Reverse design method of optical computing chip

Abstract

The invention relates to a reverse design method of an optical computing chip. The method comprises the following steps: determining a design optimization function of the optical computing chip according to an input optical field and an output optical field of the optical computing chip; optimizing a plurality of optimizable parameters of the optical computing chip, and calculating a current optimization function value of one optical computing chip according to the design optimization function of the optical computing chip after each optimizable parameter is optimized; judging whether the current optimization function value of the optical computing chip meets a preset optimization requirement of the optical computing chip or not; if the optimization function value of the optical calculation chip does not meet the preset optimization requirement of the optical calculation chip, the multiple optimizable parameters of the optical calculation chip are optimized again, and the optimization function value of the optical calculation chip is calculated till the optimization function value of the optical calculation chip meets the preset optimization requirement of the optical calculation chip. At least the structure of the optical computing chip can be optimized, and the effect of improving the performance of the optical computing chip is achieved.

Description

Technical Field

[0001] The present application relates to the technical field of optical computing chip design, and in particular to a reverse design method for an optical computing chip. Background Art

[0002] As the scale of networks expands and computing demands increase, the performance of traditional electronic computing platforms is gradually unable to meet the demand. The traditional von Neumann architecture shows obvious bottlenecks when processing these computing tasks, especially in the training of large-scale neural networks. Although the graphics processing unit (GPU) has alleviated this problem to a certain extent due to its excellent parallel computing capabilities, the demand for computing power in deep learning is almost unlimited. The challenges faced by traditional electronic technology include electrical signal interference, excessive energy consumption and physical limits, which limit further performance improvements. Optical computing technology, as a potential alternative, has shown significant advantages. The propagation speed of light reaches 300,000 kilometers per second, which is about 300 times faster than electronic signals. Optical systems can provide about 2*10^4 times higher information carrying capacity than traditional electrical channels, and the high parallelism and strong anti-interference of optical signals give them significant advantages in information transmission and processing. These characteristics make replacing electricity with light a promising research direction. In this context, optical neural networks (ONN) have received widespread attention as an emerging technology. ONN uses the high-speed transmission and highly parallel processing capabilities of light to effectively accelerate deep learning computing tasks. Optical neural networks solve the challenges of intensive computing in traditional neural networks by performing matrix multiplication at the speed of light, and use nonlinear optical elements to implement nonlinear computing in ANNs. Once trained, ONNs can process optical signals at the speed of light without the need for additional energy input, significantly improving computing efficiency. In recent years, advances in integrated photonic technology and nanophotonic processors have made the implementation of optical neural networks more feasible, marking a broad application prospect for optical computing technology in the field of deep learning.

[0003] Optical computing chips need to handle a large number of computing tasks and data channels. Wide-bandwidth devices support higher data transmission rates and allow optical signals of different wavelengths to be processed in parallel. This provides more flexibility for the chip, enabling it to handle complex computing and data tasks and support a variety of optical modulation and processing methods. Wide-bandwidth devices can also reduce signal distortion or loss caused by wavelength changes, making the system more robust to wavelength changes, thereby improving the stability and reliability of computing. Low-loss devices can significantly reduce the attenuation of optical signals during transmission, thereby maintaining the strength and quality of the signal. This not only reduces the need for additional signal amplification and reduces the energy consumption of the system, but also improves overall energy efficiency, extends the service life of the equipment and reduces operating costs. Low-loss devices also reduce the need for reprocessing due to signal attenuation, making the computing process more efficient. High-quality signal transmission reduces errors and the need for retransmission, thereby improving computing speed and system response time.

[0004] However, the working bandwidth of existing optical computing chips is insufficient and the device loss is large. The working bandwidth of existing optical computing chips is generally 50 nm, and the device loss is 0.5-1 dB, which has a great impact on the performance of optical computing chips. Summary of the invention

[0005] Based on this, it is necessary to provide a reverse design method for an optical computing chip to address the above technical problems, which can at least optimize the structure of the optical computing chip and achieve the effect of improving the performance of the optical computing chip.

[0006] In a first aspect, the present application provides a reverse design method for an optical computing chip, comprising: determining a design optimization function of the optical computing chip based on an input light field and an output light field of the optical computing chip; optimizing multiple optimizable parameters of the optical computing chip, and after optimizing each optimizable parameter, calculating a current optimization function value of the optical computing chip based on the design optimization function of the optical computing chip; judging whether the current optimization function value of the optical computing chip meets a preset optical computing chip optimization requirement; if the optimization function value of the optical computing chip does not meet the preset optical computing chip optimization requirement, re-optimizing multiple optimizable parameters of the optical computing chip and calculating the optimization function value of the optical computing chip until the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirement; if the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirement, determining the optical computing chip structure corresponding to the optimization function value of the optical computing chip as the target optical computing chip structure.

[0007] In one embodiment, the optical computing chip includes Optical beam splitters and Optical beam splitter; wherein, determining The steps of optimizing multiple function values ​​of the optical beam splitter device include: determining, in a predetermined optimization space, a first influence value of multiple analysis points in the optimization space on the output light field according to an optimization influence formula; determining a second influence value of different graphic areas on the output light field according to the first influence value of the multiple analysis points in the optimization space on the output light field; determining, according to the second influence value of different graphic areas on the output light field, The basic structure of the optical beam splitter device; the boundary of the basic structure is divided into multiple segments, and one or more of the multiple segments are adjusted multiple times, and after each adjustment, the calculation The current optimization function value of the optical beam splitter.

[0008] In one embodiment, the step of determining whether the current optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements includes: determining whether the current optimization function value is greater than the optimization function value after the optimization parameters were adjusted last time; if the current optimization function value is greater than the optimization function value after the optimization parameters were adjusted last time, determining the optical computing chip structure corresponding to the current optimization function value; determining whether the current optimization function value is greater than or equal to the preset The target optimization value of the optical beam splitter device; if the current optimization function value is greater than or equal to the preset target optimization value of the optical computing chip, it is determined that the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements; the optical computing chip structure corresponding to the optimization function value that meets the preset optical computing chip optimization requirements is determined as the target Optical beam splitting device structure.

[0009] In one embodiment, the method further includes: if the current optimization function value is less than or equal to the optimization function value after the last adjustment of the optimization parameters, multiple adjustments are made to one or more segments in the basic structure until the adjusted current optimization function value is greater than the optimization function value after the last adjustment of the optimization parameters.

[0010] In one embodiment, the method further includes: if the current optimization function value is less than or equal to a preset optical computing chip target optimization value, performing multiple adjustments in one or more sections in the basic structure until the adjusted current optimization function value is greater than or equal to the preset optical computing chip target optimization value.

[0011] In one embodiment, determining The steps of the optimization function of the optical beam splitter device include: dividing the area to be optimized into blocks based on a preset coding accuracy, and encoding each block to obtain multiple coding areas; calculating the third influence value of each coding area on the output light field; randomly assigning values ​​to the multiple coding areas respectively to obtain multiple assigned coding structures, the assignments include a first assignment and a second assignment, the first assignment indicates that the coding area is a material, and the second assignment indicates that the coding area is empty; calculating the structural optimization function value of each assigned coding structure; determining the weight value of each assigned coding structure according to the structural optimization function value of each assigned coding structure; judging whether there is a structural optimization function value of multiple assigned coding structures that is consistent with the preset The structure optimization function value with the same target optimization value of the optical beam splitter device; if the structure optimization function values ​​of multiple assignment coding structures do not exist with the preset The structure optimization function value of the optical beam splitter device is the same as the target optimization value, then the updated assignment coding structure is generated according to the weight value of each assignment coding structure until the structure optimization function value of the updated assignment coding structure is consistent with the preset The target optimization values ​​of the optical beam splitter components are the same.

[0012] In one of the embodiments, the method also includes: assigning values ​​to multiple target coding regions in the updated assigned coding structure, the multiple target coding regions being coding regions among the multiple coding regions that have a positive impact on the output light field; calculating the structural optimization function value of each assigned coding structure and the updated assigned coding structure; and determining the weight value of each assigned coding structure and the updated assigned coding structure according to the structural optimization function value of each assigned coding structure and the updated assigned coding structure.

[0013] In one of the embodiments, the weight value of each assignment coding structure and / or updated assignment coding structure is determined by the following steps: based on the optimization function value of each assignment coding structure and / or updated assignment coding structure, the optimization function ratio value of each assignment coding structure and / or updated assignment coding structure is determined; based on the optimization function ratio value of each assignment coding structure and / or updated assignment coding structure, the weight value of each assignment coding structure and / or updated assignment coding structure is determined.

[0014] In one embodiment, the method further comprises: The optical computing chip structure corresponding to the same structural optimization function value as the target optimization value of the optical beam splitter device is determined as the target Optical beam splitting device structure.

[0015] In one embodiment, the design optimization function of the optical computing chip is:

[0016] ;

[0017] Among them, FOM represents the design optimization function of the optical computing chip. Indicates the number of wavelengths of the optical computing chip, represents the transmittance of an output port of the optical computing chip, represents the transmittance of another output port of the optical computing chip, represents the i-th wavelength, represents the output at the i-th wavelength.

[0018] In a second aspect, the present application also provides a reverse design device for an optical computing chip, comprising a design optimization function determination module, an optimization function value calculation module, an optimization demand judgment module, an optical chip structure determination module, and a parameter optimization module, wherein the design optimization function determination module is used to determine the design optimization function of the optical computing chip according to the input light field and the output light field of the optical computing chip, the optimization function value calculation module is used to optimize multiple optimizable parameters of the optical computing chip, and after optimizing each optimizable parameter, calculate the current optimization function value of the optical computing chip according to the design optimization function of the optical computing chip, and the optimization demand judgment module is used to judge the optical computing chip structure. The optical computing chip structure determination module is used to determine whether the current optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements. If the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements, the optical computing chip structure corresponding to the optimization function value of the optical computing chip is determined as the target optical computing chip structure. The parameter optimization module is used to re-optimize multiple optimizable parameters of the optical computing chip and calculate the optimization function value of the optical computing chip if the optimization function value of the optical computing chip does not meet the preset optical computing chip optimization requirements, until the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements.

[0019] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0020] Determine a design optimization function of the optical computing chip according to the input light field and the output light field of the optical computing chip;

[0021] Optimizing multiple optimizable parameters of the optical computing chip, and calculating a current optimization function value of the optical computing chip according to a design optimization function of the optical computing chip after optimizing each optimizable parameter;

[0022] Determine whether the current optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements;

[0023] If the optimization function value of the optical computing chip does not meet the preset optimization requirements of the optical computing chip, the multiple optimizable parameters of the optical computing chip are optimized again, and the optimization function value of the optical computing chip is calculated until the optimization function value of the optical computing chip meets the preset optimization requirements of the optical computing chip;

[0024] If the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements, the optical computing chip structure corresponding to the optimization function value of the optical computing chip is determined as the target optical computing chip structure.

[0025] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the following steps are implemented:

[0026] Determine a design optimization function of the optical computing chip according to the input light field and the output light field of the optical computing chip;

[0027] Optimizing multiple optimizable parameters of the optical computing chip, and calculating a current optimization function value of the optical computing chip according to a design optimization function of the optical computing chip after optimizing each optimizable parameter;

[0028] Determine whether the current optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements;

[0029] If the optimization function value of the optical computing chip does not meet the preset optimization requirements of the optical computing chip, the multiple optimizable parameters of the optical computing chip are optimized again, and the optimization function value of the optical computing chip is calculated until the optimization function value of the optical computing chip meets the preset optimization requirements of the optical computing chip;

[0030] If the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements, the optical computing chip structure corresponding to the optimization function value of the optical computing chip is determined as the target optical computing chip structure.

[0031] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0032] Determine a design optimization function of the optical computing chip according to the input light field and the output light field of the optical computing chip;

[0033] Optimizing multiple optimizable parameters of the optical computing chip, and calculating a current optimization function value of the optical computing chip according to a design optimization function of the optical computing chip after optimizing each optimizable parameter;

[0034] Determine whether the current optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements;

[0035] If the optimization function value of the optical computing chip does not meet the preset optimization requirements of the optical computing chip, the multiple optimizable parameters of the optical computing chip are optimized again, and the optimization function value of the optical computing chip is calculated until the optimization function value of the optical computing chip meets the preset optimization requirements of the optical computing chip;

[0036] If the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements, the optical computing chip structure corresponding to the optimization function value of the optical computing chip is determined as the target optical computing chip structure.

[0037] The above-mentioned reverse design method of the optical computing chip determines the design optimization function of the optical computing chip according to the input light field and output light field of the optical computing chip; calculates the current optimization function value according to multiple optimizable parameters of the optical computing chip; and judges whether the current optical chip structure meets the pre-set optical computing chip optimization requirements according to the current optimization function value; and optimizes the parameters of the optical computing chip structure that does not meet the optical computing chip optimization requirements, so as to at least optimize the structure of the optical computing chip and achieve the effect of improving the performance of the optical computing chip. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments of the present application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0039] Figure 1 A schematic diagram of a flow chart of a reverse design method of an optical computing chip according to an embodiment;

[0040] Figure 2 is a schematic flow chart of a reverse design method of an optical computing chip in another embodiment;

[0041] Figure 3 is a schematic flow chart of a reverse design method of an optical computing chip in yet another embodiment;

[0042] Figure 4 is a schematic diagram of an optimized structure of a beam splitter device in one embodiment;

[0043] Figure 5 A waveform diagram of output power at both ends varying with wavelength in one embodiment;

[0044] Figure 6 is a schematic diagram of device insertion loss in one embodiment;

[0045] Figure 7is a schematic diagram of an optimized structure of a beam splitter device in another embodiment;

[0046] Figure 8 A waveform diagram of output power at both ends varying with wavelength in another embodiment;

[0047] Fig. 9 is a schematic diagram of device insertion loss in another embodiment;

[0048] Fig.10 A schematic diagram of the structure of an optical computing chip in one embodiment;

[0049] Fig.11 In one embodiment, the input light Schematic diagram of the output light of the optical beam splitter device;

[0050] Fig.12 Schematic diagram of output light after passing through an optical computing chip in one embodiment

[0051] Fig.13 is a structural block diagram of a reverse design device for an optical computing chip in one embodiment;

[0052] Fig.14 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment.

[0053] Reference numerals and descriptions:

[0054] 901, light source; 902, phase modulation area; 903, FPGA module; 904, photoelectric detection module. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0056] The reverse design method of an optical computing chip provided in the embodiment of the present application can be applied to the technical field of optical computing chip design.

[0057] Please refer to Figure 1 In an exemplary embodiment, a reverse design of an optical computing chip is provided, including the following steps S101 to S105. Among them:

[0058] Step S101: determining a design optimization function of the optical computing chip according to an input light field and an output light field of the optical computing chip.

[0059] Specifically, the design optimization function of the optical computing chip is:

[0060] ;

[0061] Among them, FOM represents the design optimization function of optical computing chip, Indicates the number of wavelengths of the optical computing chip, represents the transmittance of an output port of the optical computing chip, represents the transmittance of another output port of the optical computing chip, represents the i-th wavelength, represents the output at the i-th wavelength.

[0062] Step S102, optimizing a plurality of optimizable parameters of the optical computing chip, and calculating a current optimization function value of the optical computing chip according to a design optimization function of the optical computing chip after optimizing each optimizable parameter.

[0063] Wherein, the optical computing chip includes Optical beam splitters and Optical beam splitting device.

[0064] Specific, sure The steps of optimizing multiple function values ​​of the optical beam splitter device include: determining, in a predetermined optimization space, a first influence value of multiple analysis points in the optimization space on the output light field according to an optimization influence formula; determining a second influence value of different graphic areas on the output light field according to the first influence value of the multiple analysis points in the optimization space on the output light field; determining the second influence value of different graphic areas on the output light field according to the second influence value of different graphic areas on the output light field. The basic structure of the optical beam splitter device; dividing the boundary of the basic structure into multiple segments, and adjusting one and / or multiple segments of the multiple segments multiple times, and calculating after each adjustment The current optimization function value of the optical beam splitter.

[0065] Specific, sure The steps of the optimization function of the optical beam splitter device include: based on a preset coding accuracy, dividing the area to be optimized into blocks, and encoding each block to obtain multiple coding areas; calculating the third influence value of each coding area on the output light field; randomly assigning values ​​to the multiple coding areas respectively to obtain multiple assigned coding structures, the assignments include a first assignment and a second assignment, the first assignment indicates that the coding area is a material, and the second assignment indicates that the coding area is empty; calculating the structural optimization function value of each assigned coding structure; determining the weight value of each assigned coding structure according to the structural optimization function value of each assigned coding structure; judging whether there is a structural optimization function value of multiple assigned coding structures that is consistent with the preset The structure optimization function value with the same target optimization value of the optical beam splitter device; if the structure optimization function values ​​of multiple assignment coding structures do not exist with the preset The structure optimization function value of the optical beam splitter device is the same as the target optimization value, then the updated assignment coding structure is generated according to the weight value of each assignment coding structure until the structure optimization function value of the updated assignment coding structure is consistent with the preset The target optimization values ​​of the optical beam splitter components are the same.

[0066] Step S103, determining whether the current optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements.

[0067] Specifically, for The optical beam splitting device comprises the steps of judging whether the current optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements, comprising: judging whether the current optimization function value is greater than the optimization function value after the optimization parameter is adjusted last time; if the current optimization function value is greater than the optimization function value after the optimization parameter is adjusted last time, determining the optical computing chip structure corresponding to the current optimization function value; judging whether the current optimization function value is greater than or equal to the preset The target optimization value of the optical beam splitter device; if the current optimization function value is greater than or equal to the preset target optimization value of the optical computing chip, it is determined that the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirement; the optical computing chip structure corresponding to the optimization function value that meets the preset optical computing chip optimization requirement is determined as the target Optical beam splitting device structure.

[0068] If the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirement, step S104 is executed to determine the optical computing chip structure corresponding to the optimization function value of the optical computing chip as the target optical computing chip structure.

[0069] If the optimization function value of the optical computing chip does not meet the preset optimization requirements of the optical computing chip, step S105 is executed to re-optimize the multiple optimizable parameters of the optical computing chip and calculate the optimization function value of the optical computing chip until the optimization function value of the optical computing chip meets the preset optimization requirements of the optical computing chip.

[0070] Specifically, for The optical beam splitter device structure, the method also includes: if the current optimization function value is less than or equal to the optimization function value after the optimization parameters were adjusted last time, multiple adjustments are made to one or more sections in the basic structure until the adjusted current optimization function value is greater than the optimization function value after the optimization parameters were adjusted last time.

[0071] Specifically, for The optical beam splitting device structure, the method also includes: if the current optimization function value is less than or equal to a preset optical computing chip target optimization value, a section and / or multiple sections in the basic structure are adjusted multiple times until the adjusted current optimization function value is greater than or equal to the preset optical computing chip target optimization value.

[0072] Specifically, for The optical beam splitter device structure, the method also includes: assigning values ​​to multiple target coding areas in the updated assigned coding structure, the multiple target coding areas are coding areas in the multiple coding areas that have a positive impact on the output light field; calculating the structural optimization function value of each assigned coding structure and the updated assigned coding structure; and determining the weight value of each assigned coding structure and the updated assigned coding structure according to the structural optimization function value of each assigned coding structure and the updated assigned coding structure.

[0073] Specifically, for The optical splitter device structure determines the weight value of each assignment coding structure and / or the updated assignment coding structure through the following steps: according to the optimized function value of each assignment coding structure and / or the updated assignment coding structure, the optimized function ratio value of each assignment coding structure and / or the updated assignment coding structure is determined; according to the optimized function ratio value of each assignment coding structure and / or the updated assignment coding structure, the weight value of each assignment coding structure and / or the updated assignment coding structure is determined.

[0074] Specifically, for The optical beam splitter device structure, the method further comprises: The optical computing chip structure corresponding to the same structural optimization function value as the target optimization value of the optical beam splitter device is determined as the target Optical beam splitting device structure.

[0075] As an example, see Figure 2 In an exemplary embodiment, there is provided a The reverse design method of an optical beam splitter device includes the following steps S201 to S210.

[0076] Step S201, determining the input and output light fields, and determining the optimization function.

[0077] The optimization function is the design optimization function of the optical computing chip.

[0078] Specifically, the design optimization function of the optical computing chip is:

[0079] ;

[0080] Among them, FOM represents the design optimization function of optical computing chip, Indicates the number of wavelengths of the optical computing chip, represents the transmittance of an output port of the optical computing chip, represents the transmittance of another output port of the optical computing chip, represents the i-th wavelength, represents the output at the i-th wavelength.

[0081] It should be noted that step S201 is to determine the basic simulation conditions. The optimization function is designed according to the target. The multiplication of the first value of FOM by 10 increases the weight, hoping to focus on optimizing the value in the square brackets [ ]. The latter value is inherent. The goal is to make the output of the two output ports consistent and the power maximum (maximum 0.5). T1 and T2 are the transmittances of the two port outputs. T will correspond to different values ​​according to different input wavelengths. Here n is the number of wavelengths. The ultimate goal is to obtain the minimum value of FOM.

[0082] Step S202, determining the optimization space and analyzing the influence of each point in the space on the output.

[0083] Specifically, the optimization space can be determined according to the following formula:

[0084] ;

[0085] in, In order to optimize the influence of a point in space on the output light field, To optimize a point in space, is a field function.

[0086] Here, the optimization space will be determined before optimization. For example, in a The optimization is performed within a square of different sizes. The optimization area of ​​the simulation can be formulated according to different needs. The shape and coordinates of the simulation area are provided. In this paper, an arbitrary point is set in the optimization area to calculate the change of the output light field. , which can be summarized into a general formula to represent the influence of any point on the output light field.

[0087] Step S203, constructing a device structure that basically meets the output.

[0088] Specifically, the effect of the graph on the output can be deduced by analyzing the effect of the point on the output: , and then build a device structure that basically meets the output requirements.

[0089] Here, through the analysis of step S202, the influence of an area on the output light field can be derived. Based on the above conditions, a structure that basically meets the output target is constructed by continuously iterating different shapes. This method can quickly obtain the basic structure and save simulation time.

[0090] Step S204, dividing the device boundary into several segments (accuracy is self-determined).

[0091] Here, the obtained basic structure is further optimized, and the boundary of the obtained structure is subdivided into several small units.

[0092] Among them, the more segments there are, the higher the accuracy of the device.

[0093] Step S205: randomly select a boundary for fine-tuning.

[0094] Here, fine-tuning refers to changing the boundary conditions in the micrometer range while maintaining the smoothness of the boundary.

[0095] Specifically, one or several small units may be randomly modified to fine-tune the structural shape, for example, the amplitude of the change is within 1 micron, while maintaining the smoothness of the boundary shape to reduce losses.

[0096] Step S206, calculating the corresponding optimization function.

[0097] Here, the optimization function is the current optimization function.

[0098] As an example, each change may correspond to a value of a different wavelength of T1 or T2, which may be brought into FOM to calculate the optimization function value.

[0099] Step S207, determining whether the current optimization function is greater than the previous optimization function.

[0100] Specifically, the optimization function value of the new structure can be compared with the optimization function value of the previous structure.

[0101] If the current optimized function is greater than the previous optimized function, step S208 is executed to save the corresponding structure, and the process returns to re-execute step S205.

[0102] If the current optimization function is less than or equal to the previous optimization function, step S209 is executed to determine whether the optimization requirement is met.

[0103] If the optimization requirements are met, step S210 is executed to end the optimization and save the parameters.

[0104] If the optimization requirement is not met, step S205 is executed again.

[0105] In this way, by comparing the FOM values ​​obtained each time the shape changes, because the minimum value is required, the structure is retained when the value becomes smaller, and the process returns to step S205. Otherwise, it is first determined whether the FOM value meets the target. If it meets the requirements, the optimization is terminated and the structure is saved. Otherwise, the process returns to step S205.

[0106] For example, see Figure 4-Figure 6 , Figure 4-Figure 6 Output optimized Schematic diagram of the structure of an optical beam splitter, the variation of power with wavelength, and the insertion loss of the device.

[0107] In this way, the application first analyzes the impact of the point on the output and then expands it to the impact of the face on the output. On this basis, the device structure is optimized, which greatly saves the optimization time and improves the optimization efficiency. It also serves as a reference for subsequent optimization. The optimized structural uniformity is The insertion loss is Below, the performance is superior.

[0108] As an example, see Figure 3 In an exemplary embodiment, there is provided a The reverse design method of an optical beam splitter device includes the following steps S301 to S310.

[0109] Step S301, determining input and output light fields, and determining an optimization function.

[0110] The optimization function is the design optimization function of the optical computing chip.

[0111] Specifically, the design optimization function of the optical computing chip is:

[0112] ;

[0113] Among them, FOM represents the design optimization function of optical computing chip, Indicates the number of wavelengths of the optical computing chip, represents the transmittance of an output port of the optical computing chip, represents the transmittance of another output port of the optical computing chip, represents the i-th wavelength, represents the output at the i-th wavelength.

[0114] Step S302, encoding the optimized area (customized precision).

[0115] Specifically, this step is to divide the optimization area into a number of small squares of regular shapes. The more squares are divided, the higher the accuracy. At the same time, the small squares are encoded to facilitate the assignment in subsequent steps.

[0116] Step S303: Analyze the impact of each coding region on the output.

[0117] Specifically, the impact of each encoding region on the output can be analyzed by the following formula:

[0118] ;

[0119] Among them, among them, In order to optimize the influence of a point in space on the output light field, To optimize a point in space, is a field function.

[0120] Here, you can The same simulation method as the reverse design method of the optical beam splitter device is used to calculate the influence of each small square on the output light field, and the small squares that have a positive influence on the output light field are recorded, which will not be repeated here.

[0121] Step S304: randomly assigning several groups of values ​​to the coding region.

[0122] Exemplarily, the coding area can be randomly assigned 0 or 1, where 1 is material and 0 is etching. Several groups are set as initial optimization groups. The larger the number of groups in the group, the higher the optimization accuracy.

[0123] Among them, all small squares are randomly assigned 0 or 1. The area assigned 0 is air, that is, it is etched away, and the area assigned 1 is the device. All small squares assigned 1 constitute the device. Multiple groups of different values ​​will be generated here to assign values ​​to the coding area, that is, multiple different device structures will be generated.

[0124] Step S305, calculating the optimization function value of each group.

[0125] Here, each device structure corresponds to an optimization value.

[0126] S306, assigning a corresponding weight to each function value, wherein a function value that meets or is close to the target optimization has a higher weight.

[0127] Here, corresponding coding group weights are assigned according to different optimization values.

[0128] As an example, if ten coding groups are generated, the function values ​​corresponding to each coding group are , function value ratio , the corresponding weights are represented by ,Each interval corresponds to the function value one by one, and the value closest to the target optimization will account for the largest proportion.

[0129] Step S307, determining whether the optimization function value is equal to the objective function value.

[0130] If they are equal, step S308 is executed to save the corresponding structure and the optimization is successful.

[0131] If they are not equal, step S309 is executed to generate a new coding group according to the weight.

[0132] Execute step S310 to randomly select and assign values ​​to the regions that have output positive images, and re-execute step S305 until the optimization is completed.

[0133] Specifically, in order to find the function value closest to the optimization objective, it is determined whether the optimal value in each group is equal to the target value. If they are equal, the structure is preserved and the optimization ends. Otherwise, a group of random numbers ranging from 0 to 1 is generated, and a coding group with corresponding weights is randomly selected as the initial group. Then, part of the coding of the initial group is exchanged, and several coding groups are selected at the same time. The coding areas that have a positive impact on the output recorded in step S303 are compared, and the corresponding coding areas of the selected coding groups are assigned values ​​at the optimization stage.

[0134] See also Figure 7-Figure 9 , Figure 7-Figure 9 The optimized structure and effect diagram are shown. First, analyze the impact of each coding region on the output, then generate groups of coding regions, select the better groups for reorganization optimization by weighting the optimization function value, and randomly select coding regions that have a positive impact on the output for assignment based on the impact of each coding region on the output. Repeat this cycle until the optimization is completed. Structural uniformity after optimization The insertion loss is Below, excellent performance

[0135] See also Fig.10 , Fig.10 A schematic diagram of the structure of an optical computing chip in an embodiment is shown. The optical computing chip is composed of multiple modules such as a light source, a phase modulation area, an FPGA module, and a photoelectric detection module.

[0136] Here, the present application mainly replaces the dotted rectangular part through the reverse-engineered beam splitter device. The chip is only a small part. There are a large number of beam splitters in the visible light computing chip. The device designed in this application has the characteristics of large bandwidth, low loss, and compact structure, which can realize large-scale photonic chip integration. At the same time, it enables the photonic chip to work at a larger bandwidth, that is, the parallel transmission effect is good, the calculation amount is large and the transmission speed is fast. The characteristics of low insertion loss make the optical computing chip have more efficient computing power and reduced power consumption.

[0137] See also Figure 11-Figure 12 , Fig.11 For light to pass through The waveform of the initial input light after the beam splitter is shown in Figure 1. Fig.12 For light to pass through Fig.10 The waveform of the output light after the optical computing chip is shown.

[0138] For details, please refer to Fig.11 , Fig.11 There are two lines in the graph, but they overlap so only one line is shown in the graph, which shows the good performance of the beam splitter.

[0139] For details, please refer to Fig.12, because of the loss of the device, the intensity of the output light decreases in accordance with the basic principle, and the control of the phase realizes The light output can be changed by changing the phase to achieve multiple outputs.

[0140] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0141] Based on the same inventive concept, the embodiment of the present application also provides a reverse design device for an optical computing chip for implementing the reverse design method for an optical computing chip involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the embodiment of the reverse design device for an optical computing chip provided below can be referred to the limitations of the reverse design method for an optical computing chip above, and will not be repeated here.

[0142] Please refer to Fig.13In an exemplary embodiment, a reverse design device for an optical computing chip is provided, comprising: a design optimization function determination module 1301, an optimization function value calculation module 1302, an optimization requirement judgment module 1303, an optical chip structure determination module 1304 and a parameter optimization module 1305, wherein the design optimization function determination module is used to determine the design optimization function of the optical computing chip according to the input light field and the output light field of the optical computing chip, the optimization function value calculation module is used to optimize multiple optimizable parameters of the optical computing chip, and after optimizing each optimizable parameter, according to the design optimization function of the optical computing chip, calculate the current optimization function value of the optical computing chip, the optimization requirement judgment module 1303, and the optical chip structure determination module 1304. The judgment module is used to judge whether the current optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements. The optical chip structure determination module is used to determine the optical computing chip structure corresponding to the optimization function value of the optical computing chip as the target optical computing chip structure if the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements. The parameter optimization module is used to re-optimize multiple optimizable parameters of the optical computing chip and calculate the optimization function value of the optical computing chip if the optimization function value of the optical computing chip does not meet the preset optical computing chip optimization requirements, until the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements.

[0143] Each module in the above-mentioned reverse design device of the optical computing chip can be implemented in whole or in part by software, hardware and their combination. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the corresponding operations of each of the above modules.

[0144] In an exemplary embodiment, a computer device is provided, which may be a terminal. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, the memory, and the input / output interface are connected via a system bus, and the communication interface, the display unit, and the input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, near field communication (NFC) or other technologies. When the computer program is executed by the processor, a reverse design method of an optical computing chip is implemented. The display unit of the computer device is used to form a visually visible picture, which may be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0145] Those skilled in the art will understand that Fig.14 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0146] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0147] Determining a design optimization function of the optical computing chip according to an input light field and an output light field of the optical computing chip;

[0148] Optimizing a plurality of optimizable parameters of the optical computing chip, and calculating a current optimization function value of the optical computing chip according to a design optimization function of the optical computing chip after optimizing each optimizable parameter;

[0149] Determining whether the current optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements;

[0150] If the optimization function value of the optical computing chip does not meet the preset optimization requirements of the optical computing chip, re-optimize the multiple optimizable parameters of the optical computing chip and calculate the optimization function value of the optical computing chip until the optimization function value of the optical computing chip meets the preset optimization requirements of the optical computing chip;

[0151] If the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirement, the optical computing chip structure corresponding to the optimization function value of the optical computing chip is determined as the target optical computing chip structure.

[0152] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0153] Determining a design optimization function of the optical computing chip according to an input light field and an output light field of the optical computing chip;

[0154] Optimizing a plurality of optimizable parameters of the optical computing chip, and calculating a current optimization function value of the optical computing chip according to a design optimization function of the optical computing chip after optimizing each optimizable parameter;

[0155] Determining whether the current optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements;

[0156] If the optimization function value of the optical computing chip does not meet the preset optimization requirements of the optical computing chip, re-optimize the multiple optimizable parameters of the optical computing chip and calculate the optimization function value of the optical computing chip until the optimization function value of the optical computing chip meets the preset optimization requirements of the optical computing chip;

[0157] If the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirement, the optical computing chip structure corresponding to the optimization function value of the optical computing chip is determined as the target optical computing chip structure.

[0158] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0159] Determining a design optimization function of the optical computing chip according to an input light field and an output light field of the optical computing chip;

[0160] Optimizing a plurality of optimizable parameters of the optical computing chip, and calculating a current optimization function value of the optical computing chip according to a design optimization function of the optical computing chip after optimizing each optimizable parameter;

[0161] Determining whether the current optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements;

[0162] If the optimization function value of the optical computing chip does not meet the preset optimization requirements of the optical computing chip, re-optimize the multiple optimizable parameters of the optical computing chip and calculate the optimization function value of the optical computing chip until the optimization function value of the optical computing chip meets the preset optimization requirements of the optical computing chip;

[0163] If the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirement, the optical computing chip structure corresponding to the optimization function value of the optical computing chip is determined as the target optical computing chip structure.

[0164] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0165] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0166] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A reverse design method for an optical computing chip, characterized in that: The method comprises: Determining a design optimization function of the optical computing chip according to an input light field and an output light field of the optical computing chip; Optimizing a plurality of optimizable parameters of the optical computing chip, and calculating a current optimization function value of the optical computing chip according to a design optimization function of the optical computing chip after optimizing each optimizable parameter; Determining whether the current optimization function value of the optical computing chip meets the preset optical computing chip optimization requirements; If the optimization function value of the optical computing chip does not meet the preset optimization requirements of the optical computing chip, re-optimize the multiple optimizable parameters of the optical computing chip and calculate the optimization function value of the optical computing chip until the optimization function value of the optical computing chip meets the preset optimization requirements of the optical computing chip; If the optimization function value of the optical computing chip meets the preset optical computing chip optimization requirement, the optical computing chip structure corresponding to the optimization function value of the optical computing chip is determined as the target optical computing chip structure.

2. The method according to claim 1, characterized in that The optical computing chip includes Optical beam splitters and Optical beam splitting devices; Among them, determine The steps of optimizing multiple function values ​​of the optical beam splitter device include: In a predetermined optimization space, determining first influence values ​​of a plurality of analysis points in the optimization space on an output light field according to an optimization influence formula; Determining second influence values ​​of different graphic areas on the output light field according to first influence values ​​of multiple analysis points in the optimization space on the output light field; According to the second influence value of different graphic areas on the output light field, the Basic structure of optical beam splitting device; The boundary of the infrastructure is divided into multiple segments, and one or more of the multiple segments are adjusted multiple times, and after each adjustment, the calculation The current optimization function value of the optical beam splitter.

3. The method according to claim 2, characterized in that The step of determining whether the current optimization function value of the optical computing chip meets the preset optical computing chip optimization requirement includes: Determine whether the current optimization function value is greater than the optimization function value after the optimization parameters were adjusted last time; If the current optimization function value is greater than the optimization function value after the optimization parameter was adjusted last time, determining the optical computing chip structure corresponding to the current optimization function value; Determine whether the current optimization function value is greater than or equal to a preset value. Target optimization value of optical beam splitter; If the current optimization function value is greater than or equal to a preset optical computing chip target optimization value, then determining that the optimization function value of the optical computing chip meets a preset optical computing chip optimization requirement; The optical computing chip structure corresponding to the optimization function value that meets the preset optical computing chip optimization requirements is determined as the target Optical beam splitting device structure.

4. The method according to claim 3, characterized in that The method further comprises: If the current optimization function value is less than or equal to the optimization function value after the last adjustment of the optimization parameters, multiple adjustments are performed on one or more segments in the basic structure until the adjusted current optimization function value is greater than the optimization function value after the last adjustment of the optimization parameters.

5. The method according to claim 3, characterized in that: The method further comprises: If the current optimization function value is less than or equal to a preset optical computing chip target optimization value, one or more sections in the basic structure are adjusted multiple times until the adjusted current optimization function value is greater than or equal to the preset optical computing chip target optimization value.

6. The method according to claim 2, characterized in that Sure The steps of optimizing the function of the optical beam splitter device include: Based on the preset coding accuracy, the area to be optimized is divided into blocks, and each block is encoded to obtain multiple coding areas; Calculating a third impact value of each encoding region on the output light field; Randomly assigning values ​​to the multiple coding regions respectively to obtain multiple assigned coding structures, wherein the assignments include a first assignment and a second assignment, wherein the first assignment indicates that the coding region is material, and the second assignment indicates that the coding region is empty; Calculate the structure optimization function value of each assignment coding structure; Determine a weight value of each assignment coding structure according to a structural optimization function value of each assignment coding structure; Determine whether there is a pre-set difference in the structure optimization function value of multiple assignment coding structures The structural optimization function value with the same target optimization value of the optical beam splitter; If there is no pre-set value in the structure optimization function value of multiple assignment coding structures The structure optimization function value of the optical beam splitter device is the same as the target optimization value, then the updated assignment coding structure is generated according to the weight value of each assignment coding structure until the structure optimization function value of the updated assignment coding structure is consistent with the preset The target optimization values ​​of the optical beam splitter are the same.

7. The method according to claim 6, characterized in that The method further comprises: Assigning values ​​to multiple target coding regions in the update assignment coding structure, wherein the multiple target coding regions are coding regions in the multiple coding regions that have a positive impact on the output light field; Calculate each assignment coding structure and update the structure optimization function value of the assignment coding structure; According to the structural optimization function value of each assignment coding structure and the updated assignment coding structure, the weight value of each assignment coding structure and the updated assignment coding structure is determined.

8. The method according to claim 7, characterized in that Determine each assigned coding structure and / or update the weight value of the assigned coding structure by the following steps: Determine the optimization function ratio value of each assignment coding structure and / or the update assignment coding structure according to the optimization function value of each assignment coding structure and / or the update assignment coding structure; According to the optimization function proportion of each assignment coding structure and / or the updated assignment coding structure, the weight value of each assignment coding structure and / or the updated assignment coding structure is determined.

9. The method according to claim 6, characterized in that The method further comprises: will be with the pre-set The optical computing chip structure corresponding to the same structural optimization function value as the target optimization value of the optical beam splitter device is determined as the target Optical beam splitting device structure.

10. The method according to claim 1, characterized in that The design optimization function of the optical computing chip is: ； Among them, FOM represents the design optimization function of the optical computing chip. Represents the number of wavelengths of the optical computing chip, represents the transmittance of an output port of the optical computing chip, represents the transmittance of another output port of the optical computing chip, represents the i-th wavelength, represents the output at the i-th wavelength.

Images