Large-scale Reconfigurable Diffractive Optical Computing Chip System and Architecture

Abstract

The present disclosure relates to the field of optical computing technologies, and particularly to a large-scale reconfigurable diffractive optical computing chip system and architecture. Among them, the system includes: an acquisition module for acquiring a first large matrix and a second large matrix that need to be calculated; a calculation decomposition module for performing calculation decomposition on the first large matrix and the second large matrix to obtain a plurality of decomposed matrices; and a calculation module for performing matrix calculation through a diffractive calculation array based on the plurality of decomposed matrices to obtain a target output result. The present disclosure realizes flexible adjustment of the computing network architecture by flexibly adjusting the diffractive calculation array, so as to break through the computing scale limitation of a single reconfigurable diffractive computing core, and further greatly improve the computing scale, and can be flexibly applied in various large and complex application scenarios.

Description

Technical Field

[0001] The present disclosure relates to the field of optical computing technology, and in particular, to a large-scale reconfigurable diffractive optical computing chip system and architecture. Background Art

[0002] With the rapid development of the fields of artificial intelligence and scientific computing, the complexity and scale of computing requirements are also increasing continuously. However, existing microelectronic computing chips face performance bottlenecks (such as speed, energy consumption, etc.) when dealing with the rapidly growing computing power requirements, and it is difficult to effectively meet the increasingly stringent requirements of large-scale complex algorithms for computing power and power consumption. As a new computing paradigm, optical computing has prominent advantages such as large bandwidth, high speed, and low energy consumption. Using photons as information carriers and photon devices to accelerate neural network computing makes optical computing technology regarded as the key to breaking the existing computing bottleneck. Summary of the Invention

[0003] The present disclosure aims to at least partly solve one of the technical problems in the related art.

[0004] To this end, the first object of the present disclosure is to propose a large-scale reconfigurable diffractive optical computing chip system.

[0005] The second object of the present disclosure is to propose a large-scale reconfigurable diffractive optical computing method.

[0006] To achieve the above object, an embodiment of the first aspect of the present disclosure proposes a large-scale reconfigurable diffractive optical computing chip system, where the chip system includes:

[0007] The acquisition module is configured to acquire a first large matrix and a second large matrix that need to be calculated;

[0008] The calculation decomposition module is configured to perform calculation decomposition on the first large matrix and the second large matrix to obtain a plurality of decomposed matrices;

[0009] The calculation module is configured to perform matrix calculation based on the plurality of decomposed matrices through a diffractive calculation array to obtain a target output result.

[0010] Optionally, the diffractive calculation array includes a plurality of reconfigurable diffractive calculation cores, and the plurality of reconfigurable diffractive calculation cores are connected longitudinally and / or transversely. Among them, the reconfigurable diffractive calculation core includes a signal feeding module, a parameter modulation module, a channel synthesis module, a diffractive propagation module, and a result output module.

[0011] The signal feeding module is configured to acquire a first decomposed matrix corresponding to an input mode and obtain a signal parameter matrix through a modulator array;

[0012] The parameter modulation module is configured to obtain a second decomposition matrix of the input parameters, and process the second decomposition matrix through a modulation kernel to obtain a corresponding modulation parameter matrix;

[0013] The channel synthesis module is configured to randomly fuse the modulation parameter matrix and the signal parameter matrix to obtain a fusion matrix;

[0014] The diffraction propagation module is configured to input the fusion matrix into a target diffraction propagation model to obtain a diffraction transmission matrix;

[0015] The result output module is configured to output a target form signal based on the diffraction transmission matrix.

[0016] Optionally, the channel synthesis module is specifically configured to: perform juxtaposed fusion or cross fusion on the modulation parameter matrix and the signal parameter matrix to obtain a fusion matrix.

[0017] Optionally, the diffraction propagation module is specifically configured to: input the fusion matrix into a target diffraction propagation model, and obtain a diffraction transmission matrix through a target diffraction kernel in the target diffraction propagation model, where the target diffraction kernel is trained.

[0018] Optionally, the result output module is specifically configured to:

[0019] Determine the target form to be output;

[0020] Based on the target form, determine the corresponding signal feeding technology;

[0021] Based on the signal feeding technology and the diffraction transmission matrix, output a target form signal.

[0022] Optionally, the calculation decomposition module is specifically configured to:

[0023] Perform calculation decomposition on the first large matrix and the second large matrix through tensor decomposition to obtain a plurality of decomposition matrices.

[0024] To achieve the above object, an embodiment of the second aspect of the present disclosure proposes a large-scale reconfigurable diffraction optical calculation method, and the method includes:

[0025] Obtain a first large matrix and a second large matrix to be calculated;

[0026] Perform calculation decomposition on the first large matrix and the second large matrix to obtain a plurality of decomposition matrices;

[0027] Based on the plurality of decomposition matrices, perform matrix calculation through a diffraction calculation array to obtain a target output result.

[0028] Another object of the present disclosure is to provide a large-scale reconfigurable diffractive optical computing architecture, which is characterized by including at least one of the large-scale reconfigurable diffractive optical computing chip systems.

[0029] In summary, the large-scale reconfigurable diffractive optical computing chip system and architecture provided by the present disclosure decompose large-scale computations, perform matrix computations on the decomposed matrices through a diffractive computing array, and obtain the target output result. Thus, by flexibly adjusting the diffractive computing array, the computing network architecture can be flexibly adjusted to break through the computing scale limitation of a single reconfigurable diffractive computing core, thereby greatly improving the computing scale and enabling flexible application in various large and complex application scenarios.

[0030] Additional aspects and advantages of the present disclosure will be partially given in the following description, partially become apparent from the following description, or be understood through the practice of the present disclosure. Description of the Drawings

[0031] The above and / or additional aspects and advantages of the present disclosure will become apparent and understandable from the following description of the embodiments in conjunction with the drawings, where:

[0032] Figure 1 is a schematic structural diagram of a large-scale reconfigurable diffractive optical computing chip system provided by an embodiment of the present disclosure;

[0033] Figure 2 is a schematic diagram of a diffractive computing array provided by an embodiment of the present disclosure;

[0034] Figure 3 is a schematic diagram of a large-scale reconfigurable diffractive computing cluster provided by an embodiment of the present disclosure;

[0035] Figure 4 is a schematic flowchart of a large-scale reconfigurable diffractive optical computing method provided by an embodiment of the present disclosure. Detailed Embodiments

[0036] Embodiments of the present disclosure will be described in detail below. Examples of the embodiments are shown in the drawings, where the same or similar reference numerals denote the same or similar elements or elements with the same or similar functions throughout. The embodiments described below with reference to the drawings are exemplary and are intended to explain the present disclosure and should not be construed as limiting the present disclosure.

[0037] At present, optical neural network chips can include interference, resonance, and diffraction neural networks, etc. However, in practical applications, the current optical computing chip technology still faces multiple major challenges and problems, such as small chip network scale, single function, weak generality and flexibility, etc., which makes it difficult to fully exert its huge application potential. Specifically, for interference or resonance neural networks, they mainly use photon devices such as waveguide phase shifters, Mach-Zehnder interferometers, directional couplers, and micro-ring resonators as basic neuron devices, and the physical size of the neuron devices is about dozens - hundreds of micrometers, which limits the number scale of integratable neurons and the scale of the neural network that can be realized, and thus makes it unable to handle large-scale complex artificial intelligence tasks.

[0038] Among them, the diffraction neural network has the advantage of high-density neuron integration due to its sub-wavelength neuron physical size, and is expected to become an optical computing technology that can simultaneously achieve large network scale, high computing power density, and high computing energy efficiency. However, the current diffraction optical computing technology is limited by a single computing core, which makes the diffraction optical computing technology limited to solving small-scale simple computing tasks such as image classification and unable to handle complex tasks such as natural language processing or multi-modal information processing.

[0039] In addition, the high-density integration characteristic of diffraction neurons leads to extremely difficult reconstruction problems under the current on-chip modulation technology framework, seriously restricting the scalability of its scale and the flexibility of its architecture. Among them, the above challenges and problems lead to the problems of limited computing power, single function, and inflexible architecture in the further development of diffraction computing, so it is difficult to flexibly support multi-scenario and large-scale complex artificial intelligence tasks. Against this background, realizing an optical neural network chip based on scalable scale and adjustable architecture is considered to be the only way and an important strategic trend for this technology to move towards practical applications.

[0040] The following will explain the present disclosure in detail with specific embodiments.

[0041] Figure 1 The following is a schematic structural diagram of a large-scale reconfigurable diffraction optical computing chip system provided by an embodiment of the present disclosure. As Figure 1 shown, the large-scale reconfigurable diffraction optical computing chip system includes an acquisition module, a calculation decomposition module, and a calculation module, where,

[0042] The acquisition module 101 is used to acquire a first large matrix and a second large matrix that need to be calculated;

[0043] The calculation decomposition module 102 is used to perform calculation decomposition on the first large matrix and the second large matrix to obtain a plurality of decomposed matrices;

[0044] The calculation module 103 is used to perform matrix calculation based on the plurality of decomposed matrices through a diffraction calculation array to obtain a target output result.

[0045] In one embodiment of the present disclosure, the above-mentioned large-scale reconfigurable diffractive optical computing chip system can be applicable to large matrix calculations. Based on this, the above-mentioned large-scale reconfigurable diffractive optical computing chip system can be applicable to a variety of scenarios, such as multi-modal information processing.

[0046] Among them, in one embodiment of the present disclosure, the above-mentioned large-scale reconfigurable diffractive optical computing chip system can be used for matrix calculations in neural networks.

[0047] Moreover, in one embodiment of the present disclosure, a first large matrix and a second large matrix to be calculated in a neural network can be obtained through an acquisition module. Among them, the first large matrix and the second large matrix have the same and relatively large dimensions. For example, the dimensions of the first large matrix and the second large matrix can be 10,000×10,000.

[0048] Furthermore, in one embodiment of the present disclosure, after the first large matrix and the second large matrix are obtained through the acquisition module, the first large matrix and the second large matrix can be computationally decomposed through a computational decomposition module to obtain a plurality of decomposed matrices. Specifically, in one embodiment of the present disclosure, the above-mentioned computational decomposition module can computationally decompose the first large matrix and the second large matrix through tensor decomposition to obtain a plurality of decomposed matrices. Among them, in one embodiment of the present disclosure, the dimensions of the above-mentioned decomposed matrices can be the dimensions that can be processed by a single reconfigurable diffractive computing core in the diffractive computing array.

[0049] Among them, in one embodiment of the present disclosure, the above-mentioned diffractive computing array can include a plurality of reconfigurable diffractive computing cores, and the plurality of reconfigurable diffractive computing cores can be connected longitudinally and / or transversely (as Figure 2 shown). Moreover, in one embodiment of the present disclosure, when the above-mentioned plurality of reconfigurable diffractive computing cores are connected longitudinally and transversely, at the head and tail of the input and output ports of each reconfigurable diffractive computing core in the longitudinal direction, deep cascading is performed to achieve the longitudinal scale expansion of diffractive computing and the step-by-step processing of the input signal by the diffractive core; in the transverse direction, a plurality of reconfigurable diffractive computing cores are parallelly multiplexed, in two cases of spatial parallel multiplexing of a plurality of reconfigurable diffractive computing cores and temporal parallel multiplexing of a single reconfigurable diffractive computing core, so as to achieve the transverse scale expansion of diffractive computing and the parallel processing of the input signal by the diffractive module.

[0050] Moreover, in one embodiment of the present disclosure, if the dimensions of the first large matrix and the second large matrix to be calculated are relatively large, a plurality of diffractive computing arrays can be cascaded and multiplexed to form a diffractive cluster. Figure 3 It is a schematic diagram of a large-scale reconfigurable diffractive computing cluster provided by an embodiment of the present disclosure.

[0051] Furthermore, in an embodiment of the present disclosure, in a multi-task application scenario, the large-scale reconfigurable diffraction computing cluster can be switched on demand between different tasks by introducing an adaptive mechanism and an adaptive algorithm. Specifically, when training the large-scale reconfigurable diffraction computing cluster, the number of reconfigurable diffraction cores, the number of ports, and the basic parameters of the diffraction neurons in the diffraction core (such as the shape, number, original size, and period of the diffraction neurons, etc.) can be initialized and defined according to different tasks, and the self-structural parameters of the diffraction core (such as the structural parameters of the diffraction neurons, the number of diffraction layers, etc.) and the remote control parameters are used as variables for further optimization training and simulation to obtain a large-scale reconfigurable diffraction computing cluster suitable for different tasks.

[0052] Among them, in an embodiment of the present disclosure, the connection parameters of each reconfigurable diffraction core can be dynamically modulated to break through the network structure limitation of a single reconfigurable diffraction core and achieve flexible adjustment of the computing network architecture. Among them, the dynamic modulation methods include but are not limited to on-chip modulation devices (such as Mach-Zehnder interferometer modulators, waveguide phase shifters, micro-ring modulations, etc.) and spatial modulation devices (such as SLMs, etc.), nor are they limited to the specific modulation means used, such as phase change materials, P-N junctions, and thermal electrodes, etc., nor are they limited to the specific material platforms, such as silicon, silicon nitride, lithium niobate, etc.

[0053] Furthermore, in an embodiment of the present disclosure, the reconfigurable diffraction core may include a signal feeding module 1031, a parameter modulation module 1032, a channel synthesis module 1033, a diffraction propagation module 1034, and a result output module 1035, and,

[0054] The signal feeding module 1031 is used to obtain the first decomposition matrix corresponding to the input mode and obtain the signal parameter matrix through the modulator array;

[0055] The parameter modulation module 1032 is used to obtain the second decomposition matrix of the input parameters and process the second decomposition matrix through the modulator to obtain the corresponding modulation parameter matrix;

[0056] The channel synthesis module 1033 is used to randomly fuse the modulation parameter matrix and the signal parameter matrix to obtain the fusion matrix;

[0057] The diffraction propagation module 1034 is used to input the fusion matrix into the target diffraction propagation model to obtain the diffraction transmission matrix;

[0058] The result output module 1035 is used to output the target form signal based on the diffraction transmission matrix.

[0059] Among them, in an embodiment of the present disclosure, the above input mode can be set as needed. For example, images, texts, and audios.

[0060] In one embodiment of the present disclosure, a signal parameter matrix corresponding to the first matrix can be obtained through a modulator array. Wherein, in one embodiment of the present disclosure, the above-mentioned signal parameter matrix can be an amplitude and / or phase matrix corresponding to the first matrix.

[0061] Moreover, in one embodiment of the present disclosure, the above-mentioned modulation kernel can be trained, and different tasks correspond to different target modulation parameters. Wherein, in one embodiment of the present disclosure, the implementation method of the above-mentioned modulation kernel can include but is not limited to spatial light or on-chip, and includes but is not limited to thermal modulators, carrier modulators, or phase change materials, etc., and includes but is not limited to specific material platforms such as silicon, silicon dioxide, silicon nitride, lithium niobate, etc.

[0062] Further, in one embodiment of the present disclosure, after obtaining the target modulation parameters through the above steps, the second matrix can be processed through the modulation kernel based on the target modulation parameters to obtain a corresponding modulation parameter matrix.

[0063] Wherein, in one embodiment of the present disclosure, after obtaining the modulation parameter matrix and the signal parameter matrix through the above steps, the modulation parameter matrix and the signal parameter matrix can be randomly fused to obtain a corresponding fusion matrix. In one embodiment of the present disclosure, the modulation parameter matrix and the signal parameter matrix can be randomly fused as needed, so as to achieve the arbitrary reconfigurable characteristics of the diffraction calculation results.

[0064] Moreover, in one embodiment of the present disclosure, the method of randomly fusing the modulation parameter matrix and the signal parameter matrix to obtain a fusion matrix can include: juxtaposing or cross-fusing the modulation parameter matrix and the signal parameter matrix to obtain a fusion matrix.

[0065] Further, in one embodiment of the present disclosure, after obtaining the fusion matrix through the above steps, the fusion matrix can be input into a target diffraction propagation model to obtain a diffraction transmission matrix. Wherein, in one embodiment of the present disclosure, the method of inputting the fusion matrix into the target diffraction propagation model to obtain a diffraction transmission matrix can include: inputting the fusion matrix into the target diffraction propagation model, and obtaining the diffraction transmission matrix through the target diffraction kernel in the target diffraction propagation model, where the target diffraction kernel is trained.

[0066] Moreover, in one embodiment of the present disclosure, the specific implementation manner of the above-mentioned diffraction kernel can include but is not limited to single-layer and multi-layer diffraction, and the forms of neurons can include but are not limited to various shaped neurons such as rectangles, circles, cones, ellipses, etc. Wherein, in one embodiment of the present disclosure, the corresponding target diffraction kernel can be obtained through training of the diffraction kernel to determine the implementation manner corresponding to the target diffraction kernel and the form of the neuron.

[0067] Further, in an embodiment of the present disclosure, after obtaining the diffraction transfer matrix through the above steps, a target form signal can be output based on the diffraction transfer matrix. Specifically, in an embodiment of the present disclosure, the above result output module can specifically be used to: determine the target form to be output, determine the corresponding signal feeding technique based on the target form, and output the target form signal based on the signal feeding technique and the diffraction transfer matrix.

[0068] Among them, in an embodiment of the present disclosure, the target form to be output can be determined through a task, and the target form includes but is not limited to waveguide signals, optical radiation signals, and electrical signals.

[0069] And, in an embodiment of the present disclosure, after determining the target form through the above steps, the corresponding signal feeding technique can be determined according to the target form, so as to output the target form signal based on the signal feeding technique and the diffraction transfer matrix.

[0070] In an embodiment of the present disclosure, the signal feeding techniques corresponding to different target forms are also different.

[0071] Specifically, in an embodiment of the present disclosure, the signal feeding technique corresponding to the waveguide signal can be a mode converter based on a waveguide array; the signal feeding technique corresponding to the optical radiation signal can be an optical radiator based on a coupler array; the signal feeding technique corresponding to the electrical signal can be an optoelectronic converter based on an optodetector array.

[0072] In summary, the large-scale reconfigurable diffractive optical computing chip system provided in this embodiment performs computational decomposition on large-scale computations, and performs matrix computations on the decomposed matrices through a diffractive computing array to obtain a target output result. Thus, through flexible adjustment of the diffractive computing array, flexible adjustment of the computing network architecture can be achieved, breaking through the computational scale limitation of a single reconfigurable diffractive computing core, and further greatly improving the computational scale, enabling flexible application in various large and complex application scenarios.

[0073] To implement the above embodiment, Figure 4 A large-scale reconfigurable diffractive optical computing method provided by an embodiment of the present disclosure may include the following steps:

[0074] Step 401, obtain a first large matrix and a second large matrix that need to be computed;

[0075] Step 402, perform computational decomposition on the first large matrix and the second large matrix to obtain a plurality of decomposed matrices;

[0076] Step 403, perform matrix computation through a diffractive computing array based on the plurality of decomposed matrices to obtain a target output result.

[0077] In one embodiment of the present disclosure, the above-mentioned diffraction calculation array includes a plurality of reconfigurable diffraction calculation cores, and the plurality of reconfigurable diffraction calculation cores are connected longitudinally and / or transversely.

[0078] In addition, in one embodiment of the present disclosure, the method of performing matrix calculation through a diffraction calculation array based on a plurality of decomposition matrices to obtain a target output result may include the following steps:

[0079] Step 4031: Input the plurality of decomposition matrices into the corresponding reconfigurable diffraction calculation cores in the diffraction calculation array for calculation to obtain target-form signals output by the reconfigurable diffraction calculation cores;

[0080] Step 4032: Stitch together the target-form signals output by the plurality of reconfigurable diffraction calculation cores to obtain a target output result.

[0081] In one embodiment of the present disclosure, the method of inputting the plurality of decomposition matrices into the corresponding reconfigurable diffraction calculation cores in the diffraction calculation array for calculation to obtain target-form signals output by the reconfigurable diffraction calculation cores may include the following steps:

[0082] Step 1: Obtain a first decomposition matrix corresponding to the input mode, and obtain a signal parameter matrix through a modulator array;

[0083] Step 2: Obtain a second decomposition matrix of the input parameters, and process the second decomposition matrix through a modulator to obtain a corresponding modulation parameter matrix;

[0084] Step 3: Randomly fuse the modulation parameter matrix and the signal parameter matrix to obtain a fusion matrix;

[0085] Step 4: Input the fusion matrix into a target diffraction propagation model to obtain a diffraction transmission matrix;

[0086] Step 5: Output a target-form signal based on the diffraction transmission matrix.

[0087] The collection, storage, use, processing, transmission, provision, and disclosure of the user's personal information involved in the present disclosure all comply with the provisions of relevant laws and regulations and do not violate public order and good customs.

[0088] It should be noted that personal information from users should be collected for legal and reasonable purposes and should not be shared or sold outside of such legal uses. In addition, such collection / sharing should be carried out after obtaining the informed consent of the users, including but not limited to notifying the users to read the user agreement / user notice and sign an agreement / authorization including authorizing the relevant user information before the users use the function. In addition, any necessary steps should be taken to safeguard and protect access to such personal information data and ensure that others with access to the personal information data comply with their privacy policies and procedures.

[0089] The present disclosure anticipates embodiments that can provide users with the option to block the use or access of personal information data. That is, the present disclosure anticipates that hardware and / or software can be provided to prevent or block access to such personal information data. Once the personal information data is no longer needed, the risk can be minimized by restricting data collection and deleting the data. In addition, when applicable, personal identifiers are removed from such personal information to protect the privacy of the users.

[0090] In the technical solution of the present disclosure, the acquisition, transmission, storage, use, processing, etc. of data all comply with the relevant provisions of national laws and regulations.

[0091] It should be noted that in an embodiment of the present disclosure, certain existing industry solutions such as software, components, models, etc. may be mentioned. They should be considered exemplary, and their purpose is only to illustrate the feasibility in the implementation of the technical solution of this application, but it does not mean that the applicant has already or necessarily used this solution.

[0092] In the description of the foregoing embodiments, the descriptions referring to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0093] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" can explicitly or implicitly include at least one of the features. In the description of the present disclosure, the meaning of "a plurality" is at least two, such as two, three, etc., unless otherwise specifically and clearly defined.

[0094] Any process or method description represented in a flowchart or otherwise described herein can be understood to represent a module, segment, or portion of code including one or more executable instructions for implementing a customized logic function or process, and the scope of the preferred embodiments of the present disclosure includes additional implementations in which functions may be executed not in the order shown or discussed, including in a substantially simultaneous manner according to the functions involved or in a reverse order, which should be understood by those skilled in the art to which the embodiments of the present disclosure pertain.

[0095] Logic and / or steps represented in a flowchart or otherwise described herein, for example, can be considered a sequenced list of executable instructions for implementing a logical function, and can be embodied specifically in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device. As used in this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection having one or more wires (electronic device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable medium on which a program can be printed, as the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpretation, or otherwise processing as appropriate, and then stored in a computer memory.

[0096] It should be understood that various parts of the present disclosure can be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one of the following techniques known in the art or a combination thereof can be used: discrete logic circuits having logic gates for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), and the like.

[0097] Those of ordinary skill in the art can understand that all or part of the steps carried out in implementing the above-described embodiment methods can be completed by a program instructing relevant hardware. The program can be stored in a computer-readable storage medium, and when the program is executed, it includes one or a combination of the steps of the method embodiment.

[0098] In addition, in each of the various embodiments of the present disclosure, each functional unit may be integrated in a processing module, may exist separately physically as each unit, or two or more units may be integrated in a module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0099] The above-mentioned storage medium may be a read-only memory, a magnetic disk, an optical disc, etc. Although the embodiments of the present disclosure have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present disclosure. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present disclosure.

Claims

1. A large-scale reconfigurable diffraction optical computing chip system, characterized in that: The chip system includes an acquisition module, a calculation decomposition module and a calculation module, wherein: The acquisition module is used to acquire the first largest matrix and the second largest matrix that need to be calculated; The calculation decomposition module is used to perform calculation decomposition on the first large matrix and the second large matrix to obtain a plurality of decomposition matrices; The calculation module is used to perform matrix calculation based on the multiple decomposition matrices through a diffraction calculation array to obtain a target output result; The diffraction calculation array includes a plurality of reconfigurable diffraction calculation cores, which are connected vertically and / or horizontally, wherein the reconfigurable diffraction calculation core includes a signal feeding module, a parameter modulation module, a channel synthesis module, a diffraction propagation module and a result output module. The signal feeding module is used to obtain a first decomposition matrix corresponding to the input mode and obtain a signal parameter matrix through a modulator array; The parameter modulation module is used to obtain a second decomposition matrix of the input parameters, and process the second decomposition matrix through a modulation kernel to obtain a corresponding modulation parameter matrix; The channel synthesis module is used to randomly fuse the modulation parameter matrix and the signal parameter matrix to obtain a fusion matrix; The diffraction propagation module is used to input the fusion matrix into the target diffraction propagation model to obtain a diffraction transmission matrix; The result output module is used to output a target form signal based on the diffraction transmission matrix.

2. The system according to claim 1, characterized in that The channel synthesis module is specifically used to: perform parallel fusion or cross fusion of the modulation parameter matrix and the signal parameter matrix to obtain a fusion matrix.

3. The system according to claim 1, characterized in that The diffraction propagation module is specifically used to: input the fusion matrix into the target diffraction propagation model, and obtain the diffraction transmission matrix through the target diffraction kernel in the target diffraction propagation model, wherein the target diffraction kernel is trained.

4. The system according to claim 1, characterized in that The result output module is specifically used for: Determine the target format for output; Based on the target form, determining a corresponding signal feeding technology; Based on the signal feeding technology and the diffraction transmission matrix, a target form signal is output.

5. The system according to claim 1, characterized in that The calculation decomposition module is specifically used for: The first large matrix and the second large matrix are computationally decomposed by tensor decomposition to obtain a plurality of decomposed matrices.

6. A large-scale reconfigurable diffraction light calculation method, characterized in that: The method comprises: Get the first largest matrix and the second largest matrix to be calculated; Decomposing the first largest matrix and the second largest matrix by calculation to obtain a plurality of decomposed matrices; Based on the multiple decomposition matrices, matrix calculation is performed through a diffraction calculation array to obtain a target output result, wherein the diffraction calculation array includes multiple reconfigurable diffraction calculation cores, and the multiple reconfigurable diffraction calculation cores are connected vertically and / or horizontally to obtain a first decomposition matrix corresponding to an input mode, and a signal parameter matrix is ​​obtained through a modulator array, a second decomposition matrix of input parameters is obtained, and the second decomposition matrix is ​​processed through a modulation core to obtain a corresponding modulation parameter matrix, the modulation parameter matrix and the signal parameter matrix are randomly fused to obtain a fused matrix, the fused matrix is ​​input into a target diffraction propagation model to obtain a diffraction transmission matrix, and based on the diffraction transmission matrix, a target form signal is output, and the target form signals output by multiple reconfigurable diffraction calculation cores are spliced ​​to obtain a target output result.

Images