the current energy to the calculated energy of each of the structures, for example, the distance to the structure belonging to the current convex hull, in order to perform the evaluation.

[0076] Furthermore, as another example, the information processing device 10 can also consider the energy value, which is known, for example, published as a database, as the convex hull and use the distance between the convex hull and the convex hull in the sample-based structure as the evaluation.

[0077] Furthermore, as another example, the information processing device 10 can use the area or line length of the convex hull as the total evaluation, or can use the distance between a point specifying a configuration and the convex hull as the evaluation. (S40: save, S50: output)

[0078] The information processing device 10 stores the evaluated result and the evaluated structure in an assignment (S40). The stored structure is registered as a history and can be used as a sample in the parent generation in the next iteration.

[0079] The information processing device 10 repeats processes S10 to S40 until a final condition is met. Similar to general optimization, the final condition can be set such that the number of structures successfully evaluated reaches a predetermined number, the time required for the search exceeds a predetermined time, a predetermined physical property value of the structure being sought reaches a predetermined value, or a predetermined number of iterations are performed. After the final condition is met, the information processing device 10 outputs information about the finally acquired sample (S50).

[0080] The information processing device 10 can, for example, output a structure for the sample that forms the convex hull, and can also output a structure where the distance from the convex hull lies within a predetermined value, for example, within a value of 0.05 [eV / atom]. In this configuration, an error occurs in operations such as NNP, DFT, and the like, and the configuration within the aforementioned predetermined value can indeed be synthesized, and in such a case, the output of the candidate search result is meaningful. Furthermore, when performing a search based on the physical property value, it is conceivable to return the stable or metastable structure where the physical property value falls within the desired range.

[0081] By optimizing and outputting with respect to the convex hull that can be captured for evaluation or sampling, the information processing device 10 can perform a search based on a Pareto front in other variable optimization or a search based on the optimization of a pseudo-Pareto front and output its result.

[0082] The information processing device 10 can output information about the structure, such as the type of chemical element, the positions of atoms, and the cell structure, which are parameters that determine the crystal structure. Furthermore, the information processing device 10 can output energy or physical property values.

[0083] The information processing device 10 can also output the search history. In this case, it is possible to present the user with information about how the search was performed, for example, whether an efficient search was carried out.

[0084] Furthermore, the information processing device 10 can output information to a viewer to visualize the structure. This output can be presented, for example, via the input and output interfaces 104 and 204.

[0085] It should be noted that the information processing device 10 can also output the intermediate progress as appropriate during the execution of the search. The information processing device 10 can also output information not only about the structure, but also about the evaluation value as a graph, or output the state of the convex hull as a graph.

[0086] According to this embodiment, the information processing device 10 can automatically search for the structure of the substance composed of chemical elements, for example, simply by entering conditions such as the types of chemical elements to be used. Therefore, the in Fig. 3. Information processing system 1 depicted: performs a high-speed operation on a server-side only by inputting information about the types of chemical elements (for example, a substance containing Li and In, a crystal containing In and O) from the client-side and provides a service as the SaaS of outputting the search result to a client.

[0087] The operation based on a classical theory of capturing a value, which may be the evaluation value of energy or the like, such as the first principles calculation, a calculation by molecular dynamics or a calculation by density functional theory, requires a long time in the operation with respect to each structure.

[0088] In the case of using the genetic algorithm as a non-restrictive example of sampling, the information processing device 10 may perform the operation a plurality of times for a very long time to evaluate a plurality of samples in the child generation, and may require a time such as several days to several weeks to, for example, capture the evaluation values ​​for a generation in order to capture a stable structure.

[0089] By using the NNP method, the calculation of the evaluation value and the relaxation process of the structure can be implemented quickly and accurately. As a result, according to one embodiment, a global and robust search for the structure of a substance can be realized using an evolutionary algorithm such as a high-speed genetic algorithm.

[0090] In the case of using the genetic algorithm as an example of sampling, optimization can be performed simultaneously for multiple samples. As a result of the ability to increase the number of samples that can be acquired through sampling, high-speed and global structure searches can be achieved. The use of NNP makes it possible to quickly perform the evaluation, which was previously a bottleneck in the search, using multiple samples simultaneously. In one embodiment of the present disclosure, the use of NNP as described above makes it possible to perform the evaluation quickly while ensuring the robustness of the structure search for the substance.

[0091] Using multiple samples simultaneously allows for a more accurate process. Furthermore, the increased evaluation speed achieved through NNP makes it possible to expand the search scope more than before, using multiple accelerators such as general-purpose GPUs, without the need for a supercomputer.

[0092] Furthermore, the information that can be entered as the condition is not limited to the combination of chemical elements. For example, the information processing device 10 can accept information about the environment, such as temperature and pressure, as the conditions. In this case, the information processing device 10 can reflect the environmental information, such as temperature and pressure, in the acquisition of the physical property value in the NNP and can accurately perform the local search for the stable structure under the given conditions. As a result, the information processing device 10 can execute the search result as the entire process as closely as possible to the given conditions.

[0093] The information entered as the conditions can, for example, be entered by the user via the information processing device 20 on the client side.

[0094] The conditions can be set to include the combination of chemical elements and their composition ratio. The composition ratio can be expressed as a range. In this case, the information processing device 10 can execute the process based on the composition ratio (for example, initial sample generation, crossing, mutation) at the time of sampling as a non-restrictive example. If the number of chemical elements is large, the search range can be limited.

[0095] The information processing device 10 can perform sampling, for example determining the range of the composition ratio of Li to In. This sampling enables a search for the structure of the substance in different states, such as a charged state and a discharged state.

[0096] The information processing device 10 can perform sampling, for example, of the range of chemical element presence ratios between a plurality of layers forming a semiconductor. This sampling enables a search for a fixed structure, such as an atomic arrangement between the semiconductor layers. The semiconductor may change in physical property value, even if the number of atoms to be converted is minute, but the search can be performed while covering the change in physical property value.

[0097] The information processing device 10 can perform sampling, where the area in the curve formed by the already detected convex hulls is designated as an example of the search area. This sampling enables a search in the designated area with high granularity.

[0098] The conditions can be set to ranges of physical property values ​​such as energy and density. In this case, the information processing device 10 can perform local optimization with higher accuracy when using NNP. Furthermore, when evaluating using multi-target optimization, a structure closer to the condition can be identified.

[0099] The condition can be defined as the possibility of synthesis for the identified structure. The possibility of synthesis can be determined based on an existing process.

[0100] The evaluation function to be used can be set as a condition.

[0101] The information processing device 10 can output a new structure itself, as explained above. In addition, the information processing device 10 can also output at least one of the energy to be used for evaluation or the state of the convex hull of another variable, a visualization of which parent each structure was captured from (including any change over time), the optimization process, and the physical property value of the structure (crystal group, cell structure).

[0102] According to the above embodiment, the calculation of the physical property value for evaluation using NNP can be performed quickly, and therefore the number of individuals in a 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 becomes possible to improve the operational speed and accuracy with a wider search range.

[0103] In the case of using the configuration in Fig. 3. Information processing device 10 can perform the search using the evolutionary algorithm, and information processing device 20 can perform the evaluation. Furthermore, as another example, information processing device 10 can perform the search using the evolutionary algorithm, and another information processing device 10 can perform the evaluation. As explained above, each process can be performed in a suitable information processing device as needed.

[0104] Fig. Figure 6 is a diagram illustrating one embodiment for evaluating the result output in the embodiment described above. The information processing device 10 in Fig. 1, Fig. 2 to Fig. 3 or the information processing device 20 in Fig. 3 can evaluate the result of the search completion (S60).

[0105] For example, the information processing device 10 or the information processing device 20 can use the structure of the sample output as the search result as a candidate of the structure to be captured, in which case a most desirable structure can be extracted from the candidates by an evaluation on a plurality of candidate structures that were captured in the search.

[0106] For example, in addition to the evaluation at S30, the physical property value, such as energy, can be determined through a classical DFT operation. The evaluation calculates the energy value, or similar, for the candidate, which was searched for at high speed using NNP, using a different operational procedure. This makes it possible to increase the accuracy of detecting the desired structure.

[0107] Furthermore, if the known convex hull is used as the target of the evaluation, the information processing device 10 or the information processing device 20 can calculate the energy value, or similar value, for the structure belonging to the known convex hull using a classical method or the method using NNP. By acquiring the energy value, or similar value, of the existing structure as described above, it becomes possible to confirm that the evaluation in the information processing device 10 was performed correctly.

[0108] The information processing device 10 acquires at least the information about the search target from the information processing device 20.

[0109] The information about the search target presents the information that was captured by the information processing device 20 used by the user.

[0110] The information about the search target can include information such as the types and composition ratios of chemical elements that constitute the substance described in the embodiment. The information is not limited to this, but can also include information specifying the substance that is the search target and information specifying search conditions (temperature, pressure, convex hull area, range of physical property values, and so on). Furthermore, the information processing device 10 can acquire the information about the search target from the information processing device 20 via another device.

[0111] The information processing device 10 detects a plurality of structures of a substance based on the information about the detected search target and updates the detected plurality of structures of the substance.

[0112] Capturing the majority of the substance's structures can be a concept that includes at least one of the following.

[0113] Specifically, the information processing device 10 can detect the majority of the substance's structures by sampling.

[0114] Furthermore, the information processing device 10 can detect the majority of the substance's structures through a random arrangement.

[0115] Furthermore, the information processing device 10 can detect the majority of the substance's structures based on the sequence information. Even when performing elite selection using the sequence, the information processing device 10 can identify at least one of the candidates to be selected as the substance's structure generated by the random arrangement. This allows a global search to be integrated into a local search.

[0116] Furthermore, the information about the structure of the substance can be the concept that includes at least one of the following.

[0117] Specifically, the information about the structure of the substance can include information about the type of atom (chemical element) and the positional information about the atom (chemical element).

[0118] Furthermore, the information about the structure of the substance can include information about the cell structure.

[0119] It should be noted that the information processing device 10 may be configured to capture at least one structure of the substance based on the information acquired about the search target and to update the at least one captured structure of the substance.

[0120] Information processing device 10 can evaluate the updated plurality of structures of the substance using the neural network (for example, NNP) and transmit the information about the desired structure of the substance to information processing device 20 based on the evaluation result. Furthermore, information processing device 10 can transmit the information about the structure of the substance sought by information processing device 20 via another device.

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

[0122] It is not excluded that NNP is formed using one neural network model as well as a neural network that forms a graph based on a so-called atomic arrangement, and that the structure is evaluated by the other neural network model. In other words, the NNP in the present disclosure does not preclude the use of the neural network model for capturing information about the structure of the substance as a broader concept.

[0123] The information processing device 10 can update the detected majority of structures of the substance using the neural network.

[0124] This neural network can be a neural network of the above NNP or it can be a different model. It should be noted that the information processing device 10 can be set up to evaluate the information about the updated at least one structure of the substance using the neural network.

[0125] Furthermore, the information processing device 10 can be configured to transfer the information about the one or more structures of the substance being sought to the information processing device 20.

[0126] A plurality of information processing devices 20 may be provided, as also explained above.

[0127] Similarly, the information processing device 10 can be set up as a system that includes a plurality of 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 also include one or more storage devices 30.

[0129] In Information Processing System 1, these configurations are connected by a network link using a wireless or wired means, which typically includes the Internet, but the connection method is not limited to this, and any means capable of properly connecting them may be used.

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

[0131] Some or all of each device (information processing device 10 or information processing device 20) in the above embodiment may be configured as hardware or as software (program) information processing, executed, for example, by a CPU (Central Processing Unit) or GPU (Graphics Processing Unit). In the case of software information processing, software enabling at least some of the functions of each device in the above embodiments may be stored on a non-volatile storage medium (non-volatile computer-readable medium) such as CD-ROM (Compact Disc Read Only Memory) or USB (Universal Serial Bus) storage, and the software information processing may be performed by loading the software onto a computer. Additionally, the software may also be downloaded via a communication network.Furthermore, all or part of the software can be implemented in a circuit such as an ASIC (Application Specific Integrated Circuit) or FPGA (Field Programmable Gate Array), whereby the information processing of the software can be performed by hardware.

[0132] A storage medium for the software can be a removable storage medium, such as an optical disc, or a fixed storage medium, such as a hard drive or memory. The storage medium can be located inside the computer (in a main memory device or an auxiliary storage device) or outside the computer.

[0133] Fig. Figure 7 is a block diagram illustrating an example of a hardware configuration for each device (the information processing device 10 or the information processing device 20) in the embodiments described above. As an example, each device may be implemented as a computer 7 comprising a processor 71, a main memory device 72, an auxiliary memory device 73, a network interface 74, and a device interface 75, all connected via a bus 76.

[0134] The computer 7 from Fig. 7 is equipped with each component individually, but can be equipped with multiple of the same components. Although a computer 7 in Fig.As shown in Figure 7, the software can be installed on a plurality of computers, and each of the plurality of computers can perform the same or a different part of the software processing. In this case, it can be in a form of distributed computing, with each of the computers communicating with each other, for example, via the network interface 74, to perform the processing. That is, each device (the information processing device 10 or the information processing device 20) in the above embodiments can be set up as a system in which one or more computers execute the instructions stored in one or more memories to enable functions.Each device can be configured so that information transmitted by an end device is processed by one or more computers provided in a cloud, and the results of the processing are transmitted back to the end device.

[0135] Various arithmetic operations of each device (the information processing device 10 or the information processing device 20) in the above embodiments can be performed in parallel processing using one or more processors or using a plurality of computers over a network. The various arithmetic operations can be assigned to a plurality of arithmetic cores in the processor and performed in parallel processing. Some or all of the processes, means, or the like of this disclosure can be implemented by at least one of the processors or the storage devices provided on a cloud that can communicate with the computer 7 over a network. Thus, each device in the above embodiments can be in a form of parallel computing by one or more computers.

[0136] The processor 71 can be an electronic circuit (such as a processor, processing circuit, processing circuits, CPU, GPU, FPGA, or ASIC) that performs at least the control of the computer or arithmetic calculations. The processor 71 can also be, for example, a general-purpose processing circuit, a dedicated processing circuit designed to perform specific operations, or a semiconductor device that incorporates both the general-purpose and dedicated processing circuits. Furthermore, the processor 71 can also include, for example, an optical circuit or an arithmetic function based on quantum computing.

[0137] Processor 71 can perform arithmetic operations based on data and / or software input from, for example, any device in the internal configuration of Computer 7, and can output an arithmetic result and a control signal to, for example, any device. Processor 71 can control any component of Computer 7, for example, by running an operating system or application of Computer 7.

[0138] Each device (the information processing device 10 or the information processing device 20) in the above embodiments can be enabled by one or more processors 71. The processor 71 can refer to one or more electronic circuits arranged on a single chip, or to one or more electronic circuits arranged on two or more chips or devices. In the case where multiple electronic circuits are used, each electronic circuit can communicate via wired or wireless connections.

[0139] The main memory device 72 can, for example, store instructions to be executed by the processor 71 or various data, and the information stored in the main memory device 72 can be read by the processor 71. The auxiliary memory device 73 is a different type of memory device than the main memory device 72. These memory devices are intended to refer to any electronic component capable of storing electronic information and can be semiconductor memory. The semiconductor memory can be either volatile or non-volatile.The storage device for storing various data or the like in each device (the information processing device 10 or the information processing device 20) in the above embodiments can be provided by the main memory device 72 or the auxiliary memory device 73, or it can be implemented by built-in memory integrated into the processor 71. For example, the memory circuit 102 or the memory circuit 202 in the above embodiments can be implemented in the main memory device 72 or the auxiliary memory device 73.

[0140] In the case where each device (the information processing device 10 or the information processing device 20) in the above embodiments is equipped by at least one storage device (memory) and at least one processor connected / coupled to that at least one storage device, the at least one processor may be connected to a single storage device. Or the at least one memory may be connected to a single processor. Or each device may comprise a configuration in which at least one of the plurality of processors is connected to at least one of the plurality of storage devices. Furthermore, this configuration may be implemented by a storage device and a processor contained in a plurality of computers.Furthermore, each device can include a configuration in which a storage device is integrated into a processor (for example, a cache memory that includes an L1 cache or an L2 cache).

[0141] Network interface 74 is an interface for wireless or wired connection to a communication network 8. Network interface 74 can be a suitable interface, such as one compatible with existing communication standards. Information can be exchanged via network interface 74 with an external device 9A connected through the communication network 8. It should be noted that the communication network 8 can be configured, for example, as a WAN (Wide Area Network), LAN (Local Area Network), PAN (Personal Area Network), or a combination thereof, and can be such that information can be exchanged between the computer 7 and the external device 9A. The Internet is an example of a WAN, IEEE 802.11 or Ethernet (registered trademark) is an example of LAN, and Bluetooth (registered trademark) or NFC (Near Field Communication) is an example of PAN.

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

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

[0144] The external device 9A or the external device 9B can, for example, be an input device. The input device is, for example, a device such as a camera, a microphone, a motion capture device, at least one of several sensors, a keyboard, a mouse, or a touch panel, and it transmits the captured information to the computer 7. Furthermore, it can be a device that includes an input unit, such as a personal computer, a tablet device, or a smartphone, which may have an input unit, memory, and a processor.

[0145] The external device 9A or the external device 9B can, for example, be an output device. The output device can be, for example, a display device, such as an LCD (Liquid Crystal Display) or an organic EL (Electroluminescent) panel, or a loudspeaker that outputs audio. It can also be a device that includes an output unit, such as a personal computer, a tablet device, or a smartphone, which may have an output unit, memory, and a processor.

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

[0147] Furthermore, the external device 9A or the external device 9B can be a device that has at least one function of the configuration element of each device (the information processing device 10 or the information processing device 20) in the embodiments described above. That is, the computer 7 can transmit some or all of the processing results to the external device 9A or the external device 9B, or receive some or all of the processing results from the external device 9A or the external device 9B.

[0148] In the present specification (including the claims), the representation (including similar expressions) of "at least one of a, b and c" or "at least one of a, b or c" includes all combinations of a, b, c, a - b, a - c, b - c and a - b - c. It also covers combinations with multiple instances of any element, such as a - a, a - b - b or a - a - b - b - c - c. It further covers, for example, the addition of another element d beyond a, b and / or c, such as a - b - c - d.

[0149] In the present specification (including the claims), expressions such as "data as input", "using data", "based on data", "according to data" or "in accordance with data" (including similar expressions) are used, unless otherwise specified, this includes cases where data itself is used or cases where data is processed in any way (for example, data with added noise, normalized data, feature sizes extracted from the data or intermediate representation of the data).When it is stated that some results can be obtained “by inputting data”, “by using data”, “based on data”, “according to data”, “in accordance with data” (including similar expressions), unless otherwise specified, this may include cases where the result is obtained solely based on the data, and may also include cases where the result is obtained by being influenced by factors, conditions and / or states or the like through data other than the data.Unless otherwise specified, when “output / output of data” (including similar expressions) is stated, this includes cases where the data itself is used as output, or cases where the data is processed in some way (for example, the data with added noise, the normalized data, feature size extracted from the data, or intermediate representation of the data) as the output.

[0150] When terms such as "connected (connection)" and "coupled (coupling)" are used in this specification (including the claims), they are intended as non-restrictive terms that include any of the following: "direct connection / coupling," "indirect connection / coupling," "electrical connection / coupling," "communicative connection / coupling," "operational connection / coupling," "physical connection / coupling," or the like. The terms should be interpreted accordingly, depending on the context in which they are used, but all forms of connection / coupling not intentionally or naturally excluded should be considered included in the terms and interpreted in a non-exclusive manner.

[0151] In the present specification (including the claims), when the expression "A is configured to do B" is used, it can mean that a physical structure of A has a configuration capable of performing operation B, and that a permanent or temporary setting / configuration of element A is configured / set to actually perform operation B. For example, if element A is a general-purpose processor, the processor may have a hardware configuration capable of performing operation B and may be configured to actually perform operation B by setting the permanent or temporary program (instructions).Furthermore, if element A is a dedicated processor, a dedicated arithmetic circuit, or the like, a circuit structure of the processor or the like may be implemented to actually perform operation B, regardless of whether control instructions and data are actually attached to it or not.

[0152] Whenever a term relating to inclusion or possession (for example, "comprising / encompassing," "possessing," or the like) is used in this specification (including the claims), it is to be understood as an open term that includes the case of including or possessing an object different from the object specified by the term. If the object of such terms implying inclusion or possession is an expression that does not specify a quantity or suggests a singularity (an expression with an indefinite article such as "a / an"), the expression should be interpreted as not being limited to a specific number.

[0153] In the present specification (including the claims), even though the expression "one or more," "at least one," or the like is used in some places, and the expression that does not specify a quantity or suggest a singularity (the expression with an indefinite article such as "one") is used elsewhere, it is not intended that this expression means "one." In general, the expression that does not specify a quantity or suggest a singularity (the expression with an indefinite article such as "one") should be interpreted as not necessarily being limited to a specific number.

[0154] If the present specification states that a particular configuration of an example leads to a certain effect (advantage / result), provided there are no other reasons, it should be understood that the effect is also obtained for one or more other embodiments with the same configuration. However, it should be understood that the presence or absence of such an effect generally depends on various factors, conditions, and / or states, etc., and that such an effect is not always achieved by the configuration. The effect is only achieved by the configuration in the embodiments when various factors, conditions, and / or states, etc., are met, but the effect is not always obtained in the claimed invention that defines the configuration or a similar configuration.

[0155] In this specification (including the claims), when the term "maximize" is used as in "maximize," it includes finding a global maximum value, finding an approximate value of the global maximum value, finding a local maximum value, and finding an approximate value of the local maximum value. It should be interpreted appropriately depending on the context in which the term is used. It also includes probabilistically or heuristically finding the approximate value of these maximum values. Similarly, when the term is used as in "minimize," it includes finding a global minimum value, finding an approximate value of the global minimum value, finding a local minimum value, and finding an approximate value of the local minimum value. It should be interpreted appropriately depending on the context in which the term is used.It also includes the probabilistic or heuristic finding of the approximate value of these minimum values. Similarly, when the term is used as "optimize / optimization," it includes finding a global optimal value, finding an approximate value of the global optimal value, finding a local optimal value, and finding an approximate value of the local optimal value, and should be interpreted appropriately depending on the context in which the term is used. It also includes the probabilistic or heuristic finding of the approximate value of these optimal values.

[0156] If, in this specification (including the claims), a plurality of hardware performs a predetermined process, the respective hardware may cooperate to perform the predetermined process, or some hardware may perform the entire predetermined process. Furthermore, some hardware may perform part of the predetermined process, and the other hardware may perform the remainder of the predetermined process. If, in this specification (including the claims), an expression (including similar expressions) such as "one or more hardware performs a first process and one or more hardware performs a second process" or the like is used, the hardware performing the first process and the hardware performing the second process may be the same hardware or may be different hardware.This means that the hardware performing the first process and the hardware performing the second process can be contained within one or more pieces of hardware. It should be noted that the hardware can include an electronic circuit, a device containing the electronic circuit, or the like.

[0157] If, in the present specification (including the claims), a plurality of storage devices (memories) store data, an individual storage device among the plurality of storage devices may store only a portion of the data or may store all of the data. Furthermore, some storage devices among the plurality of storage devices may include a configuration for storing data.

[0158] While certain 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, partial deletions, etc., are possible, provided they do not deviate from the technical teaching and purpose of the present disclosure, which are derived from the content specified in the claims and their equivalents. For example, where numerical values ​​or mathematical formulas are used in the description of the embodiments described above, they are shown for illustrative purposes only and do not limit the scope of the present disclosure. Furthermore, the sequence of each operation shown in the embodiments is also an example and does not limit the scope of the present disclosure.

[0159] The above embodiments can be summarized or extended as follows. (1) Information processing device comprising: at least one processor; and at least one storage device, where at least one processor Information about a search target is captured by another information processing device, a plurality of structures of a substance based on information about the search target, the recorded majority of the substance's structures updated, the updated majority of the substance's structures were evaluated using a neural network, and Information about a sought-after structure of a substance is transferred to the other information processing device based on a result of the evaluation. (2) Information processing device according to (1), wherein which at least one processor updates the detected majority of structures of the substance using the neural network. (3) Information processing device according to (2), wherein which at least one processor performs the update by converting the detected majority of structures of the substance into stable structures in neighborhoods. (4) Information processing device according to (1), wherein which at least one processor evaluates the updated plurality of structures of the substance based on a multi-target optimization procedure. (5) Information processing device according to (4), wherein which at least one processor captures the majority of structures of the substance based on a Pareto front. (6) Information processing device according to (5), wherein which at least one processor captures the majority of the substance's structures using an evolutionary algorithm. (7) Information processing device according to one of (1) to (6), wherein where at least one processor repeatedly performs the acquisition, updating, and evaluation until a predetermined condition is met. (8) Information processing device according to (7), wherein where at least one processor performs the acquisition, updating, and evaluation in parallel. (9) Information processing device according to one of (1) to (8), wherein The information about the search target must contain information about at least either one type of chemical element or a composition ratio of chemical elements. (10) Information processing device according to (9), wherein The information about the search target must also include information about at least one of a temperature, a pressure, an area of ​​a convex hull, and an area of ​​a physical property value. (11) Information processing device according to one of (1) to (10), wherein which transmits at least one storage device information about convex hulls of a plurality of sought substances to the other information processing device. (12) Information processing device according to any of (1) to (11) wherein the neural network is a Neural Network Potential (NNP). (13) Information processing device according to (12), wherein which at least one processor evaluates the updated majority of structures of the substance using energy. (14) Information processing device according to (12) or (13), wherein at least one memory evaluates the sought-after structure of the substance using a density functional theory (DFT) calculation. (15) Information processing system, comprising: the information processing device according to one of (1) to (14); and at least the other information processing device. (16) Information processing system according to (15), wherein the neural network is a Neural Network Potential (NNP). (17) Information processing system according to (16), wherein at least either the information processing device or the other information processing device evaluates the sought-after structure of the substance using a density functional theory (DFT) calculation. (18) Information processing techniques, comprehensive: at least one processor Retrieving information about a search target from another information processing device, Capturing a plurality of structures of a substance based on information about the search target, Updating the captured majority of substance structures, Evaluating the updated majority of the substance's structures using a neural network, and Transferring information about a sought-after structure of a substance to another information processing device based on a result of the evaluation. (19) Information processing method according to (18) wherein at least part of the information processing device according to one of (1) to (17) is executed by the at least one processor. [Explanation of reference symbols] 1 Information processing system, 10 Information processing device, 100 processing circuits, 102 memory circuit, 104 Input and output interface, 20 Information processing device, 200 processing circuits, 202 Memory circuit, 204 Input and output interface, 30 storage QUOTES INCLUDED IN THE DESCRIPTION

[0000] This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited non-patent literature

[0000] LB Vilhelmsen et al., “A genetic algorithm for first principles global structure optimization of supported nano structures”, J. Chem. Phys., Vol. 141, Issue 4, July 28, 2014

[0003]

Claims

[1] Information processing device, comprising: at least one storage device and at least one processor configured to: Retrieving information about a search target from another information processing device; Capturing multiple structures of a substance based on information about the search target; Updating the captured majority of substance structures; Evaluating the updated majority of substance structures using a neural network; and Transferring information about a sought-after structure of a substance to another information processing device based on a result of the evaluation. [2] Information processing device according to claim 1, wherein the at least one processor updates the detected plurality of structures of the substance using the neural network. [3] Information processing device according to claim 2, wherein the at least one processor performs the update by converting the detected plurality of structures of the substance into stable structures in neighborhoods. [4] Information processing device according to claim 1, wherein the at least one processor evaluates the updated plurality of structures of the substance based on a method of multiple target optimization. [5] Information processing device according to claim 4, wherein the at least one processor detects the plurality of structures of the substance based on a Pareto front. [6] Information processing device according to claim 5, wherein the at least one processor detects the plurality of structures of the substance using an evolutionary algorithm. [7] Information processing device according to claim 1, wherein the at least one processor repeatedly performs the acquisition, updating and evaluation until a predetermined condition is met. [8] Information processing device according to claim 7, wherein the at least one processor performs the acquisition, updating and evaluation in parallel. [9] Information processing device according to claim 1, wherein the information about the search target includes information about at least either one type of chemical element or a composition ratio of chemical elements. [10] Information processing device according to claim 9, wherein the information about the search target further includes information about at least one of a temperature, a pressure, an area of ​​a convex hull and an area of ​​a physical property value. [11] Information processing device according to claim 1, wherein the at least one memory transmits information about convex hulls of a plurality of substances sought to the other information processing device. [12] Information processing device according to any one of claims 1 to 11, wherein the neural network is a Neural Network Potential, NNP. [13] Information processing device according to claim 12, wherein the at least one processor evaluates the updated plurality of structures of the substance by means of energy. [14] Information processing device according to claim 12, wherein the at least one memory evaluates the sought-after structure of the substance using a density functional theory (DFT) calculation. [15] Information processing system, including: the information processing device according to any one of claims 1 to 11; and at least one other information processing device. [16] Information processing system according to claim 15, wherein the neural network is a Neural Network Potential, NNP. [17] Information processing system according to claim 16, wherein at least either the information processing device or the other information processing device evaluates the desired structure of the substance using a density functional theory (DFT) calculation. [18] Information processing techniques, comprehensive: Acquisition, by at least one processor, of information about a search target from another information processing device, Detection, by which at least one processor, a plurality of structures of a substance are based on information about the search target, Update, by means of at least one processor, the detected majority of structures of the substance, Evaluate, by means of at least one processor, the updated plurality of structures of the substance using a neural network, and Transferred, by at least one processor, of information about a sought-after structure of a substance to the other information processing device based on a result of the evaluation.

Images