Processing apparatus, system, method, and program

Abstract

To provide a processing unit, a system, a method, and a program that perform non-negative value matrix factorization of a measurement profile of X-ray powder diffraction based upon known information.SOLUTION: The present invention relates to a processing unit 400 that performs non-negative value matrix factorization of a measurement profile of X-ray powder diffraction, and comprises: a measurement profile acquisition part 410 which acquires one or more measurement profiles; a known information acquisition part 420 which acquires known information including the shape of a background included in the measurement profile or a predetermined profile corresponding to a predetermined substance, or limitations of a coefficient matrix of the predetermined profile; and a decomposition part 430 which decomposes the measurement profile into non-negative value matrix factors based upon the known information.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] The present invention relates to a processing device, a system, a method, and a program.

Background Art

[0002] X-ray powder diffraction is used in various fields. By analyzing the measurement profile of X-ray powder diffraction, for example, it is possible to identify (qualitative analysis) and quantify the constituent components of a powder sample. Conventionally, the crystal phase has been identified by comparing the measurement profile or the d-I list created from the measurement profile with the diffraction pattern of a known substance.

[0003] Patent Document 1 discloses a crystal phase identification method for identifying a crystal phase contained in a sample from the powder diffraction pattern of the sample using a database, including: {a full pattern fitting step of performing full pattern fitting on a first diffraction pattern which is the powder diffraction pattern of the sample using sample-containing crystal phase information to calculate a theoretical diffraction pattern of the already identified crystal phase}; {a residual information generation step of generating residual information of the sample based on the difference between the theoretical diffraction pattern and the first diffraction pattern}; and {a residual information search match step of selecting a new crystal phase contained in the sample by comparing the residual information with the database}.

[0004] Patent Document 2 discloses a spectral data analysis device that obtains a plurality of basis spectral data and activation data representing the magnitude of each basis spectrum by performing a non-negative matrix decomposition of a set of observed spectral data obtained for a signal to be analyzed. The spectral data analysis device obtains the plurality of basis spectral data and the activation data by searching for a minimum value of an objective function that includes a degree of deviation between the set of observed spectral data and a set of estimated spectral data calculated from the plurality of basis spectral data and the activation data, as well as a regularization term that evaluates the linear independence of the plurality of basis spectral data or the activation data. [Prior art documents] [Patent Documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2014-178203 [Patent Document 2] Japanese Patent Publication No. 2019-87042 [Overview of the Initiative] [Problems that the invention aims to solve]

[0006] When there is a large amount of mixture, the measurement profiles of X-ray powder diffraction show significant overlap of peaks in each profile. However, if the method described in Patent Document 1, which uses a dI list for search and match without processing the measurement profiles, is applied in such cases, the accuracy of the qualitative analysis deteriorates.

[0007] Furthermore, the technology described in Patent Document 2 improves the accuracy of decomposition by assuming a high degree of independence between profiles. However, if there is a large overlap in the peaks of each profile, imposing regularization based on first-order independence increases the risk that each profile cannot be accurately decomposed, and the accuracy of subsequent qualitative analysis deteriorates.

[0008] As a result of diligent research, the inventors of this invention have discovered that X-ray powder diffraction measurement profiles often contain associated known information, and that performing non-negative matrix factorization of the measurement profile based on this known information results in higher accuracy of the factorization and subsequent qualitative analysis compared to performing non-negative matrix factorization without using the known information. Based on these findings, the inventors have completed the present invention.

[0009] The present invention has been made in view of these circumstances, and aims to provide a processing apparatus, system, method, and program for performing non-negative matrix factorization of an X-ray powder diffraction measurement profile based on known information. [Means for solving the problem]

[0010] (1) To achieve the above objective, the present invention provides an apparatus for performing non-negative matrix factorization of an X-ray powder diffraction measurement profile, comprising: a measurement profile acquisition unit for acquiring one or more of the measurement profiles; a known information acquisition unit for acquiring known information including the shape of a predetermined profile corresponding to a background or a predetermined substance included in the measurement profile, or constraints on the coefficient matrix of the predetermined profile; and a decomposition unit for performing non-negative matrix factorization of the measurement profile based on the known information.

[0011] (2) Furthermore, in the apparatus of the present invention, the decomposition unit is characterized in that it selectively performs either a normal non-negative matrix factorization or a constrained non-negative matrix factorization based on the known information, depending on whether or not the known information is present.

[0012] (3) Furthermore, the apparatus of the present invention is characterized in that the known information is information that the coefficient matrix values ​​of the predetermined profile which is included in common with the plurality of measurement profiles are equal.

[0013] (4) Furthermore, the apparatus of the present invention is characterized in that the known information is information about the shape of the predetermined profile included in the measurement profile.

[0014] (5) Furthermore, the apparatus of the present invention is characterized in that the information on the shape of the predetermined profile is based on information obtained from a database or on measured data.

[0015] (6) The apparatus of the present invention further comprises a dendrogram creation unit that calculates statistics between a plurality of measurement profiles and creates a dendrogram, wherein the decomposition unit is characterized by performing non-negative matrix factorization of clusters from the dendrogram that include a group of similar profiles selected by the apparatus or selected by the user.

[0016] (7) The apparatus of the present invention is further characterized by comprising: a peak search unit that performs a peak search on the profile after the non-negative matrix factorization and creates a dI list; and a qualitative unit that performs a qualitative analysis using the dI list.

[0017] (8) The apparatus of the present invention is further characterized by comprising a quantitative unit that performs quantitative analysis using the qualitatively analyzed data.

[0018] (9) The present invention is characterized by comprising an X-ray diffractometer having an X-ray generating unit for generating X-rays, a detector for detecting X-rays, and a goniometer for controlling the rotation of a sample, and a processing apparatus according to any one of (1) to (8) above.

[0019] (10) The present invention also provides a method for non-negative matrix factorization of an X-ray powder diffraction measurement profile, comprising the steps of: acquiring one or more of the measurement profiles; acquiring known information including the shape of a predetermined profile corresponding to a background or a predetermined substance included in the measurement profiles, or constraints on the coefficient matrix of the predetermined profile; and performing non-negative matrix factorization of the measurement profiles based on the known information.

[0020] (11) Further, the program of the present invention is a program for performing non - negative matrix factorization on the measurement profile of X - ray powder diffraction, and includes a process of acquiring one or more of the measurement profiles, a process of acquiring known information including the shape of a background or a predetermined profile corresponding to a predetermined substance included in the measurement profile, or a limitation on the coefficient matrix of the predetermined profile, and a process of performing non - negative matrix factorization on the measurement profile based on the known information, and is characterized by causing a computer to execute these processes.

Brief Description of the Drawings

[0021] [Figure 1] (a) and (b) are conceptual diagrams showing the state of non - negative matrix factorization and the state of non - negative matrix factorization when including known information, respectively. [Figure 2] It is a conceptual diagram showing an example of the configuration of an X - ray diffraction measurement system. [Figure 3] It is a block diagram showing an example of the configuration of a control device and a processing device. [Figure 4] It is a block diagram showing a modified example of the configuration of a control device and a processing device. [Figure 5] It is a block diagram showing a modified example of the configuration of a control device and a processing device. [Figure 6] It is a block diagram showing a modified example of the configuration of a processing device. [Figure 7] It is a block diagram showing a modified example of the configuration of a processing device. [Figure 8] It is a schematic diagram showing an example of a UI for calling the function of cluster analysis. (b) is a schematic diagram showing an example of a UI for setting the function of cluster analysis, etc. [Figure 9] (a) is a schematic diagram describing a part of the functions of the UI of cluster analysis. [Figure 10] It is a schematic diagram showing an example of a dialog for setting known information. [Figure 11] It is a schematic diagram showing an example of a UI for setting the creation of a dendrogram, etc. [Figure 12]This is a schematic diagram showing an example of a UI for setting up search and match, quantitative functions, etc. [Figure 13] This is a flowchart illustrating an example of the operation of the processing unit. [Figure 14] This flowchart shows a variation of the operation of the processing unit. [Figure 15] This flowchart shows a variation of the operation of the processing unit. [Modes for carrying out the invention]

[0022] Next, embodiments of the present invention will be described with reference to the drawings. To facilitate understanding of the description, the same reference numerals are used for identical components in each drawing, and redundant descriptions are omitted.

[0023] [principle] X-ray powder diffraction measurement profiles contain overlapping profiles and backgrounds from multiple substances. When there are many mixtures, peak overlap increases significantly. In such cases, the accuracy of peak search deteriorates, and conventional search-matching using dI lists is often unsuitable.

[0024] Non-negative matrix factorization (NMF) is the process of decomposing a non-negative matrix into a product of non-negative matrices. To facilitate search and match, we consider decomposing the measurement profile of X-ray powder diffraction into a weighted sum of multiple profiles (including the background profile). Since each profile and its weight are non-negative, non-negative matrix factorization is suitable for representing the measurement profile of X-ray powder diffraction as a weighted sum of multiple profiles.

[0025] Figure 1(a) is a conceptual diagram illustrating the non-negative matrix factorization. The left side of Figure 1(a) shows a matrix of n X-ray powder diffraction measurement profiles, each with m measurement points. The right side of Figure 1(a) is the result of non-negative matrix factorization of this matrix. However, the wavy equality in Figure 1 includes not only exact agreement but also cases where the degree of deviation, which indicates the degree of closeness between the left and right sides, is less than or equal to a predetermined value.

[0026] X-ray powder diffraction measurement profiles can sometimes exhibit significant profile overlap. Significant profile overlap occurs, for example, when there is a large overlap of peaks in each profile, when amorphous material is present, or when there is a large background. In such cases, [patent document] 2 Imposing regularization based on linear independence, as shown above, when performing non-negative matrix factorization is not appropriate because it can worsen the accuracy of subsequent search and match operations.

[0027] Furthermore, X-ray powder diffraction measurement profiles often have associated known information. The method of the present invention performs non-negative matrix factorization of the X-ray powder diffraction measurement profile based on known information. Non-negative matrix factorization of a measurement profile based on known information means that the measurement profile is factorized into a non-negative matrix using the known information as constraints, such that the constraints are satisfied. The method of the present invention can perform non-negative matrix factorization with high accuracy even when there is a large overlap in profiles. Although various methods have been proposed for factorizing a given non-negative matrix into a non-negative matrix, the present invention can use a general method.

[0028] In Figure 1(a), we assume that known information is associated with the X-ray powder diffraction measurement profile. Known information refers to information associated with the X-ray powder diffraction measurement profile. More details will be provided later. In such cases, a decomposition that appropriately expresses the known information as a result of the non-negative matrix factorization will be a more accurate decomposition. If the right-hand side of Figure 1(a) contains known information, it can be extracted and represented, for example, as the second term on the right-hand side of Figure 1(b). Figure 1(b) is a conceptual diagram showing the non-negative matrix factorization when known information is included.

[0029] The method of the present invention sets the known information to one or more of the following on the right-hand side of Figure 1(b): matrix W (N rows, R columns), matrix W' (N rows, S columns), or matrix B' (S rows, M columns). Under these constraints, W', W, and matrix B (R rows, M columns) are optimized. Optimization means finding a basis matrix and coefficient matrix whose deviation is less than or equal to a predetermined value. In this way, a decomposition can be performed such that the known information is appropriately represented as a result of non-negative matrix factorization.

[0030] As described above, the method of the present invention can perform non-negative matrix factorization of the measurement profile of X-ray powder diffraction based on known information. The detailed method of the present invention will be described in detail in the embodiments.

[0031] [Embodiment] The method of the present invention will be described in detail below. The method for non-negative matrix factorization of the measurement profile of X-ray powder diffraction measured by an X-ray diffractometer will be described below. Furthermore, methods for qualitative analysis and quantitative analysis using this method will be described.

[0032] Let X be an N x M matrix containing the measurement profiles of N X-ray powder diffractions, each with M measurement points. The non-negative matrix factorization of X is given by the following equation (1): W is the coefficient matrix, and B is the basis matrix. W represents the weights of B. B consists of basis vectors in each row. R is a hyperparameter indicating the number of basis vectors.

[0033]

number

[0034] Assume that a portion of W or B on the right-hand side of equation (1) contains known information associated with the X-ray powder diffraction measurement profile. If we extract the coefficient matrix or basis matrix representing the known information from W or B in equation (1) and denote it as W' or B', the non-negative matrix factorization of X can be rewritten as shown in equation (2) below. However, assume that one or more of W, W', or B' contain known information. S is a hyperparameter indicating the number of basis vectors of the known information. Also, the equations from which W' and B' were extracted are again represented as W and B.

[0035]

number

[0036] Known information in an X-ray powder diffraction measurement profile includes background information included in the measurement profile, the shape of a predetermined profile corresponding to a given substance, or constraints on the coefficient matrix of a given profile. One or more of these known information parameters (W, W', or B') are set, and W', W, and B are optimized under these constraints. In this way, non-negative matrix factorization can be performed using the known information of the X-ray powder diffraction measurement profile as constraints. Note that equation (2) is a format intended to clearly indicate the existence of known information; the known information may also be set in equation (1) for optimization.

[0037] The known information is preferably that the coefficient matrix values ​​of a predetermined profile, which is included in multiple measurement profiles, are equal. This is a non-negative matrix factorization constrained by the fact that equal amounts of components are included in multiple measurement profiles. For example, it can be applied when multiple measurement profiles share a background from the instrument or when equal amounts of standard substances are included. It can also be applied, for example, when multiple measurement profiles measure a single sample over time and there are profiles derived from components that do not react during measurement.

[0038] For example, multiple measurement profiles measured under the same conditions using a single X-ray diffractometer contain the same background. In such cases, where it is known that multiple measurement profiles share a common profile shape, but that shape is unknown, the constraint that the values ​​of a certain column in the coefficient matrix W are equal is imposed, and W and the basis matrix B are optimized.

[0039] The known information is preferably information about the shape of a predetermined profile included in the measurement profile. This is a non-negative matrix factorization constrained by the fact that one or more measurement profiles contain a known profile. For example, it can be applied when the background shape is known or when the contents are known. The information about the shape of the predetermined profile is preferably information obtained from a database or information based on measured data.

[0040] For example, if the shape of the background of the measurement profile is known, the profile representing the background shape is set as the basis matrix B', and the coefficient matrix W', W, and B are optimized.

[0041] For non-negative matrix factorization, it is preferable to selectively perform either normal non-negative matrix factorization or constrained non-negative matrix factorization based on known information, depending on the presence or absence of known information. This allows for accurate non-negative matrix factorization of the measurement profile depending on the type and content of the known information, when known information is available. Conversely, when no known information is available, the measurement profile can be non-negative matrix factorized without constraints.

[0042] Non-negative matrix factorization can be performed using optimization methods such as alternating least squares, multiplicative update, and coordinate descent, with added regularization. Regularization can be applied by imposing weights or sparsity on the decomposed profiles.

[0043] When there are multiple measurement profiles, it is preferable to calculate statistics between the measurement profiles and create a dendrogram. Furthermore, it is preferable to perform non-negative matrix factorization on clusters containing similar profile groups selected by the processing device described later from the dendrogram, or selected by the user.

[0044] To accurately perform non-negative matrix factorization of multiple measurement profiles, it is desirable that most measurement profiles include characteristic profiles. By creating dendrograms and selecting clusters containing similar profile groups from the created dendrograms, it is expected that measurement profiles with a low probability of containing characteristic profiles can be excluded. As a result, multiple measurement profiles can be accurately performed non-negative matrix factorization.

[0045] It is preferable to perform qualitative analysis after non-negative matrix factorization of the measurement profile. For qualitative analysis, it is preferable to perform a peak search on the profile (basis vectors) after non-negative matrix factorization and create a dI list. Qualitative analysis can then be performed by performing a search match using the created dI list. Qualitative analysis can be performed using known methods. Since the profile after non-negative matrix factorization has less peak overlap than the measurement profile before non-negative matrix factorization, performing qualitative analysis using the dI list created from the factorized profile often makes component identification easier or improves the accuracy of identification.

[0046] After performing qualitative analysis on the measurement profile using non-negative matrix factorization, it is preferable to perform further quantitative analysis as needed. The quantitative analysis is performed using the data obtained from the qualitative analysis. The quantitative analysis can be performed using known methods.

[0047] In this way, the measurement profile of X-ray powder diffraction measured by an X-ray diffractometer can be factorized into a non-negative matrix based on known information. Furthermore, qualitative and quantitative analyses can be performed using this factorization.

[0048] [Overall System] Figure 2 is a conceptual diagram showing an example of the configuration of an X-ray diffraction measurement system 100. System 100 includes an X-ray diffractometer 200, a control device 300, and a processing device 400. The X-ray diffractometer 200 comprises an optical system that incidents X-rays onto a sample and detects the diffracted X-rays generated from the sample, and the optical system includes a goniometer. Note that the configuration shown in Figure 2 is just one example, and various other configurations can be adopted.

[0049] The control device 300 is connected to the X-ray diffractometer 200 and controls the X-ray diffractometer 200 and processes and stores the acquired data. The processing device 400 performs non-negative matrix factorization of the X-ray powder diffraction measurement profile. The control device 300 and processing device 400 are devices equipped with a CPU and memory, and may be PC terminals or servers on the cloud. Furthermore, not only the entire system, but also some of the devices or some of the functions within the system may be provided on the cloud. The input device 510 is, for example, a keyboard and mouse, and provides input to the control device 300 and processing device 400. The display device 520 is, for example, a display, and displays the measurement profile and the results of the non-negative matrix factorization.

[0050] By using such a system 100, the X-ray powder diffraction profile can be measured, and the measured profile can be factorized into a non-negative matrix. Furthermore, qualitative and quantitative analyses can be performed using the non-negative matrix-factorized profile.

[0051] In Figure 2, the control device 300 and the processing device 400 are shown as the same PC. However, as explained above, the method of the present invention can acquire a measurement profile and perform non-negative matrix factorization independently of the X-ray diffractometer 200 and the control device 300. Therefore, as shown in Figure 3, the processing device 400 may be configured as a different device from the control device 300. Figure 3 is a block diagram showing an example of the configuration of the control device 300 and the processing device 400. Also, as shown in Figure 4, the processing device 400 may be configured as a part of the functions included in the control device 300. Furthermore, as shown in Figure 5, the processing device 400 and the control device 300 may be configured as an integrated device. Figures 4 and 5 are block diagrams showing modified configurations of the control device 300 and the processing device 400. Below, we will describe the case where the control device 300 and the processing device 400 are configured as different devices.

[0052] [X-ray diffractometer] The X-ray diffractometer 200 comprises an X-ray generator 210 that generates X-rays from the X-ray focal point, i.e., the X-ray source; an incident optical unit 220; a goniometer 230; a sample stage 240 for placing the sample; an exit optical unit 250; and a detector 260 for detecting X-rays. The X-ray generator 210, incident optical unit 220, goniometer 230, sample stage 240, exit optical unit 250, and detector 260 that make up the X-ray diffractometer 200 can be general-purpose components, so their explanation will be omitted.

[0053] [Control device] The control unit 300 is composed of a computer consisting of a CPU (Central Processing Unit), ROM (Read Only Memory), RAM (Random Access Memory), and memory connected to a bus. The control unit 300 is connected to the X-ray diffractometer 200 to receive information.

[0054] The control device 300 comprises a control unit 310, a device information storage unit 320, a measurement data storage unit 330, and a display unit 340. Each unit can send and receive information via the control bus L. The input device 510 and the display device 520 are connected to the CPU via appropriate interfaces.

[0055] The control unit 310 controls the operation of the X-ray diffractometer 200. The device information storage unit 320 stores device information acquired from the X-ray diffractometer 200. The device information includes information about the X-ray diffractometer 200 such as the device name, type of radiation source, wavelength, and background. In addition, it may include information necessary for non-negative matrix factorization of the measurement profile of X-ray powder diffraction, such as the type and composition of constituent elements of the sample.

[0056] The measurement data storage unit 330 stores the measurement profile acquired from the X-ray diffractometer 200. Along with the measurement profile, it may also store information necessary for non-negative matrix factorization of the X-ray powder diffraction measurement profile, such as the type of radiation source, wavelength, background, type of constituent elements of the sample, and composition. The display unit 340 displays the measurement profile on the display device 520. This allows the user to confirm the measurement profile. Furthermore, the user can give instructions and specifications to the control device 300, processing device 400, etc., based on the measurement data.

[0057] [Processing device] The processing unit 400 is composed of a computer with a CPU, ROM, RAM, and memory connected to a bus. The processing unit 400 may be connected to the X-ray diffractometer 200 via the control device 300.

[0058] The processing unit 400 comprises a measurement profile acquisition unit 410, a known information acquisition unit 420, and a decomposition unit 430. Each unit can send and receive information via the control bus L. If the processing unit 400 and the control device 300 have different configurations, the input device 510 and the display device 520 are also connected to the CPU of the processing unit 400 via an appropriate interface. In this case, the input device 510 and the display device 520 may be different from those connected to the control device 300.

[0059] The measurement profile acquisition unit 410 acquires one or more measurement profiles. The measurement profile acquisition unit 410 may acquire the measurement profiles directly from the X-ray diffractometer 200 or via the control device 300.

[0060] The known information acquisition unit 420 acquires known information including the background included in the measurement profile, the shape of a predetermined profile corresponding to a predetermined substance, or the constraints of the coefficient matrix of a predetermined profile.

[0061] Preferably, the known information acquired by the known information acquisition unit 420 is information that the coefficient matrix values ​​of a predetermined profile, which are commonly included in multiple measurement profiles, are equal. This allows for non-negative matrix factorization using this known information, for example, when the background originating from the X-ray diffractometer 200 is common or when equal amounts of standard substances are included.

[0062] The known information acquired by the known information acquisition unit 420 is preferably information about the shape of a predetermined profile included in the measurement profile. This allows for non-negative matrix factorization using this known information, for example, when the background shape or the contents are known. Furthermore, the information about the shape of the predetermined profile is preferably information acquired from a database or information based on measured data.

[0063] The decomposition unit 430 performs non-negative matrix factorization of the measurement profile based on known information. Performing non-negative matrix factorization of the measurement profile based on known information means using the known information as constraints and performing non-negative matrix factorization of the measurement profile such that those constraints are satisfied.

[0064] The decomposition unit 430 preferably selectively performs either a normal non-negative matrix factorization or a constrained non-negative matrix factorization based on known information, depending on whether or not known information is available.

[0065] Figure 6 is a block diagram showing a modified configuration of the processing unit 400. As shown in Figure 6, the processing unit 400 preferably includes a dendrogram creation unit 440. The dendrogram creation unit 440 calculates statistics between multiple profiles and creates a dendrogram.

[0066] When the dendrogram creation unit 440 creates a dendrogram, it is preferable for the decomposition unit 430 to perform non-negative matrix factorization on clusters containing similar profiles selected by the processing unit 400 or selected by the user from the dendrogram. This improves the accuracy of the non-negative matrix factorization.

[0067] Figure 7 is a block diagram showing a modified configuration of the processing apparatus 400. As shown in Figure 7, the processing apparatus 400 preferably comprises a dendrogram creation unit 440, a peak search unit 450, a qualitative unit 460, and a quantitative unit 470. The dendrogram creation unit 440 is the same as described above.

[0068] The peak search unit 450 performs a peak search on the profiles (basis vectors) after non-negative matrix factorization and creates a dI list. The peak search is performed on one selected profile from the results of the non-negative matrix factorization. The selection of the profile may be done by the user, or by the peak search unit 450 or another function unit of the processing unit 400. A dI list is also created for each profile on which a peak search has been performed. It is preferable to perform the peak search on all profiles except those determined to be background.

[0069] The qualitative analysis unit 460 performs qualitative analysis using the dI list. Qualitative analysis can be performed by performing a search match on the created dI list. The quantitative analysis unit 470 performs quantitative analysis using the data analyzed qualitatively. In the configuration shown in Figure 7, the dendrogram creation unit 440, peak search unit 450, qualitative analysis unit 460, and quantitative analysis unit 470 are optional functional units, and the configuration may be one or more of them omitted.

[0070] [User Interface] Parameters of the processing unit 400, etc. SettingsWhen this is done according to user instructions, it is preferable to use a user interface (UI) function that allows various settings to be made using, for example, mouse or keyboard operations. Furthermore, it is preferable that the functions of the processing unit 400 be configured to cooperate with the functions of other devices. Below, an example of a UI for setting the parameters of the processing unit 400 and an example of a UI when the functions of the processing unit 400 cooperate with the functions of other devices will be described. Assume that the functions of the processing unit 400 are implemented as software.

[0071] Figure 8(a) is a schematic diagram showing an example of a UI for calling the cluster analysis function. Figure 8(b) is a schematic diagram showing an example of a UI for configuring the cluster analysis function, etc. The cluster analysis function in Figure 8(b) is assumed to include the functions of the processing unit 400. By pressing the cluster analysis button on the screen in Figure 8(a), the screen in Figure 8(b) will appear.

[0072] Figure 9 is a schematic diagram illustrating some of the functions of the cluster analysis UI (UI in Figure 8(b)). 9 On the screen, the user can give instructions to the processing unit 400, such as acquiring a measurement profile, setting parameters, performing non-negative matrix factorization, inputting compound information, outputting the results of non-negative matrix factorization, and transferring data. The parameter setting panel allows setting the optimization method for non-negative matrix factorization, hyperparameter values, number of iterations, regularization, and whether or not known information is included. The hyperparameter values ​​may be automatically estimated and set using the Akaike information criterion, Bayesian information criterion, etc., or the estimated values ​​may be displayed using the Estimate button. The data transfer button transfers the data of the non-negative matrix factorization results to the search match and quantitative functions. Figure 10 is a schematic diagram showing an example of a dialog for setting known information. The content of the known information can be set on the dialog in Figure 10.

[0073] Figure 11 is a schematic diagram showing an example of a UI for setting up dendrogram creation. On the screen in Figure 11, the user can give instructions to the processing unit 400 for setting up statistics for dendrogram creation, data processing settings, cluster selection, etc.

[0074] Figure 12 is a schematic diagram showing an example of a UI for setting functions such as search matching and quantitative analysis. Pressing the data transfer button on the screen in Figure 9 brings up the screen in Figure 12. On the screen in Figure 12, the user can give instructions to the processing device 400 or other devices, such as executing a search match or setting compound information.

[0075] Note that the settings shown in Figures 8 to 12 are just examples, and even when users configure these settings, they may choose to configure only some of them, or all of them. Furthermore, there may be settings and functions not shown in Figures 8 to 12.

[0076] [Measurement method] The sample S is placed in the X-ray diffractometer 200, and the goniometer is driven under predetermined conditions based on the control of the control device 300. X-rays are also incident on the sample, and the diffracted X-rays generated from the sample are detected. Diffraction data is then acquired. The X-ray diffractometer 200 transmits the device information and the acquired diffraction data as measurement data to the control device 300.

[0077] [Disassembly method] (Explanation of the flow when performing only non-negative matrix factorization) Figure 13 is a flowchart illustrating an example of the operation of the processing unit 400. Figure 13 shows an example of operation when only non-negative matrix factorization is performed. First, the processing unit 400 acquires a measurement profile (step S1). Next, it acquires known information from the acquired measurement profile (step S2). The known information may be acquired from the X-ray diffractometer 200 or the control device 300, or it may be acquired by user input. Since the known information is associated with the measurement profile, it may be acquired at the same time as the acquisition of the measurement profile. Known information common to multiple measurement profiles used when performing non-negative matrix factorization may not be determined at the time of acquiring the known information, so the known information to be used may be determined from among the acquired known information.

[0078] Next, the parameters are set (step S3). The parameters to be set are the number of basis vectors to be decomposed (hyperparameter values), the optimization method, and other parameters necessary for optimization. The parameters may be set based on input from the user, or the processing device 400 may determine and set them based on the measurement profile or known information. Next, non-negative matrix factorization is performed (step S4). Non-negative matrix factorization is performed by creating a matrix consisting of the measurement profile, setting the known information as a basis matrix or coefficient matrix, and optimizing the basis matrix and coefficient matrix other than the known information. Then, the results are output as needed (step S5), and the process ends. Alternatively, the system may be configured to simply store the results and output them only when instructed by the user. In this way, the measurement profile of X-ray powder diffraction can be subjected to non-negative matrix factorization based on known information.

[0079] In step S2 of the flowchart in Figure 13, when acquiring known information, if the acquired measurement profile does not have any associated known information or if there is no user input, the subsequent processing may be carried out without acquiring the known information. In other words, when acquiring known information, the system may verify the presence or absence of known information and selectively perform either a normal non-negative matrix factorization or a constrained non-negative matrix factorization based on the known information. Furthermore, if it is confirmed that known information exists, the system may decide whether to impose constraints on the basis matrix, the coefficient matrix, or both, depending on the content of the known information. In this way, for X-ray powder diffraction measurement profiles, the system can selectively perform either a normal non-negative matrix factorization or a constrained non-negative matrix factorization based on the known information, depending on the presence or absence of known information.

[0080] (Explanation of the flow for creating a dendrogram) Figure 14 is a flowchart illustrating a modified operation of the processing unit 400. Figure 14 shows an example of operation when creating a dendrogram. In the following flowchart description, characteristic operations will be described in detail, and descriptions of operations already described may be omitted. First, the processing unit 400 acquires a measurement profile (step T1). Next, it acquires known information of the acquired measurement profile (step T2).

[0081] Next, a dendrogram is created (step T3). The dendrogram is created by calculating statistics between multiple acquired measurement profiles. Next, clusters are selected (step T4). Clusters may be selected by the user or by the processing unit. Selecting clusters, which are groups of similar profiles from the dendrogram, improves the resolution accuracy. When performing non-negative matrix factorization, the known information common to multiple measurement profiles (clusters) may not be determined at the time of acquiring the known information. Therefore, the known information to be used may be determined from the known information acquired after cluster selection. Alternatively, the known information may be acquired after cluster selection.

[0082] Next, the parameters are set (step T5). Then, non-negative matrix factorization is performed (step T6). Finally, the results are output as needed (step T7), and the process terminates. In this way, For data measured with an X-ray diffraction device, After creating a dendrogram and selecting clusters, non-negative matrix factorization can be performed.

[0083] (Explanation of the flowchart for variations when performing qualitative analysis or further quantitative analysis) Figure 15 is a flowchart showing a modified operation of the processing device 400. Figure 15 shows a modified operation when qualitative analysis or further quantitative analysis is performed after non-negative matrix factorization. The steps from acquiring the measurement profile (step U1) to performing non-negative matrix factorization (step U4) are the same as steps S1 to S4.

[0084] Next, a peak search is performed (step U5). The peak search is performed on one selected profile (basis vector) from the results of non-negative matrix factorization. The selection of the profile may be done by the user or by the processing unit 400. A dI list is also created for each profile that has undergone a peak search. It is preferable to perform the peak search on all profiles except those that have been determined to be background.

[0085] Next, perform a qualitative analysis (step U6). The qualitative analysis can be performed by performing a search match based on the created dI list. If the search match does not find any profiles with a matching degree equal to or greater than a predetermined value, you may return to step U3 and perform non-negative matrix factorization again, starting from setting the parameters.

[0086] Next, quantitative analysis is performed (step U7). Quantitative analysis determines the content of the substance identified by qualitative analysis using various methods. For example, the DD (Direct Derivation) method, RIR method, and Rietveld method can be used. Then, the results are output as needed (step U8), and the process is completed. In this way, the measurement profile of X-ray powder diffraction can be factorized into a non-negative matrix based on known information, qualitative analysis can be performed, and this can be used to perform quantitative analysis.

[0087] In the flowchart of Figure 15, after the qualitative analysis in step U6, the results may be output as needed (step U8) without performing a quantitative analysis, and the process may be terminated. In this way, the measurement profile of the X-ray powder diffraction can be factorized into a non-negative matrix based on known information, and this can be used for qualitative analysis.

[0088] Furthermore, in the flowchart of Figure 15, after obtaining the measurement profile in step U1, or obtaining the known information in step U2, a flow for creating a dendrogram and selecting clusters may be added.

[0089] The order of the steps in each flowchart described above is not fixed; the order may be changed or the steps may be processed in parallel, as long as they can be processed correctly. Furthermore, each flowchart may be applied in combination with other flowcharts.

[0090] [Examples] Using the system 100 configured as described above, X-ray diffraction data was measured for a mixture of indomethacin. Non-negative matrix factorization was performed on the measurement profile, using the shape of the background profile as known information. By setting the hyperparameters other than the known information to 3, the profiles of indomethacin α-type, indomethacin γ-type, amorphous profile, and background profile were appropriately decomposed, enabling qualitative analysis.

[0091] Non-negative matrix factorization was performed on the same measurement profile using a conventional method. As a result, it was not possible to properly decompose the indomethacin α-type, indomethacin γ-type, amorphous profile, and background profile. This is thought to be because the profiles derived from the background and amorphous components overlap significantly with the crystalline phase profile. The method of the present invention was confirmed to be able to extract profiles derived from the background and amorphous components even in such cases.

[0092] Based on the above results, it has been confirmed that the apparatus, system, method, and program of the present invention can perform non-negative matrix factorization of the X-ray powder diffraction measurement profile based on known information. [Explanation of Symbols]

[0093] 100 Systems 200 X-ray diffractometer 210 X-ray generating unit 220 Incident Optical Unit 230 Goniometer 240 Sample stage 250 Output-side optical unit 260 detectors 300 Control device 310 Control Unit 320 Device information storage unit 330 Measurement data storage unit 340 Display section 400 Processing Units 410 Measurement Profile Acquisition Unit 420 Known Information Acquisition Unit 430 Disassembly section 440 Dendrogram Creation Department 450 Peak Search Unit 460 Qualitative Section 470 Quantification section 510 Input device 520 Display device

Claims

1. A processing apparatus for performing non-negative matrix factorization of the measurement profile of X-ray powder diffraction, A measurement profile acquisition unit that acquires multiple measurement profiles, A known information acquisition unit that acquires known information including the background included in the measurement profile or the shape of a predetermined profile corresponding to a predetermined substance, The system includes a decomposition unit that performs non-negative matrix factorization of the measurement profile into a basis matrix and a coefficient matrix based on the known information, The processing apparatus is characterized in that the decomposition unit uses the known information as a constraint on the basis matrix to determine the basis matrix and the coefficient matrix such that the degree of deviation is less than or equal to a predetermined value.

2. The processing apparatus according to claim 1, characterized in that the decomposition unit selectively performs either a normal non-negative matrix factorization or a constrained non-negative matrix factorization based on the known information, depending on whether or not the known information is present.

3. The apparatus according to claim 1, characterized in that the information on the shape of the predetermined profile is information obtained from a database or information based on measured data.

4. The unit further comprises a dendrogram creation unit that calculates statistics between multiple measurement profiles and creates a dendrogram, The apparatus according to claim 1, characterized in that the decomposition unit performs non-negative matrix factorization of clusters including a group of similar profiles selected by the apparatus or selected by the user from the dendrogram.

5. A peak search unit performs a peak search on the profile after the non-negative matrix factorization and creates a d-I list. The apparatus according to claim 1, further comprising a qualitative unit that performs qualitative analysis using the d-I list.

6. The apparatus according to claim 5, further comprising a quantitative unit for performing quantitative analysis using the qualitatively analyzed data.

7. An X-ray diffractometer comprising an X-ray generating unit for generating X-rays, a detector for detecting X-rays, and a goniometer for controlling the rotation of a sample, A system comprising the processing apparatus according to any one of claims 1 to 6.

8. A method for non-negative matrix factorization of the measurement profile of X-ray powder diffraction, The steps include acquiring one or more of the aforementioned measurement profiles, The steps include obtaining known information including the background included in the measurement profile or the shape of a predetermined profile corresponding to a predetermined substance, The step of performing non-negative matrix factorization of the measurement profile into a basis matrix and a coefficient matrix based on the known information, The step of performing non-negative matrix factorization is characterized by determining the basis matrix and the coefficient matrix such that the degree of deviation is less than or equal to a predetermined value, using the known information as a constraint on the basis matrix.

9. A program for performing non-negative matrix factorization of X-ray powder diffraction measurement profiles, A process for acquiring one or more of the aforementioned measurement profiles, A process for acquiring known information including the background included in the measurement profile or the shape of a predetermined profile corresponding to a predetermined substance, The computer is instructed to perform a process of non-negative matrix factorization of the measurement profile into a basis matrix and a coefficient matrix based on the known information. The process of performing non-negative matrix factorization is a program characterized by using the known information as a constraint on the basis matrix to find the basis matrix and the coefficient matrix such that the degree of deviation is less than or equal to a predetermined value.

10. An apparatus for performing non-negative matrix factorization of a measurement profile of X-ray powder diffraction, A measurement profile acquisition unit that acquires multiple measurement profiles, A known information acquisition unit that acquires known information including constraints on the coefficient matrix of a predetermined profile corresponding to the background or a predetermined substance included in the measurement profile, The system includes a decomposition unit that performs non-negative matrix factorization of the measurement profile into a basis matrix and a coefficient matrix based on the known information, The processing apparatus is characterized in that the decomposition unit uses the known information as a constraint on the coefficient matrix to determine the basis matrix and the coefficient matrix such that the degree of deviation is less than or equal to a predetermined value.

11. The apparatus according to claim 10, characterized in that the known information is information that the values ​​of the coefficient matrices of the predetermined profile, which are commonly included in the plurality of measurement profiles, are equal.

Images