Crystal structure representation method, computer device and storage medium

Abstract

The invention discloses a representation method of a crystal structure, computer equipment and a storage medium, and relates to the technical field of computational materiation.The method comprises the steps that neighbor atoms corresponding to all atoms are determined according to the initial spatial positions of all the atoms extracted from an initial representation file, and then the atom pair distances between all the atoms and the neighbor atoms of all the atoms are determined; and carrying out dimension reduction processing on the initial spatial position of each atom by utilizing the distance of each atom pair, so as to obtain the target spatial position of each atom, thereby representing the crystal structure by utilizing the target spatial position of each atom, and generating the target representation of the crystal structure. According to the method, the dimension of the initial spatial position of each atom is reduced through the atom pair distance between each atom and the adjacent atom, so that the high-dimensional initial spatial position can be mapped into low-dimensional, continuous and comparable vector representation, and the geometric similarity and symmetry between crystal structures are kept; therefore, the problem that the effect of expressing the crystal structure is poor is solved.

Description

Technical Field

[0001] This application relates to the technical field of computational materials science, and particularly to a method for representing crystal structures, a computer device, and a storage medium. Background Art

[0002] With the proposal of the concept of materials genome engineering, machine learning has gradually been incorporated into materials science research. Using machine learning, researchers can deeply explore the band structure, charge distribution, stability, and various physical and chemical properties of materials from the atomic scale, providing a solid theoretical support for the design and performance prediction of new materials.

[0003] For example, carbon materials, due to their rich structural forms and unique physical and chemical properties, show broad application prospects in the fields of energy storage, electronic devices, catalysis, and biomedicine. For carbon crystal structures, such as graphite, diamond-like carbon, porous carbon, carbon nanotubes, fullerenes, etc., their atomic arrangements are highly complex and diverse. By training on the structure and property data of carbon crystal structures, it is possible to achieve the modeling of structure-property relationships, the prediction of material properties, and reverse design. However, this method poses relatively high requirements for the representation of carbon crystal structures, and the input features need to be able to accurately express the essence of carbon crystal structures and be convenient for the understanding and processing of machine learning algorithms.

[0004] However, traditional methods for representing crystal structures are based on lattices and atomic coordinates. Although they can describe all the geometric information of crystal structures, they lack the internal structural information of crystals, which is not conducive to directly comparing, modeling, and searching crystal structures. Summary of the Invention

[0005] This application provides a method for representing crystal structures, a computer device, and a storage medium to at least solve the problem of poor representation effect of crystal structures.

[0006] This application provides a method for representing crystal structures, including: obtaining an initial representation file of a crystal structure, and extracting the initial spatial positions of each atom in the initial representation file; using the initial spatial positions of each atom to determine the corresponding neighboring atoms of each atom and the atomic pair distances between each atom and its neighboring atoms; using each atomic pair distance to perform dimensionality reduction processing on each initial spatial position to obtain the target spatial positions corresponding to each atom and its neighboring atoms; using each target spatial position to represent the crystal structure and generating a target representation of the crystal structure.

[0007] This application also provides a computer device, including: a memory for storing a computer program; a processor for implementing the steps of any of the above methods for representing crystal structures when executing the computer program.

[0008] The present application also provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the steps of any of the above-mentioned crystal structure representation methods are implemented.

[0009] The present application also provides a computer program product including a computer program, and when the computer program is executed by a processor, the steps of any of the above-mentioned crystal structure representation methods are implemented.

[0010] Through the crystal structure representation method of the present application, using the initial spatial positions of each atom extracted from the initial representation file, the neighboring atoms corresponding to each atom are determined, and then the atomic pair distances between each atom and its neighboring atoms are determined; using each atomic pair distance, dimensionality reduction processing is performed on the initial spatial positions of each atom, and the target spatial positions of each atom can be obtained, so that the crystal structure can be represented using the target spatial positions of each atom, and a target representation of the crystal structure is generated. Through the atomic pair distances between each atom and its neighboring atoms, the present application performs dimensionality reduction on the initial spatial positions of each atom, which can not only map the high-dimensional initial spatial positions into low-dimensional, continuous and comparable vector representations, but also retain the geometric similarity and symmetry between crystal structures, thus solving the problem of poor representation effect of crystal structures. Further, it breaks through the limitations of traditional high-dimensional discrete structure representations in structure search, performance modeling and computational efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0012] Figure 1 A schematic diagram of a crystal structure representation method provided by an embodiment of the present application; Figure 2 A schematic diagram of another crystal structure representation method provided by an embodiment of the present application; Figure 3 A schematic diagram of yet another crystal structure representation method provided by an embodiment of the present application; Figure 4 A schematic diagram of generating a target representation of a crystal structure based on manifold learning provided by an embodiment of the present application; Figure 5 A method for representing a carbon crystal structure and representing a carbon crystal structure based on manifold learning by taking the carbon crystal structure as an example provided by an embodiment of the present application; Figure 6 A schematic diagram of the structure of a crystal structure representation device provided by an embodiment of the present application; Figure 7 A schematic structural diagram of a computer device provided by an embodiment of the present application. Specific implementation manners

[0013] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0014] It should be noted that in the description of the present application, the terms "include", "comprise" or any other variant thereof are intended to cover a non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. The terms "first", "second", etc. in the present application are used to distinguish similar objects and are not used to describe a specific order or sequence.

[0015] In order to enable those skilled in the art of the present technology to better understand the solution of the present application, the present application will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners.

[0016] Combined with the specific application environment architecture or specific hardware architecture on which the execution of the representation method of the crystal structure depends, the specific application environment architecture or specific hardware architecture is described herein.

[0017] With the proposal of the concept of materials gene engineering, machine learning has gradually been integrated into materials science research. Using machine learning, researchers can deeply explore the energy band structure, charge distribution, stability, and various physical and chemical properties of materials from the atomic scale, providing a solid theoretical support for the design and performance prediction of new materials.

[0018] For example, carbon materials have shown broad application prospects in the fields of energy storage, electronic devices, catalysis, and biomedicine due to their rich structural forms and unique physical and chemical properties. For carbon crystal structures, such as graphite, diamond-like carbon, porous carbon, carbon nanotubes, fullerenes, etc., their atomic arrangements are highly complex and diverse. By training the structure and property data of carbon crystal structures, it is possible to realize the modeling of the structure-property relationship, material property prediction, and inverse design. However, this method places high requirements on the representation of carbon crystal structures, and the input features need to be able to accurately express the essence of carbon crystal structures and be convenient for the understanding and processing of machine learning algorithms.

[0019] However, the traditional representation method of crystal structures is based on lattices and atomic coordinates. Although it can describe all the geometric information of crystal structures, it lacks the internal structure information of crystals, which is not conducive to directly comparing, modeling, and searching crystal structures. Especially in intelligent modeling tasks such as machine learning, it is not input-friendly, restricting the further development of materials informatization and intelligent research. In recent years, deep learning techniques such as Graph Neural Network (GNN), AutoEncoder, Variational Autoencoder (VAE), Diffusion Model, and Transformer (a deep learning architecture) have been gradually introduced into the material structure modeling task, and important progress has been made in compound generation, property prediction, and structure search. However, the representation methods of crystal structures mostly rely on graph structures and do not have good geometric interpretability, especially performing poorly in irregular structures such as amorphous and porous structures. Among them, the representation method of crystal structures based on graph neural networks is a way to model crystal structures as graphs, with atoms as nodes of the graph and the interactions between atoms as edges, and combines graph neural networks to automatically extract structure features. This method can effectively capture local and global atomic environment information, but there are still some limitations. For example, graph neural networks have limited ability to handle long-range interactions (such as van der Waals forces or charge transfer), and can often only capture the structure features within the local neighborhood. In addition, it is difficult to directly retain or express the symmetry information between different structures, which may lead to inaccurate prediction of physical properties sensitive to crystal symmetry. Moreover, graph neural networks are sensitive to data quality and training scale, with high training costs and poor interpretability.

[0020] In view of this, the present application proposes a representation method of crystal structures, a computer device, and a storage medium. The method includes: obtaining an initial representation file of a crystal structure and extracting the initial spatial positions of each atom in the initial representation file; based on the initial spatial positions of each atom, determining the neighboring atoms corresponding to each atom and the atomic pair distances between each atom and its neighboring atoms; using the atomic pair distances to perform dimensionality reduction processing on the initial spatial positions to obtain the target spatial positions corresponding to each atom; using the target spatial positions to represent the crystal structure and generating a target representation of the crystal structure.

[0021] The method for representing the crystal structure of the present application determines the neighboring atoms corresponding to each atom by using the initial spatial positions of each atom extracted from the initial representation file, and then determines the atomic pair distance between each atom and its neighboring atoms; by using the atomic pair distances of each atom, the initial spatial positions of each atom are subjected to dimensionality reduction processing, and the target spatial positions of each atom can be obtained, so that the crystal structure can be represented by using the target spatial positions of each atom, and the target representation of the crystal structure is generated. By using the atomic pair distances between each atom and its neighboring atoms, the present application reduces the dimension of the initial spatial positions of each atom, which can not only map the high-dimensional initial spatial positions into low-dimensional, continuous and comparable vector representations, but also retain the geometric similarity and symmetry between crystal structures, thus solving the problem of poor representation effect of crystal structures. Further, it breaks through the limitations of traditional high-dimensional discrete structure representations in structure search, performance modeling and computational efficiency.

[0022] In this embodiment, a method for representing a crystal structure is provided, which can be used in a server, a personal computer, a cloud platform, etc. Figure 1 It is a flowchart of the method for representing a crystal structure according to an embodiment of the present application, as Figure 1 shown, and this process includes the following steps.

[0023] Step S101: Obtain the initial representation file of the crystal structure, and extract the initial spatial positions of each atom in the initial representation file.

[0024] The initial representation file of the crystal structure can be a standardized record file of the crystal structure. As a specific example, the file format of the initial representation file can be CIF (Crystallographic Information File), POSCAR or XYZ. In addition, the initial representation file of the crystal structure can be directly input into a computer device so that the computer device can obtain the initial representation file of the crystal structure.

[0025] The initial spatial positions of each atom can be three-dimensional spatial coordinates or spatial coordinates of other dimensions, and the present application does not limit this. As a specific example, the process of extracting the three-dimensional spatial coordinates of each atom in the initial representation file can be: parsing the file format of the initial representation file to obtain the target file format of the initial representation file; obtaining the preset extraction strategy corresponding to the target file format; using the preset extraction strategy to extract the three-dimensional spatial coordinates of each atom from the initial representation file. For example, if the target file format is POSCAR, the first part of the initial representation file is generally 3 The unit cell parameter matrix of 3, and the second part of the content is the three-dimensional spatial coordinates of each atom. Such a preset extraction strategy can extract the data information after the unit cell parameter matrix to obtain the three-dimensional spatial coordinates of each atom. Another example is that if the target file format is CIF, the first part of the initial representation file generally corresponds to the specific values of the unit cell parameters through the unit cell parameter names, and the second part of the content is the three-dimensional spatial coordinates of each atom. Such a preset extraction strategy can extract the specific values of the unit cell parameters corresponding to the unit cell parameter names to obtain the three-dimensional spatial coordinates of each atom.

[0026] As a specific example, the crystal structure can be a carbon crystal structure or other crystal structures, and the present application does not limit this.

[0027] Step S102: Using the initial spatial positions of each atom, determine the nearest neighbor atoms corresponding to each atom and the atomic pair distances between each atom and its nearest neighbor atoms.

[0028] For the determination of the nearest neighbor atoms of each atom, methods such as the distance threshold method, the coordination environment method, the polyhedron method, or the triangulation method can be used for determination, and the present application does not limit this.

[0029] The atomic pair distance can be the distance between an atom and its nearest neighbor atom. For the atomic pair distances between each atom and its nearest neighbor atoms, they can be directly calculated through the initial spatial positions between each atom and its nearest neighbor atoms. Of course, it is also possible to process the initial spatial positions between each atom and its nearest neighbor atoms and then calculate the atomic pair distances.

[0030] Step S103: Use each atomic pair distance to perform dimensionality reduction processing on each initial spatial position to obtain the target spatial positions corresponding to each atom and its nearest neighbor atoms.

[0031] During the process of performing dimensionality reduction processing on each initial spatial position, the goal can be to minimize the difference between the atomic pair distances of each atom and its nearest neighbor atoms in the low-dimensional space and in the high-dimensional space, and perform dimensionality reduction processing on each initial spatial position to obtain the target spatial positions of each atom and its nearest neighbor atoms in the low-dimensional space. For example, if the initial spatial position is 3-dimensional, through dimensionality reduction processing, the initial spatial position can be reduced to 2-dimensional, and each atom and its nearest neighbor atoms can be transformed into the 2-dimensional space to obtain the target spatial positions of each atom and its nearest neighbor atoms in the 2-dimensional space.

[0032] Here, by performing dimensionality reduction processing on the initial spatial positions, the data structure complexity of each atom can be reduced, and the distance relationship between each atom and its nearest neighbor atoms in the original space can be maintained as much as possible.

[0033] Step S104: Represent the crystal structure using each target spatial position to generate a target representation of the crystal structure.

[0034] The target representation is used to describe the composition of the crystal structure. Specifically, the basic repeating unit of the crystal structure is the unit cell. The edge lengths and angles of the unit cell define the space group of the crystal structure, and the atoms within the unit cell occupy specific positions to form lattice points. When the target spatial positions of each atom and its neighboring atoms are determined, the positions of each lattice point are determined. Therefore, multiple unit cells can be generated according to the positions occupied by each target spatial position. Subsequently, by combining the edge lengths and angles of each unit cell, the crystal structure can be described, and a target representation for describing the crystal structure can be obtained.

[0035] As a specific example, each target spatial position can also be input into a target structure file to form a target representation of the crystal structure. Of course, each target spatial position can also be used to represent the crystal structure in a representation form as Figure 4 shown. In this way, representing the crystal structure using each target spatial position can obtain a unified representation of the crystal structure that is low-dimensional, continuous, comparable, and inputtable, facilitating subsequent applications in scenarios such as indexing and retrieval of material databases, structural classification and clustering, supervised / self-supervised modeling, input space of generative models, material recommendation, and optimization design using the target representation of the crystal structure.

[0036] The method for representing a crystal structure provided in this embodiment uses the initial spatial positions of each atom extracted from the initial representation file to determine the neighboring atoms corresponding to each atom, and then determines the atomic pair distances between each atom and its neighboring atoms; using each atomic pair distance, dimensionality reduction processing is performed on the initial spatial positions of each atom, and the target spatial positions of each atom can be obtained, so that the crystal structure can be represented using the target spatial positions of each atom, and a target representation of the crystal structure can be generated. In this application, dimensionality reduction is performed on the initial spatial positions of each atom through the atomic pair distances between each atom and its neighboring atoms, which can not only map the high-dimensional initial spatial positions to low-dimensional, continuous, and comparable vector representations, but also retain the geometric similarity and symmetry between crystal structures, thus solving the problem of poor representation effect of crystal structures. Further, it breaks through the limitations of traditional high-dimensional discrete structure representations in structure search, performance modeling, and computational efficiency.

[0037] In this embodiment, a method for representing a crystal structure is provided, which can be used in servers, personal computers, cloud platforms, etc. Figure 2 is a flowchart of the method for representing a crystal structure according to an embodiment of the present application, as Figure 2 shown, and this process includes the following steps.

[0038] Step S201: Obtain the initial representation file of the crystal structure and extract the initial spatial positions of each atom in the initial representation file.

[0039] Specifically, the above step S201 includes: Step S2011: Obtain the initial representation file of the crystal structure. For details, please refer to Figure 1 Step S101 of the illustrated embodiment, which will not be elaborated here.

[0040] Step S2012: Obtain the atomic arrangement type of the crystal structure.

[0041] The atomic arrangement type can be the periodic arrangement pattern of atoms (or ions, molecules) in the crystal in three-dimensional space. As a specific example, the atomic arrangement type of the crystal structure can be determined according to the file format of the initial representation file. Of course, it can also be determined according to the crystal structure itself. For example, periodic crystal structures can be graphite, carbon nanotubes, diamond, etc.; non-periodic crystal structures can be carbon clusters, diamond-like carbon, porous carbon, fullerenes, etc.

[0042] Step S2013: If the atomic arrangement type is a periodic type, determine the coordinates of each atom extracted from the initial representation file as the initial spatial positions of each atom.

[0043] Periodicity can be the characteristic that atoms (or ions, molecules) in the crystal are arranged infinitely and repeatedly in a certain fixed pattern in three-dimensional space. The smallest unit of this repetition can be called a unit cell, and the entire crystal structure can be filled by translating the unit cell in space. If the crystal structure itself has periodicity, the coordinates of each atom extracted from the initial representation file can be directly determined as its initial spatial positions, and then the unit cell parameters can be extracted from the initial representation file for coordinate transformation or cell expansion processing. Specifically, the unit cell parameters can include lattice constants and lattice angles.

[0044] Step S2014: If the atomic arrangement type is a non-periodic type, perform coordinate transformation on the coordinates of each atom extracted from the initial representation file to obtain the initial spatial positions of each atom.

[0045] Non-periodicity can be that the arrangement of atoms or molecules in three-dimensional space lacks a long-range ordered repetition pattern, that is, there is no infinitely extendable unit cell, that is, it is impossible to locate atoms relying on unit cell parameters. Thus, for crystal structures without periodicity, it is necessary to calculate the geometric center of all atoms, that is, the centroid of all atomic coordinates, and use it as the new coordinate origin. Then, perform coordinate translation on the positions of all atoms so that the crystal structure is represented with the geometric center as the origin. This not only helps in the consistent comparison between crystal structures but also improves the accuracy of the target representation of the finally generated crystal structure.

[0046] In some alternative embodiments, the above step S2014 includes: Step a1, determining the centroid using the coordinates of each atom extracted from the initial representation file.

[0047] Step a2, taking the centroid as the coordinate origin, translating the coordinates of each atom to obtain the initial spatial positions of each atom.

[0048] The centroid is the geometric center of the atom. Combining the shape of the atom and the coordinates of the atom, the geometric center of the atom can be directly calculated, or the integral method can be used to calculate the centroid, or the finite element analysis method can be used to simulate the mass distribution of the atom to calculate the position of the centroid. Of course, the computer-aided design (CAD) can also be used to automatically calculate and display the position of the centroid. Here, the method for determining the centroid is not specifically limited.

[0049] The centroid can be more accurately determined through the coordinates of each atom. Then, taking the centroid as the coordinate origin, translating the coordinates of each atom makes the crystal structure able to be represented with the geometric center as the origin. This not only helps in the consistent comparison between crystal structures but also improves the accuracy of the target representation of the finally generated crystal structure.

[0050] Step S202, using the initial spatial positions of each atom, determining the neighboring atoms corresponding to each atom, and the inter-atomic pair distances between each atom and its neighboring atoms.

[0051] Specifically, the above step S202 includes: Step S2021, for any target atom among each atom, expanding a preset distance outward with the target atom as the center to obtain a target region.

[0052] Among them, the preset distance can be flexibly set according to actual needs, and this application does not limit this.

[0053] As a specific example, a circular target region can be determined with the target atom as the center and the preset distance as the radius. Since the circular target region can ensure that the radii in all directions are the same, the neighboring atoms of each atom can be more accurately determined using the target region subsequently.

[0054] Step S2022, determining the atoms within the target region as the neighboring atoms of the target atom.

[0055] The number of atoms in the target region can be one or more, so the neighboring atoms of the target atom can be one or more. Using the target region to determine the neighboring atoms of each atom can more accurately determine the neighboring atoms of each atom. That is to say, the same atoms can be processed in the way of expanding the unit cell. That is, the same atom originally in different unit cells will be regarded as different atoms in the same "large lattice" after the expansion of the unit cell, so as to avoid the influence of the periodic boundary conditions of the crystal structure and the situation that there are multiple atom-pair distances corresponding to two atoms.

[0056] Determining the neighboring atoms of each atom through the constructed preset region can more accurately determine the neighboring atoms of each atom. Subsequently, the atom-pair distance between each atom and its neighboring atoms can be more accurately obtained. Further, the accuracy of the determined target spatial positions of each atom can be improved.

[0057] Specifically, step 202 further includes: Step S2023: Perform a symmetry transformation on the initial spatial positions of each atom and its neighboring atoms to obtain the candidate spatial positions of each atom and its neighboring atoms.

[0058] As a specific example, a rotation operation, a reflection operation, an inversion operation, or the like can be used to perform a symmetry transformation on the initial spatial positions of each atom and its neighboring atoms to obtain the candidate spatial positions of each atom and its neighboring atoms. Performing a symmetry transformation on the initial spatial positions of each atom and its neighboring atoms can transform the physical symmetry of the crystal into mathematical constraints, so that the target representation of the crystal structure can accurately retain the essential characteristics of the crystal structure.

[0059] Step S2024: Based on the initial spatial positions of each atom and its neighboring atoms, determine the initial atom-pair distance between each atom and its neighboring atoms.

[0060] For the initial atom-pair distance between each atom and its neighboring atoms, it can be the Euclidean distance calculated using the initial spatial positions of each atom and its neighboring atoms. Of course, it is not limited to the Euclidean distance, and it can also be other distances, such as the Hamming distance, the Manhattan distance, and so on.

[0061] Step S2025: Based on the initial spatial positions and candidate spatial positions of each atom and its neighboring atoms, perform distance optimization on each initial atom-pair distance to determine each atom-pair distance.

[0062] Since the original crystal structure contains several sets of symmetrically equivalent atom pairs, the distances between them should be exactly equal in three-dimensional space. By using the initial spatial positions and candidate spatial positions of each atom and its neighboring atoms to perform distance optimization on each initial atom pair distance, symmetrically non-equivalent atom pair distances can be obtained. That is to say, the purpose of the distance optimization process is to obtain symmetrically non-equivalent atom pair distances. Subsequently, through the symmetrically non-equivalent atom pair distances, the target spatial positions of each atom in the low-dimensional space can be determined, which can better maintain the geometric characteristics of the crystal structure itself and reduce the complexity of the data. Further, it can simplify the target representation of the crystal structure.

[0063] In an alternative embodiment, the above step S2023 further includes: Step b1, obtaining pre-constructed symmetry transformation parameters.

[0064] Step b2, using the symmetry transformation parameters to perform symmetry transformation on the initial spatial positions of each atom and its neighboring atoms to obtain the candidate spatial positions of each atom and its neighboring atoms.

[0065] The symmetry transformation parameters are parameters used for performing position symmetry transformation; the candidate spatial positions are spatial positions obtained through preliminary spatial position transformation. Specifically, the symmetry transformation parameters can be represented by a symmetry matrix For example, this symmetry matrix can be a three-dimensional symmetry matrix. In this application, the specific representation form of the symmetry matrix is not limited and can be flexibly adjusted according to actual needs. Among them, is used to represent the symmetry matrix, is the special orthogonal group (Special Orthogonal Group) in three-dimensional space, which consists of all real matrices W that satisfy the following two conditions. One condition is orthogonality, that is, there is , where is the transpose matrix of W, is the identity matrix; the other condition is that the determinant is 1, that is, there is .

[0066] Continuing with the previous example, if the initial spatial positions of each atom and its neighboring atoms are represented by three-dimensional coordinate vectors, then the initial spatial positions of each atom and its neighboring atoms can be represented in matrix form by as follows:

[0067] where the x-axis, y-axis, and z-axis are the basic orthogonal coordinate axes in three-dimensional space, and their directions are defined according to the right-hand rule, which is used to uniquely determine the spatial position of the atom. ,​ and The three-dimensional coordinate vector used to represent the first atom, and N is used to represent the total number of all atoms and their neighboring atoms. Thus, the process of performing a symmetry transformation on the initial spatial positions of each atom and its neighboring atoms to obtain the candidate spatial positions of each atom and its neighboring atoms can be expressed as:

[0068] wherein, is used to represent the candidate spatial positions of each atom and its neighboring atoms.

[0069] Using the symmetry transformation parameters, perform a symmetry transformation on the initial spatial positions of each atom and its neighboring atoms, so that the obtained candidate spatial positions of each atom and its neighboring atoms can be used to construct constraint conditions, and further can more accurately determine the distance between each pair of atoms, and further ensure the physical rationality of the dimensionality reduction representation of the crystal structure.

[0070] In an alternative embodiment, the above step S2025 includes: Step c1, constructing constraint conditions by using the positional relationship between the initial spatial positions and the candidate spatial positions of each atom and its neighboring atoms.

[0071] Step c2, performing distance optimization processing on each initial pair of atom distances under the constraint of the constraint conditions to determine each pair of atom distances.

[0072] For a crystal structure, there are multiple symmetric and equivalent pairs of atoms, and these symmetric and equivalent pairs of atoms are exactly equal in a high-dimensional space (such as a three-dimensional space). In order to retain symmetry in a low-dimensional space and also be able to briefly perform a target representation of the crystal structure, it is necessary to perform distance optimization processing on multiple symmetric and equivalent pairs of atoms to retain symmetric and non-equivalent pairs of atoms, and use the pair of atom distances of these symmetric and non-equivalent pairs of atoms to determine the target spatial positions of each atom.

[0073] Using the initial and candidate spatial positions of each atom and its neighboring atoms can reasonably construct constraint conditions. At the same time, under the constraint of the constraint conditions, performing distance optimization processing on each initial pair of atom distances can obtain multiple symmetric but non-equivalent pairs of atom distances.

[0074] In an alternative embodiment, the above step c1 includes: determining the candidate pair of atom distances between each atom and its neighboring atoms based on the candidate spatial positions of each atom and the candidate spatial positions of its neighboring atoms; and taking the equality of the initial pair of atom distances and the candidate pair of atom distances between each atom and its neighboring atoms as the constraint condition.

[0075] As a specific example, an atom and its neighboring atoms The initial atomic pair distance can be expressed as For an atom and its neighboring atoms The candidate atomic pair distance can be expressed as Thus, the constraint condition can be expressed as .

[0076] Construct a constraint condition with the initial atomic pair distance and the candidate atomic pair distance of each atom being equal. Subsequently, utilize the constraint condition to perform distance optimization processing on each initial atomic pair distance, and further, multiple symmetric but non-equivalent atomic pair distances can be obtained preferably.

[0077] In an alternative implementation manner, the above step c2 includes: performing fusion processing on each initial atomic pair distance to obtain a target total distance; under the constraint of the constraint condition, performing distance optimization processing on the target total distance based on each initial atomic pair distance to obtain an optimal target total distance; determining the initial atomic pair distances corresponding to the optimal target total distance as the atomic pair distances of each atom.

[0078] Among them, the fusion processing of each initial atomic pair distance can be direct addition fusion or weighted fusion processing. This application does not limit this, and it can be flexibly selected according to the actual situation.

[0079] Under the constraint of the constraint condition, each initial atomic pair distance can be continuously adjusted, so that the target total distance can be adjusted, and then the optimal target total distance under the constraint condition can be obtained.

[0080] As a specific example, each initial atomic pair distance can be directly added to construct a distance optimization function, which can be expressed as:[[]]

[0081] Among them, is used to represent the initial atomic pair distance between an atom and its neighboring atoms The distance optimization function is constrained by . That is to say, under the constraint condition, the initial atomic pair distance corresponding to the minimum sum of the initial raw distances between each atom and its neighboring atoms is determined as the atomic pair distance. It should be understood that this application does not limit the optimization algorithm corresponding to the distance optimization function, and it can be any suitable optimization algorithm. For example, the distance optimization function can be optimized and solved by Lagrange multiplication.

[0082] Step S203: Use the distances between each pair of atoms to perform dimensionality reduction on each initial spatial position, and obtain the target spatial positions corresponding to each atom and its neighboring atoms. For details, please refer to Figure 1 Step S103 of the embodiment shown in this, which will not be elaborated here.

[0083] Step S204: Use each target spatial position to represent the crystal structure and generate a target representation of the crystal structure. For details, please refer to Figure 1 Step S104 of the embodiment shown in this, which will not be elaborated here.

[0084] The method for representing a crystal structure in this application uses the initial spatial positions of each atom extracted from the initial representation file to determine the neighboring atoms corresponding to each atom, and then determines the distance between each pair of atoms between each atom and its neighboring atoms by optimizing the target total distance; using the distances between each pair of atoms to perform dimensionality reduction on the initial spatial positions of each atom can obtain the target spatial positions of each atom, so that the crystal structure can be represented by using the target spatial positions of each atom, and a target representation of the crystal structure is generated. In this application, the initial spatial positions of each atom are reduced in dimension by the distances between each atom and its neighboring atoms, which can not only map the high-dimensional initial spatial positions into low-dimensional, continuous and comparable vector representations, but also retain the geometric similarity and symmetry between crystal structures, thus solving the problem of poor representation effect of crystal structures. Further, it breaks through the limitations of traditional high-dimensional discrete structure representation in structure search, performance modeling and computational efficiency.

[0085] In this embodiment, a method for representing a crystal structure is provided, which can be used in computer devices such as servers and computers. Figure 3 is a flowchart of the method for representing a crystal structure according to an embodiment of the present invention, as Figure 3 shown, and this process includes the following steps.

[0086] Step S301: Obtain the initial representation file of the crystal structure and extract the initial spatial positions of each atom in the initial representation file. For details, please refer to Figure 1 Step S101 of the embodiment shown in this, which will not be elaborated here.

[0087] Step S302: Use the initial spatial positions of each atom to determine the neighboring atoms corresponding to each atom and the distances between each pair of atoms between each atom and its neighboring atoms. For details, please refer to Figure 1 Step S102 of the embodiment shown in this, which will not be elaborated here.

[0088] Step S303: Use the distances between each pair of atoms to perform dimensionality reduction on each initial spatial position, and obtain the target spatial positions corresponding to each atom and its neighboring atoms.

[0089] After the atomic pair distance matrix is established, in order to reduce the complexity of the structural data and facilitate visualization, classification or machine learning modeling, the multidimensional scaling algorithm (MDS) can be used to reduce the dimensionality of the three-dimensional structure. Figure 4 As shown in the figure, MDS is a classic nonlinear manifold learning method, whose core goal is to embed high-dimensional data into low-dimensional space while maintaining the distance relationship between data points in the original space as much as possible. Specifically in the modeling of crystal structures, the goal of MDS is to make the distance between atomic pairs as consistent as possible after dimensionality reduction. By optimizing the objective function, the MDS algorithm outputs a set of low-dimensional coordinate representations, in which the position of each atom on its low-dimensional plane reflects its local geometric structure in the original dimensional space as much as possible. In this way, not only the core features of the structure are retained, but also the data dimension is greatly reduced, which is helpful for the subsequent classification, clustering analysis or graph neural network input of the structure.

[0090] Specifically, the above step S303 includes: Step S3031, normalizing the distances of each atom pair to obtain the normalized distances of each atom pair.

[0091] As a specific example, each atomic pair distance can be normalized by maximum value normalization or normalization by average atomic distance, so that the atomic pair distances of all samples remain consistent. Since the scale differences of different crystal structures may lead to different ranges of atomic pair distances, normalizing the distances of each atomic pair can ensure the comparability of the results after dimensionality reduction.

[0092] Before normalizing the distances of each atom pair, the distances of each atom pair can be screened using a preset distance threshold. For example, all atom pair distances less than the distance threshold are screened out based on the distance threshold. The atom pairs corresponding to the remaining atom pair distances are considered to have potential chemical interactions or local structural dependencies. All atom pair distances that meet the distance threshold screening conditions can be used to construct an atom pair distance matrix, which is a symmetric sparse matrix that records the local geometric information between atoms. The atom pair distance matrix plays a role similar to a "graph structure" in this scheme, providing key geometric constraints for subsequent dimensionality reduction. As a specific example, the preset distance threshold can be between 1.5 angstroms (Å) and 3.5Å. Of course, the preset distance threshold is not limited to between 1.5.Å and 3.5Å, and can also be other distance thresholds.

[0093] Step S3032, performing dimensionality reduction processing on the initial spatial positions of each atom and its neighboring atoms, and determining the target atom pair distance with the smallest difference from the normalized distance of the atom pair.

[0094] There are various methods for dimensionality reduction of the initial spatial positions of each atom and its neighboring atoms. For example, an atomic pair distance matrix can be used to perform dimensionality reduction on the initial spatial positions of each atom and its neighboring atoms. Another example is that a graph neural network can be used to perform dimensionality reduction on the initial spatial positions of each atom and its neighboring atoms.

[0095] Step S3033: Determine the target spatial positions corresponding to each atom and its neighboring atoms according to the distances of each target atomic pair.

[0096] As a specific example, after determining the distance of the target atomic pair, the two spatial positions corresponding to the distance of the target atomic pair can be used as the target spatial positions of the corresponding atom and its neighboring atoms.

[0097] As a specific example, after obtaining the atomic pair distance matrix, in order to reduce the representation dimension of the crystal structure and maintain the geometric characteristics of the structure itself, dimensionality reduction can be performed on the initial spatial positions of each atom based on the MDS algorithm. The specific process can be as follows: First, construct a target stress function:

[0098] where is the matrix element in the atomic pair distance matrix, that is, the normalized distance of the atomic pair, is the atom 's target spatial position in the low-dimensional space, is the atom 's neighboring atom 's target spatial position in the low-dimensional space, is a weight factor, which can usually be set to 1. Of course, it can also be weighted according to the distance size to emphasize the neighboring relationship between the atom and its neighboring atom .

[0099] Second, use an optimization solution algorithm to optimize and solve the target stress function. For example, use the SMACOF (Scaling by MAjorizing a COmplicated Function) iterative optimization algorithm to optimize and solve the target stress function. This iterative optimization algorithm can effectively process large-scale samples and improve the convergence speed by gradually approaching the optimal solution. It should be understood that at the initial solution, the atom and its neighboring atom The target spatial positions can be randomly initialized or use the preprocessing results of the Principal Component Analysis (PCA) as the initial values, and then perform iterative optimization. Finally, the target spatial positions of each atom and its neighboring atoms are obtained.

[0100] Step S304: Use each target spatial position to represent the crystal structure and generate the target representation of the crystal structure.

[0101] Specifically, the above step S304 includes: Step S3041: Perform projection processing on each target spatial position onto multiple projection planes to obtain multiple projection spatial positions under each projection plane, and the projected atom pair distances between each atom and its neighboring atoms.

[0102] As a specific example, if the target spatial positions after dimensionality reduction are three-dimensional coordinates, then each target spatial position can be projected onto the xy plane, xz plane, and yz plane respectively. After projection, the projection spatial positions corresponding to each target spatial position under the xy plane can be obtained, the projection spatial positions corresponding to each target spatial position under the xz plane can also be obtained, and the projection spatial positions corresponding to each target spatial position under the xz plane can also be obtained.

[0103] Continuing with the previous example, after obtaining the projection spatial positions corresponding to each target spatial position on the xy plane, xz plane, and yz plane, the projected atom pair distances between each atom and its neighboring atoms under the xy plane, the projected atom pair distances between each atom and its neighboring atoms under the xz plane, and the projected atom pair distances between each atom and its neighboring atoms under the yz plane can be determined using the projection spatial positions. That is, under each projection plane, there are corresponding projection spatial positions and projected atom pair distances for each atom.

[0104] Step S3042: Determine multiple target projection spatial positions under the target projection plane according to the number of projected atom pair distances greater than the first preset threshold under each projection plane.

[0105] Continuing with the previous example, according to the projected atom pair distances under each projection plane, the number M1 of projected atom pair distances greater than the first preset threshold under the xy plane can be obtained; the number M2 of projected atom pair distances greater than the first preset threshold under the xz plane, and the number M3 of projected atom pair distances greater than the first preset threshold under the yz plane. For example, if M1 is greater than M2 and M1 is greater than M3, then the xy plane is determined as the target projection plane, and the projection spatial positions under the target projection plane are determined as the corresponding target projection spatial positions.

[0106] It should be understood that, since the distance range corresponding to each projection surface is different, each projection surface can correspond to a first preset threshold value. Of course, if the distance range corresponding to each projection surface is the same, multiple projection surfaces can correspond to the same first preset threshold value. This is not limited in the present application, and can be flexibly adjusted according to actual needs. For the specific setting of the first preset threshold value, this application does not limit this, and can be flexibly adjusted according to actual needs.

[0107] Step S3043, representing the crystal structure according to multiple target projection space positions to generate a target representation of the crystal structure.

[0108] As a specific example, multiple target projection space positions can be input into a target structure file to form a target representation of the crystal structure. The target projection space position is further reduced in dimension relative to the target space position, further reducing the complexity of the structure data and facilitating visualization, classification or machine learning modeling.

[0109] The spatial positions of multiple target projections under the target projection plane are determined by the number of projected atom pairs under each projection plane whose distance is greater than a first preset threshold. That is, the standard embedding dimension is selected through the projection principle, which can facilitate subsequent visualization and modeling.

[0110] As an optional implementation, the above step S3043 further includes: Step d1, determining the target projected atomic pair distance between each atom and its neighboring atoms based on the target projected spatial position of each atom and its neighboring atoms.

[0111] Step d2, removing atoms whose target projection atom pair distance is greater than a second preset threshold, to obtain multiple target atoms.

[0112] Step d3, representing the crystal structure according to the target projection spatial position of each target atom, and generating a target representation of the crystal structure.

[0113] Atoms whose distance to the target projected atom is greater than the second preset threshold are determined as isolated atoms. Eliminating isolated atoms can prevent isolated atoms (i.e., too far away from other atoms) from affecting the quality of dimensionality reduction, and when generating a low-dimensional target representation of the crystal structure, the main connected subgraph can be guaranteed to be complete to enhance the consistency of subsequent modeling. It should be understood that the specific setting of the second preset threshold is not limited in this application, and can be flexibly adjusted according to actual needs.

[0114] The method for representing the crystal structure of the present application uses the initial spatial positions of each atom extracted from the initial representation file to determine the neighboring atoms corresponding to each atom, and then determines the atomic pair distances between each atom and its neighboring atoms; uses each atomic pair distance to perform dimensionality reduction processing on the initial spatial positions of each atom, and can obtain the target spatial positions of each atom. By means of projection processing, the target spatial positions of each atom and its neighboring atoms are processed to determine the target projected spatial positions of each atom and its neighboring atoms, and uses the magnitude relationship between the target projected atomic pair distance and the second preset threshold to eliminate isolated atoms, so that the crystal structure can be represented by using the target projected spatial positions of each target atom, and a target representation of the crystal structure is generated. In this way, not only can the high-dimensional initial spatial positions be mapped into low-dimensional, continuous and comparable vector representations, but also the geometric similarity and symmetry between crystal structures can be retained, thus solving the problem of poor representation effect of crystal structures. Further, it breaks through the limitations of traditional high-dimensional discrete structure representation in structure search, performance modeling and computational efficiency.

[0115] As a specific example, such as Figure 4 and Figure 5 shown, it is a method for representing the carbon crystal structure by taking the carbon crystal structure as an example and based on manifold learning. This method can realize the unified expression of low-dimensional, continuous, comparable and inputtable complex carbon material structures, and break through the limitations of traditional high-dimensional discrete structure representation in structure search, performance modeling and computational efficiency. This method is based on the carbon atom coordinates and lattice information of carbon materials. First, local structure features and adjacency relationships are extracted, and then nonlinear dimensionality reduction is performed through the manifold learning method to generate a target representation of the low-dimensional carbon crystal structure that retains structural similarity and geometric characteristics, and at the same time maintains the symmetry of the crystal itself during the generation process. The template representation of this carbon crystal structure can be used in application scenarios such as indexing and retrieval of material databases, structure classification and clustering, supervised / self-supervised modeling, input space of generative models, material recommendation and optimization design, and specifically includes the following steps.

[0116] Step S501, data preprocessing.

[0117] Input the carbon crystal structure file into a computer device to extract the high-dimensional coordinate information of each carbon atom in the carbon crystal structure file (i.e., the initial spatial positions shown above). Among them, the carbon crystal structure file usually contains key information such as carbon atom types, three-dimensional coordinates of carbon atoms, and lattice parameters. During the process of the computer device reading the carbon crystal structure file, first, analyze the file format of the carbon crystal structure file, and then determine the preset extraction strategy corresponding to the file format; finally, use the preset extraction strategy to extract the three-dimensional spatial coordinate information of all carbon atoms in the carbon crystal structure file. For periodic crystal structures, such as graphene, diamond, etc., the unit cell parameters will be further read, including lattice constants (a, b, c) and lattice angles (α, β, γ), for coordinate transformation or unit cell expansion processing. In non-periodic carbon structures, such as carbon clusters, fullerenes (such as C60), etc., since their structures themselves do not have periodicity, it is impossible to rely on unit cell information to locate carbon atoms. At this time, use the three-dimensional spatial coordinate information of all carbon atoms to calculate the geometric center of all carbon atoms, that is, the centroid of all carbon atom coordinates, and use it as the new reference origin. Then, perform coordinate translation on the positions of all carbon atoms so that the carbon crystal structure is represented with the geometric center as the origin, which not only helps the consistency comparison between structures but also improves the effect of subsequent manifold learning.

[0118] Before generating the target representation of the carbon crystal structure, perform sufficient and standardized preprocessing on the data. The goal is to extract unified and standardized information from carbon crystal structure files of different sources or formats so that subsequent algorithms can run smoothly and maintain high efficiency and accuracy.

[0119] Step S502, carbon crystal structure feature extraction.

[0120] For many carbon crystal structures, especially those with high symmetry (such as cubic diamond, hexagonal graphite, etc.), space group symmetry is a fundamental element that constitutes the material properties. By introducing a space group symmetry constraint mechanism, the physical rationality of the dimensionality reduction representation can be ensured. Specifically, the original three-dimensional crystal structure contains several sets of symmetrically equivalent carbon atom pairs, and the distances between them in three-dimensional space should be exactly equal. To make the reduced two-dimensional planar structure still retain this symmetry, it is necessary to ensure that the distances between all symmetrically equivalent carbon atom pairs in the two-dimensional plane also remain consistent. In this way, by constructing a distance optimization function and constraint conditions, the distances between each carbon atom and its neighboring carbon atoms are determined, which not only enhances the structural rationality of the dimensionality reduction representation but also provides a more reliable input representation for subsequent machine learning models sensitive to symmetry. Among them, the distance optimization function and constraint conditions can be expressed as:

[0121] 。

[0122] After determining the carbon atom pair distances between each carbon atom and its neighboring carbon atoms, all carbon atom pairs whose distances are less than the distance threshold can be screened out through a preset distance threshold. If the distances are less than the preset distance threshold, it is considered that the carbon atom pairs have potential chemical interactions or local structural dependencies. All carbon atom pair distances that meet the conditions will be used to construct a carbon atom pair distance matrix, which is a symmetric sparse matrix that records the local geometric information between carbon atoms. It plays a role similar to a "graph structure" in this method and provides key geometric constraints for subsequent dimensionality reduction. It is worth noting that due to the periodic boundary conditions of carbon crystals, there may be multiple distances between two carbon atoms. In order to include as much crystal structure information as possible, the same carbon atom can be processed by cell expansion. That is, the same carbon atom that was originally in different lattices will be treated as different carbon atoms in the same "large lattice" after cell expansion.

[0123] Step S503: manifold learning reduces the representation dimension of the crystal structure and generates a target representation of the crystal structure.

[0124] Step S5031, since the scale difference of different crystal structures may lead to different ranges of distance matrix, maximum value normalization or standardization according to the average carbon atom distance is adopted to normalize the matrix elements in the carbon atom distance matrix so that the average carbon atom distance of all samples remains consistent.

[0125] Step S5032, constructing a target stress function .

[0126] Step S5033, using the SMACOF iterative optimization algorithm to solve and obtain the target spatial position of each carbon atom.

[0127] Step S5034, perform projection processing on multiple projection planes for each target spatial position, obtain multiple projection spatial positions under each projection plane and the projection atomic pair distances between each atom and its neighboring atoms, and determine the multiple target projection spatial positions under the target projection plane based on the number of projection atomic pair distances under each projection plane that are greater than a first preset threshold.

[0128] Step S5035, using the target projection spatial position of each atom and its neighboring atoms, determine the target projection atom pair distance between each atom and its neighboring atoms; eliminate atoms whose target projection atom pair distance is greater than a second preset threshold to obtain multiple target atoms; use the target projection spatial position of each target atom to represent the crystal structure and generate a target representation of the crystal structure.

[0129] Through the above processing, the present application can effectively map high-dimensional atomic arrangements to low-dimensional space while retaining the local structural similarity of the original carbon crystals, generating continuous and comparable structural feature vectors, and providing a high-quality input data foundation for subsequent machine learning, material screening, performance prediction and other applications.

[0130] Step S504, performing downstream model application based on the target representation of the low-dimensional carbon crystal structure.

[0131] 1) Input characteristics of carbon material property prediction model The low-dimensional target representation is used as the input feature of the machine learning model (such as support vector machine, graph neural network, Transformer, etc.), and the model is trained to predict the key physical and chemical properties of carbon materials, such as band width, electron mobility, specific surface area, thermal stability, etc.

[0132] 2) Reverse design and generation of carbon crystals Combined with generative models (such as variational autoencoders, diffusion models, etc.), low-dimensional target representation is used as spatial constraints to perform inverse generation and optimization search of carbon structures, and design new carbon material structures that meet specific performance indicators.

[0133] 3) Unsupervised classification and cluster analysis of carbon structure Unsupervised learning methods such as K-means, DBSCAN, and spectral clustering are applied in low-dimensional space to automatically classify and cluster carbon crystal structures, and to explore the potential evolutionary relationships, stability partitions, or functional partitions in carbon structure systems.

[0134] Through the above applications, this application not only realizes an efficient and comparable representation method for carbon crystal structure, but also establishes a complete machine learning modeling link from structural representation to material discovery, greatly improving the intelligence level and efficiency of carbon material research and development.

[0135] The embodiment of the present application also provides a crystal structure display device, such as Figure 6 As shown, the device includes an acquisition module 610, a determination module 620, a dimension reduction processing module 630 and a structure representation module 640.

[0136] An acquisition module 610 is used to acquire an initial representation file of the crystal structure and extract the initial spatial position of each atom in the initial representation file; A determination module 620 is used to determine the neighboring atoms corresponding to each atom and the atomic pair distances between each atom and its neighboring atoms using the initial spatial position of each atom; A dimension reduction processing module 630 is used to perform dimension reduction processing on each initial spatial position using each atomic pair distance to obtain a target spatial position corresponding to each atom and its neighboring atoms; A structure representation module 640 is used to represent the crystal structure using respective target spatial positions, generating a target representation of the crystal structure.

[0137] As an alternative implementation, the acquisition module 610 is further configured to acquire the atomic arrangement type of the crystal structure; if the atomic arrangement type is a periodic type, the coordinates of each atom extracted from the initial representation file are determined as the initial spatial positions of each atom; if the atomic arrangement type is a non-periodic type, coordinate transformation is performed on the coordinates of each atom extracted from the initial representation file to obtain the initial spatial positions of each atom.

[0138] As an alternative implementation, the acquisition module 610 is further configured to determine the centroid using the coordinates of each atom extracted from the initial representation file; with the centroid as the coordinate origin, the coordinates of each atom are translated to obtain the initial spatial positions of each atom.

[0139] As an alternative implementation, the determination module 620 is further configured to, for any target atom among each atom, expand a preset distance outward from the target atom to obtain a target region; the atoms within the target region are determined as the neighboring atoms of the target atom.

[0140] As an alternative implementation, the determination module 620 is further configured to perform symmetry transformation processing on the initial spatial positions of each atom and its neighboring atoms to obtain the candidate spatial positions of each atom and its neighboring atoms; based on the initial spatial positions of each atom and its neighboring atoms, determine the initial atom pair distances between each atom and its neighboring atoms; based on the positional relationship between the initial spatial positions and the candidate spatial positions of each atom and its neighboring atoms, perform distance optimization processing on each initial atom pair distance to determine each atom pair distance.

[0141] As an alternative implementation, the determination module 620 is further configured to acquire pre-constructed symmetry transformation parameters; using the symmetry transformation parameters, perform symmetry transformation on the initial spatial positions of each atom and its neighboring atoms to obtain the candidate spatial positions of each atom and its neighboring atoms.

[0142] As an alternative implementation, the determination module 620 is further configured to construct constraint conditions using the positional relationship between the initial spatial positions and the candidate spatial positions of each atom and its neighboring atoms; under the constraint of the constraint conditions, perform distance optimization processing on each initial atom pair distance to determine each atom pair distance.

[0143] As an alternative implementation, the determination module 620 is further configured to determine the candidate atomic pair distances between each atom and its neighboring atoms based on the positional relationship between the candidate spatial positions of each atom and the candidate spatial positions of its neighboring atoms; and determine the constraint conditions when the initial atomic pair distances between each atom and its neighboring atoms are equal to the candidate atomic pair distances.

[0144] As an alternative implementation, the determination module 620 is further configured to perform a fusion process on each initial atomic pair distance to obtain a target total distance; perform a distance optimization process on the target total distance based on each initial atomic pair distance under the constraint of the constraint conditions to obtain an optimal target total distance; and determine each initial atomic pair distance corresponding to the optimal target total distance as each atomic pair distance.

[0145] As an alternative implementation, the dimensionality reduction processing module 630 is further configured to perform dimensionality reduction processing on the initial spatial positions of each atom and its neighboring atoms to determine a target atomic pair distance with the smallest difference from the atomic pair normalized distance; and determine the target spatial positions corresponding to each atom and its neighboring atoms according to each target atomic pair distance.

[0146] As an alternative implementation, the structure representation module 640 is further configured to perform projection processing on each target spatial position on multiple projection planes to obtain multiple projected spatial positions under each projection plane and the projected atomic pair distances between each atom and its neighboring atoms; determine multiple target projected spatial positions under the target projection plane according to the number of projected atomic pair distances greater than a first preset threshold under each projection plane; and represent the crystal structure according to the multiple target projected spatial positions to generate a target representation of the crystal structure.

[0147] As an alternative implementation, the structure representation module 640 is further configured to determine the target projected atomic pair distances between each atom and its neighboring atoms based on the target projected spatial positions of each atom and its neighboring atoms; remove the atoms with target projected atomic pair distances greater than a second preset threshold to obtain multiple target atoms; and represent the crystal structure according to the target projected spatial positions of each target atom to generate a target representation of the crystal structure.

[0148] For the description of the features in the corresponding embodiments of the crystal structure representation device, reference can be made to the relevant descriptions in the corresponding embodiments of the crystal structure representation method, which will not be elaborated here one by one.

[0149] An embodiment of the present application further provides a computer device, as Figure 7 shown, including a memory 710 and a processor 720. The memory 710 stores a computer program, and the processor 720 is configured to run the computer program to execute the steps in any of the above embodiments of the crystal structure representation method.

[0150] Embodiments of the present application also provide a computer-readable storage medium storing a computer program, wherein the computer program is configured to execute the steps in any of the above-described embodiments of the method for representing a crystal structure when running.

[0151] In an exemplary embodiment, the above computer-readable storage medium may include, but is not limited to: various media such as USB flash drives, read-only memories (ROM), random access memories (RAM), external hard disks, magnetic disks, or optical discs that can store computer programs.

[0152] Embodiments of the present application also provide a computer program product, where the computer program product includes a computer program, and the steps in any of the above-described embodiments of the method for representing a crystal structure are implemented when the computer program is executed by a processor.

[0153] Embodiments of the present application also provide another computer program product, including a non-volatile computer-readable storage medium storing a computer program, and the steps in any of the above-described embodiments of the method for representing a crystal structure are implemented when the computer program is executed by a processor.

[0154] Those skilled in the art can further realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been generally described according to their functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Skilled professionals can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0155] The above has introduced in detail a method for representing a crystal structure, a computer device, and a storage medium provided by the present application. Specific examples are used herein to elaborate on the principle and implementation manner of the present application. The description of the above embodiments is only used to help understand the method and its core idea of the present application. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the protection scope of the claims of the present application.

Claims

1. A method for representing a crystal structure, characterized in that, Including: Obtain an initial representation file of the crystal structure, and extract the initial spatial positions of each atom in the initial representation file; Using the initial spatial positions of each atom, determine the neighboring atoms corresponding to each atom, and the atomic pair distances between each atom and its neighboring atoms; Perform dimensionality reduction processing on each of the initial spatial positions using the atomic pair distances to obtain the target spatial positions corresponding to each atom and its neighboring atoms; Use each of the target spatial positions to represent the crystal structure and generate a target representation of the crystal structure.

2. The method according to claim 1, wherein Extracting the initial spatial positions of each atom in the initial representation file includes: Obtain the atomic arrangement type of the crystal structure; If the atomic arrangement type is a periodic type, determine the coordinates of each atom extracted from the initial representation file as the initial spatial position of each atom; If the atomic arrangement type is a non-periodic type, perform coordinate transformation on the coordinates of each atom extracted from the initial representation file to obtain the initial spatial position of each atom.

3. The method according to claim 2, wherein Performing coordinate transformation on the coordinates of each atom extracted from the initial representation file to obtain the initial spatial position of each atom includes: Determine the centroid using the coordinates of each atom extracted from the initial representation file; Taking the centroid as the coordinate origin, translate the coordinates of each atom to obtain the initial spatial position of each atom.

4. The method according to claim 1, wherein Using the initial spatial positions of each atom to determine the neighboring atoms corresponding to each atom includes: For any target atom among each atom, expand a preset distance outward with the target atom as the center to obtain a target region; Determine the atoms within the target region as the neighboring atoms of the target atom.

5. The method according to any one of claims 1 or 4, characterized in that Determining the atomic pair distances between each atom and its neighboring atoms includes: Perform symmetric transformation processing on the initial spatial positions of each atom and its neighboring atoms to obtain the candidate spatial positions of each atom and its neighboring atoms; Based on the initial spatial positions of each atom and its neighboring atoms, determine the initial atomic pair distances between each atom and its neighboring atoms; Based on the initial spatial positions and candidate spatial positions of each atom and its neighboring atoms, perform distance optimization processing on each of the initial atomic pair distances to determine each atomic pair distance.

6. The method according to claim 5, wherein Performing symmetric transformation processing on the initial spatial positions of each atom and its neighboring atoms to obtain the candidate spatial positions of each atom and its neighboring atoms includes: Obtain pre-constructed symmetric transformation parameters; Using the symmetric transformation parameters, perform symmetric transformation on the initial spatial positions of each atom and its neighboring atoms to obtain the candidate spatial positions of each atom and its neighboring atoms.

7. The method according to claim 5, characterized in that, Based on the initial spatial positions and candidate spatial positions of each atom and its neighboring atoms, performing distance optimization processing on each of the initial atomic pair distances to determine each atomic pair distance includes: Construct constraint conditions by using the positional relationships between the initial spatial positions and the candidate spatial positions of each of the atoms and its neighboring atoms; Under the constraint of the constraint conditions, perform distance optimization processing on each of the initial atomic pair distances to determine each of the atomic pair distances.

8. The method according to claim 7, wherein Construct constraint conditions by using the positional relationships between the initial spatial positions and the candidate spatial positions of each of the atoms and its neighboring atoms, including: Based on the candidate spatial positions of each of the atoms and the candidate spatial positions of its neighboring atoms, determine the candidate atomic pair distances between each of the atoms and its neighboring atoms; Determine the constraint conditions with the initial atomic pair distances between each of the atoms and its neighboring atoms being equal to the candidate atomic pair distances.

9. The method according to claim 7, wherein Under the constraint of the constraint conditions, perform distance optimization processing on each of the initial atomic pair distances to determine each of the atomic pair distances, including: Perform fusion processing on each of the initial atomic pair distances to obtain a target total distance; Under the constraint of the constraint conditions, perform distance optimization processing on the target total distance based on each of the initial atomic pair distances to obtain an optimal target total distance; Determine each of the initial atomic pair distances corresponding to the optimal target total distance as each of the atomic pair distances.

10. The method according to claim 1, characterized in that, Perform dimensionality reduction processing on each of the initial spatial positions by using each of the atomic pair distances to obtain the target spatial positions corresponding to each of the atoms and its neighboring atoms, including: Perform normalization processing on each of the atomic pair distances to obtain each atomic pair normalized distance; Perform dimensionality reduction processing on the initial spatial positions of each of the atoms and its neighboring atoms to determine the target atomic pair distance with the smallest difference from the atomic pair normalized distance; Determine the target spatial positions corresponding to each of the atoms and its neighboring atoms according to each of the target atomic pair distances.

11. The method according to claim 1, wherein Represent the crystal structure by using each of the target spatial positions to generate a target representation of the crystal structure, including: Perform projection processing on each of the target spatial positions on multiple projection planes to obtain multiple projected spatial positions under each projection plane, and the projected atomic pair distances between each of the atoms and its neighboring atoms; Determine multiple target projected spatial positions on the target projection plane according to the number of the projected atomic pair distances greater than a first preset threshold under each projection plane; Represent the crystal structure according to the multiple target projected spatial positions to generate a target representation of the crystal structure.

12. The method according to claim 11, wherein Represent the crystal structure according to the multiple target projected spatial positions to generate a target representation of the crystal structure, including: Based on the target projected spatial positions of each of the atoms and its neighboring atoms, determine the target projected atomic pair distances between each of the atoms and its neighboring atoms; Remove the atoms with the target projected atomic pair distances greater than a second preset threshold to obtain multiple target atoms; Represent the crystal structure according to the target projected spatial positions of each of the target atoms to generate a target representation of the crystal structure.

13. A computer device, characterized in that, Including: A memory for storing a computer program; A processor for implementing the steps of the method for representing a crystal structure according to any one of claims 1 to 12 when executing the computer program.

14. A computer-readable storage medium, characterized in that, A computer program is stored in the computer-readable storage medium, wherein the computer program, when executed by a processor, implements the steps of the method for representing a crystal structure according to any one of claims 1 to 12.

15. A computer program product, comprising a computer program, characterized in that, The computer program, when executed by a processor, implements the steps of the method for representing a crystal structure according to any one of claims 1 to 12.

Images