Optical computing method based on optical computing network structure, device, system, and medium

Abstract

The present application discloses an optical computing method based on an optical computing network structure, a device, a system, and a medium. The optical computing network structure comprises a plurality of preset computing units, and each preset computing unit comprises a Mach-Zehnder interferometer, a broadband phase shifter, and a plurality of micro-ring resonators. The method comprises: determining first modulation parameters for micro-ring resonators from a preset micro-ring modulation parameter range, wherein during matrix-matrix multiplication, the micro-ring resonators on an upper arm and a lower arm of an interferometer have the same first modulation parameter, and during different matrix-vector multiplications, any two of the micro-ring resonators on the upper arm and the lower arm of the interferometer have different first modulation parameters; according to a second modulation parameter, each broadband phase shifter performing phase delay modulation on light passing through the Mach-Zehnder interferometer, and according to a first modulation parameter, each micro-ring resonator performing phase modulation on the light passing through the Mach-Zehnder interferometer; and inputting an optical signal to be computed into an optical computing network structure to obtain an optical computing result output by the optical computing network structure.

Description

Optical computing method, device, system and medium based on optical computing network structure TECHNICAL FIELD

[0001] The present application relates to the technical field of optical computing, in particular to an optical computing method, device, system and medium based on an optical computing network structure. BACKGROUND

[0002] In recent years, artificial intelligence (AI) technology has developed rapidly and is widely used in many fields such as language translation, image classification, audio recognition, autonomous driving and cancer diagnosis. With the continuous expansion of AI model size, the demand for high computing power, fast processing and energy efficiency is also increasing.

[0003] However, due to the basic limitations of joule heat effect, electron tunneling effect, parasitic capacitance and radio frequency crosstalk, it becomes more and more difficult to improve the performance and energy efficiency of traditional electronic integrated circuits (EIC) through semiconductor technology.

[0004] As a promising solution, photonic computing has emerged, which makes full use of the advantages of low delay, low power consumption, no electromagnetic interference and parallel processing of light, and is particularly suitable for performing large-scale matrix operations, which are the basic operations of many AI algorithms such as feedforward neural networks, convolutional neural networks and spiking neural networks. Photonic computing uses interference, diffraction or dispersion of light to realize large-scale matrix operations, with extremely low power consumption and delay.

[0005] At present, the implementation methods of photonic computing can be mainly divided into the following three categories: spatial light diffraction method, Mach-Zehnder interferometer method and wavelength division multiplexing method based on resonators (such as micro rings). Among them, the Mach-Zehnder interferometer method is an ideal method for photonic matrix operation because of its on-chip integration, efficient processing of real and complex matrix operations, and easy execution of singular value decomposition (SVD). However, the Mach-Zehnder interferometer method still has the disadvantages of low modulation efficiency, large footprint and high power consumption. Moreover, when implementing photonic computing on an integrated platform, a single optical processor core can only implement a single matrix-vector multiplication, which greatly limits the flexibility of processing diversified computing tasks. SUMMARY

[0006] The present application aims to solve one of the technical problems in the related art to some extent. To this end, the present application provides an optical computing method, device, system and medium based on an optical computing network structure.

[0007] As a first aspect of the present application, an optical computing method based on an optical computing network structure is provided, wherein the optical computing network structure comprises a plurality of pre-designed computing units,

[0008] The pre-designed computing unit comprises a Mach-Zehnder interferometer, a broadband phase shifter and a plurality of microring resonators, the broadband phase shifter is arranged on the upper arm of the Mach-Zehnder interferometer, and at least part of the microring resonators are arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer, and the method comprises:

[0009] According to the type to be calculated, a first modulation parameter is determined for each of the microring resonators from a pre-set range of microring modulation parameters; in the case of matrix-by-matrix multiplication, each pair of microring resonators arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computing unit has the same first modulation parameter; in the case of different matrix-by-vector multiplication, any two pairs of microring resonators arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computing unit have different first modulation parameters;

[0010] The determined first modulation parameters are input into the corresponding microring resonators, and a pre-set second modulation parameter is input into each of the broadband phase shifters, so that each of the broadband phase shifters performs phase delay modulation on the light passing through the corresponding Mach-Zehnder interferometer according to the second modulation parameter, and each of the microring resonators performs phase modulation on the light passing through the corresponding Mach-Zehnder interferometer according to the first modulation parameter thereof;

[0011] An optical signal to be calculated is input into the optical computing network structure, and an optical computing result output by the optical computing network structure is obtained.

[0012] Optionally, the optical computing network structure comprises an input modulation module, a matrix multiplication module and an output detection module connected in cascade;

[0013] The input modulation module comprises a number of pre-designed computing units consistent with the number of input ports of the optical computing network structure;

[0014] The matrix multiplication module comprises a number of pre-designed computing units consistent with the square of the number of input ports of the optical computing network structure;

[0015] The output detection module comprises a number of addition-subtraction microring filters and balanced detectors, the through end and the selected end of one of the addition-subtraction microring filters are connected to one of the balanced detectors.

[0016] Optionally, a nonlinear activation module is cascaded between the matrix multiplication module and the output detection module, the nonlinear activation module comprising a plurality of pre-designed computing units corresponding to the number of input ports of the optical computing network structure;

[0017] The microring resonators above the upper arms of the pre-designed computing units of the nonlinear activation module and the microring resonators above the lower arms of the pre-designed computing units of the nonlinear activation module are respectively in one of over-coupling state and under-coupling state and the other;

[0018] The microring resonators above the pre-designed computing units of the nonlinear activation module work in different nonlinear working zones.

[0019] Optionally, the type of the optical computing network structure is any one of singular value decomposition structure for complex number computation, mapping structure for complex number computation, singular value decomposition structure for real number computation, and mapping structure for real number computation.

[0020] In the singular value decomposition structure for complex number computation, the type of the pre-designed computing units in the input modulation module comprises a two-port complex number computation unit and / or a single-port complex number computation unit, and the type of the pre-designed computing units in the matrix multiplication module comprises a two-port real number computation unit and / or a single-port real number computation unit, and a two-port complex number computation unit.

[0021] In the mapping structure for complex number computation, the type of the pre-designed computing units in the input modulation module comprises a two-port complex number computation unit and / or a single-port complex number computation unit, and the type of the pre-designed computing units in the matrix multiplication module comprises a single-port complex number computation unit and / or a two-port complex number computation unit.

[0022] In the singular value decomposition structure for real number computation, the type of the pre-designed computing units in the input modulation module comprises a two-port real number computation unit and / or a single-port real number computation unit, and the type of the pre-designed computing units in the matrix multiplication module comprises a two-port real number computation unit and / or a single-port real number computation unit, and a two-port complex number computation unit.

[0023] In the mapping structure for real number computation, the type of the pre-designed computing units in the input modulation module comprises a two-port real number computation unit and / or a single-port real number computation unit, and the type of the pre-designed computing units in the matrix multiplication module comprises a single-port real number computation unit and / or a two-port real number computation unit.

[0024] Optionally, the two-port real number computation unit has two input ports and two output ports.

[0025] The double-port complex number calculation unit has a double-input port and a double-output port, and further comprises a plurality of pairs of micro-ring resonators arranged opposite to the output waveguide of the Mach-Zehnder interferometer, the number of the plurality of pairs of micro-ring resonators is consistent compared with the number of the plurality of pairs of micro-ring resonators arranged opposite to the upper arm and the lower arm of the same Mach-Zehnder interferometer, and the plurality of pairs of micro-ring resonators have the same first modulation parameter in the case of the type to be calculated being matrix and matrix multiplication, and have different first modulation parameters between any two pairs of micro-ring resonators in the case of the type to be calculated being different matrix and vector multiplication.

[0026] The single-port real number calculation unit has a single-input port and a single-output port.

[0027] The single-port complex number calculation unit has a single-input port and a single-output port, and further comprises a plurality of micro-ring resonators arranged opposite to the output waveguide of the Mach-Zehnder interferometer, the number of the plurality of micro-ring resonators is consistent compared with the number of the plurality of micro-ring resonators arranged on the upper arm or the lower arm of the same Mach-Zehnder interferometer, and the plurality of micro-ring resonators have the same first modulation parameter in the case of the type to be calculated being matrix and matrix multiplication, and have different first modulation parameters between any two micro-ring resonators in the case of the type to be calculated being different matrix and vector multiplication.

[0028] As a second aspect of the present application, an electronic device is provided, wherein the electronic device comprises:

[0029] one or more processors;

[0030] a memory having one or more computer programs stored thereon, when the one or more computer programs are executed by the one or more processors, the one or more processors implement the optical computing method based on the optical computing network structure provided by the first aspect of the present application.

[0031] As a third aspect of the present application, an optical computing system is provided, wherein the optical computing system comprises an optical computing network structure and the electronic device provided by the second aspect of the present application,

[0032] The optical computing network structure comprises a plurality of pre-designed calculation units, the pre-designed calculation unit comprises a Mach-Zehnder interferometer, a wideband phase shifter and a plurality of micro-ring resonators, the wideband phase shifter is arranged on the upper arm of the Mach-Zehnder interferometer, and at least part of the micro-ring resonators are arranged opposite to the upper arm and the lower arm of the Mach-Zehnder interferometer;

[0033] The wideband phase shifter is configured to perform phase delay modulation on light passing through the corresponding Mach-Zehnder interferometer according to a preset second modulation parameter; each micro-ring resonator is configured to perform phase modulation on light passing through the corresponding Mach-Zehnder interferometer according to a first modulation parameter of the micro-ring resonator; and the optical computation is performed on the to-be-computed optical signal; in a case where the to-be-computed type is matrix-matrix multiplication, each pair of micro-ring resonators arranged above the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computation unit has the same first modulation parameter; in a case where the to-be-computed type is different matrix-vector multiplication, any two pairs of micro-ring resonators arranged above the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computation unit have different first modulation parameters.

[0034] Optionally, the optical computing network structure comprises a cascaded input modulation module, a matrix multiplication module, a nonlinear activation module, and an output detection module.

[0035] The input modulation module comprises a number of pre-designed computation units equal to a number of input ports of the optical computing network structure.

[0036] The matrix multiplication module comprises a number of pre-designed computation units equal to a square of the number of input ports of the optical computing network structure.

[0037] The nonlinear activation module comprises a number of pre-designed computation units equal to the number of input ports of the optical computing network structure; a micro-ring resonator above the upper arm of each pre-designed computation unit of the nonlinear activation module and a micro-ring resonator above the lower arm of each pre-designed computation unit of the nonlinear activation module are in one of an over-coupled state and an under-coupled state and the other, respectively; and each micro-ring resonator on each pre-designed computation unit of the nonlinear activation module operates in a different nonlinear operating region.

[0038] The output detection module comprises a number of addition-subtraction micro-ring filters equal to the number of input ports of the optical computing network structure and a number of balanced detectors equal to the number of addition-subtraction micro-ring filters; a pass-through end and a selected end of one addition-subtraction micro-ring filter are connected to one balanced detector at the same time.

[0039] Optionally, the optical computing network structure is of any one of the following types: a singular value decomposition structure for complex number computation, a mapping structure for complex number computation, a singular value decomposition structure for real number computation, and a mapping structure for real number computation.

[0040] In the singular value decomposition structure of the complex number calculation, the type of the pre-designed calculation unit in the input modulation module comprises a double-port complex number calculation unit and / or a single-port complex number calculation unit, and the type of the pre-designed calculation unit in the matrix multiplication module comprises a double-port real number calculation unit and / or a single-port real number calculation unit and a double-port complex number calculation unit;

[0041] In the mapping structure of the complex number calculation, the type of the pre-designed calculation unit in the input modulation module comprises a double-port complex number calculation unit and / or a single-port complex number calculation unit, and the type of the pre-designed calculation unit in the matrix multiplication module comprises a single-port complex number calculation unit and / or a double-port complex number calculation unit;

[0042] In the singular value decomposition structure of the real number calculation, the type of the pre-designed calculation unit in the input modulation module comprises a double-port real number calculation unit and / or a single-port real number calculation unit, and the type of the pre-designed calculation unit in the matrix multiplication module comprises a double-port real number calculation unit and / or a single-port real number calculation unit and a double-port complex number calculation unit;

[0043] In the mapping structure of the real number calculation, the type of the pre-designed calculation unit in the input modulation module comprises a double-port real number calculation unit and / or a single-port real number calculation unit, and the type of the pre-designed calculation unit in the matrix multiplication module comprises a single-port real number calculation unit and / or a double-port real number calculation unit;

[0044] The double-port real number calculation unit has a double-input port and a double-output port;

[0045] The double-port complex number calculation unit has a double-input port and a double-output port, and further comprises a plurality of pairs of micro-ring resonators oppositely arranged above the output waveguide of the Mach-Zehnder interferometer, the number of the plurality of pairs of micro-ring resonators is consistent compared with the number of the plurality of pairs of micro-ring resonators oppositely arranged above the upper arm and the lower arm of the same Mach-Zehnder interferometer, and in the case of the type to be calculated being a matrix and a matrix multiplication, the plurality of pairs of micro-ring resonators have the same first modulation parameter, and in the case of the type to be calculated being a different matrix and a vector multiplication, any two pairs of micro-ring resonators in the plurality of pairs of micro-ring resonators have different first modulation parameters;

[0046] The single-port real number calculation unit has a single-input port and a single-output port;

[0047] The single-port complex calculation unit has a single input port and a single output port, and further comprises a plurality of micro-ring resonators arranged oppositely above the output waveguide of the Mach-Zehnder interferometer, the plurality of micro-ring resonators are consistent in number compared with the plurality of micro-ring resonators arranged above the upper arm or the lower arm of the same Mach-Zehnder interferometer, and in the case of the type to be calculated being matrix and matrix multiplication, the plurality of micro-ring resonators have the same first modulation parameter, and in the case of the type to be calculated being different matrix and vector multiplication, any two micro-ring resonators in the plurality of micro-ring resonators have different first modulation parameters.

[0048] As a fourth aspect of the present application, a computer readable medium is provided, and a computer program is stored on the computer readable medium, wherein the computer program is executed by a processor to implement the optical computing method based on the optical computing network structure provided in the first aspect of the present application.

[0049] In the optical computing method based on the optical computing network structure provided in the embodiments of the present application, a micro-ring resonator is arranged as a resonant phase shifter on the Mach-Zehnder interferometer, the narrow-band and high-efficiency modulation characteristics of the micro-ring resonator are used to realize low-power phase adjustment of the Mach-Zehnder interferometer on different wavelength optical signals, and a wide-band phase shifter is arranged to make the wide-band phase shifter provide common-mode phase shift / phase bias and work with the micro-ring resonator to overcome the loss caused by resonance. In this way, the Mach-Zehnder interferometer, the micro-ring resonator and the wide-band phase shifter are constructed as a basic computing unit, which can effectively improve the modulation efficiency of the Mach-Zehnder interferometer, thereby reducing the footprint and power consumption. Further, the optical computing network structure is constructed by using the basic computing unit, and according to the type to be calculated, a first modulation parameter is determined for each micro-ring resonator from a preset micro-ring modulation parameter range; in the case that the type to be calculated is matrix and matrix multiplication, each pair of micro-ring resonators arranged above the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computing unit has the same first modulation parameter; in the case that the type to be calculated is different matrix and vector multiplication, any two pairs of micro-ring resonators arranged above the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computing unit have different first modulation parameters; the determined first modulation parameters are respectively input to the corresponding micro-ring resonators, and a preset second modulation parameter is input to each wide-band phase shifter, so that each wide-band phase shifter respectively performs phase delay modulation on the light passing through the corresponding Mach-Zehnder interferometer according to the second modulation parameter, each micro-ring resonator performs phase modulation on the light passing through the corresponding Mach-Zehnder interferometer according to the first modulation parameter thereof, the optical signal to be calculated is input to the optical computing network structure, and the optical computing result output by the optical computing network structure is obtained. The programmable free switching of matrix and matrix multiplication and matrix and vector multiplication can also be realized, thereby improving the computing scale, the flexibility of processing diversified computing tasks and the applicable scenarios of optical computing. BRIEF DESCRIPTION OF DRAWINGS

[0050] The present application will be further described below in conjunction with the drawings:

[0051] FIG. 1 is a flowchart of an embodiment of the optical computing method based on the optical computing network structure provided in the embodiments of the present application;

[0052] FIG. 2 is a schematic diagram of a pre-designed computing unit provided in the embodiments of the present application;

[0053] FIG. 3 is a schematic diagram of an embodiment of a pre-designed computing unit in a nonlinear activation module provided in the embodiments of the present application;

[0054] FIG. 4a is a schematic diagram of an embodiment of a singular value decomposition structure of complex number calculation provided in the embodiments of the present application;

[0055] FIG. 4b is a schematic diagram of an embodiment of a mapping structure for complex computation according to an embodiment of the present application;

[0056] FIG. 4c is a schematic diagram of an embodiment of a singular value decomposition structure for real computation according to an embodiment of the present application;

[0057] FIG. 4d is a schematic diagram of an embodiment of a mapping structure for real computation according to an embodiment of the present application;

[0058] FIG. 5a is a schematic diagram of an embodiment of a dual-port real computation unit according to an embodiment of the present application;

[0059] FIG. 5b is a schematic diagram of an embodiment of a dual-port complex computation unit according to an embodiment of the present application;

[0060] FIG. 5c is a schematic diagram of an embodiment of a single-port real computation unit according to an embodiment of the present application;

[0061] FIG. 5d is a schematic diagram of an embodiment of a single-port complex computation unit according to an embodiment of the present application;

[0062] FIG. 6 is a schematic diagram of an embodiment of an electronic device according to an embodiment of the present application;

[0063] FIG. 7 is a schematic diagram of an embodiment of a computer-readable medium according to an embodiment of the present application.

[0064] 101: processor 102: memory

[0065] 103: I / O interface 104: bus DETAILED DESCRIPTION

[0066] Embodiments of the present application are described in detail below with reference to the attached drawing figures, wherein the same or like reference numerals and characters throughout the figures denote the same or like components or elements having the same or similar functions. The embodiments described below are presented by way of example to explain the present application, and are not intended to limit the present application.

[0067] "one embodiment" or "an embodiment" or "example" or "exemplary" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the application. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment.

[0068] Although the Mach-Zehnder interferometer method is the mainstream method for implementing photonic computing, it has been successfully applied to optical neural networks (ONNs), optical deconvolution, modular linear optical circuits, optical convolutional neural networks, equalizers, digital-to-analog converters (DACs), Ising machines, pattern analysis, and many other fields. On the one hand, the footprint of each interferometer exceeds 10,000 μm², which is a significant obstacle to the realization of super-large-scale integration design. On the other hand, due to the low modulation efficiency, the modulation arm length of the interferometer can reach hundreds of μm. Therefore, when cascading multiple Mach-Zehnder interferometers, the footprint and power consumption will increase significantly. At the same time, when implementing photonic computing on an integrated platform, a single optical processor core can only implement a single matrix-vector multiplication, greatly limiting the flexibility of processing diversified computing tasks.

[0069] Therefore, the applicant of the present application proposes to set a microring resonator as a resonant phase shifter on the Mach-Zehnder interferometer, use the narrowband and high-efficiency modulation characteristics of the microring resonator to realize low-power phase adjustment of the Mach-Zehnder interferometer on different wavelength optical signals, and set a broadband phase shifter to make the broadband phase shifter work with the microring resonator to overcome the loss caused by resonance. In this way, the Mach-Zehnder interferometer, the microring resonator, and the broadband phase shifter are constructed as a basic computing unit, which can effectively improve the modulation efficiency of the Mach-Zehnder interferometer, thereby reducing the footprint and power consumption. Further, using such a basic computing unit to construct an optical computing network structure can also realize programmable free switching of matrix-matrix multiplication and matrix-vector multiplication, thereby improving the computing scale and the flexibility of processing diversified computing tasks.

[0070] As a first aspect of the present application, an optical computing method based on an optical computing network structure is provided, wherein the optical computing network structure includes a plurality of pre-designed computing units, the pre-designed computing units include a Mach-Zehnder interferometer, a broadband phase shifter, and a plurality of microring resonators, the broadband phase shifter is arranged on the upper arm of the Mach-Zehnder interferometer, and at least part of the microring resonators are arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer, as shown in FIG. 1, the method can include:

[0071] In step S110, according to the type to be calculated, first modulation parameters are determined for each of the micro-ring resonators from a preset range of micro-ring modulation parameters; in the case that the type to be calculated is matrix-by-matrix multiplication, each pair of micro-ring resonators arranged oppositely above the upper arm and the lower arm of the Mach-Zehnder interferometer in the preset computing unit has the same first modulation parameter; in the case that the type to be calculated is different matrix-by-vector multiplication, any two pairs of micro-ring resonators arranged oppositely above the upper arm and the lower arm of the Mach-Zehnder interferometer in the preset computing unit have different first modulation parameters;

[0072] In step S120, the determined first modulation parameters are input to the corresponding micro-ring resonators respectively, and a preset second modulation parameter is input to each of the wideband phase shifters, so that each of the wideband phase shifters performs phase delay modulation on the light passing through the corresponding Mach-Zehnder interferometer according to the second modulation parameter, and each of the micro-ring resonators performs phase modulation on the light passing through the corresponding Mach-Zehnder interferometer according to the first modulation parameter thereof;

[0073] In step S130, the optical signal to be calculated is input to the optical computing network structure, and an optical computing result output by the optical computing network structure is obtained.

[0074] As shown in FIG. 2, a schematic diagram of the preset computing unit provided by the embodiment of the present application is shown, a plurality of pairs of micro-ring resonators (micro-ring 1, micro-ring 1', …, micro-ring N, micro-ring N' shown in FIG. 2) are arranged oppositely above the upper arm and the lower arm of the Mach-Zehnder interferometer, for performing low-power phase adjustment on optical signals of different wavelengths, of course, only one pair of micro-ring resonators can be arranged for performing low-power phase adjustment on optical signals of a certain wavelength, and a wideband phase shifter is further arranged above the Mach-Zehnder interferometer, for cooperating with the micro-ring resonator to overcome the loss caused by resonance.

[0075] In the embodiment of the present application, the first modulation parameters and the preset second modulation parameters are provided to the micro-ring resonators and the wideband phase shifters respectively. When the preset computing unit is working, first, each of the wideband phase shifters performs phase delay on the light passing through the upper arm of the corresponding Mach-Zehnder interferometer according to the second modulation parameter thereof, so that the entire Mach-Zehnder interferometer can be placed at the quadrature operating point.

[0076] Then, each pair of micro-ring resonators arranged above the upper arm and the lower arm of the Mach-Zehnder interferometer performs one of the phase advance modulation and the phase delay modulation on the light passing through the corresponding Mach-Zehnder interferometer according to the first modulation parameter of the micro-ring resonator itself, for example, the micro-ring resonators arranged above the upper arm of the Mach-Zehnder interferometer gradually blue-shift (i.e. move in the short wave direction) the resonance wavelength of the micro-ring resonator itself according to the first modulation parameter, so that the light passing through the lower arm of the corresponding Mach-Zehnder interferometer is phase advanced by the first modulation parameter, and the micro-ring resonators arranged above the lower arm of the Mach-Zehnder interferometer gradually red-shift (i.e. move in the long wave direction) the resonance wavelength of the micro-ring resonator itself according to the first modulation parameter, so that the light passing through the upper arm of the corresponding Mach-Zehnder interferometer is phase delayed by the first modulation parameter. Conversely, the micro-ring resonators arranged above the upper arm of the Mach-Zehnder interferometer can perform the phase delay modulation, and the micro-ring resonators arranged above the lower arm of the Mach-Zehnder interferometer can perform the phase advance modulation.

[0077] The micro-ring resonator can cause a significant phase change even if it slightly deviates from the resonance condition when it is near the resonance point, thereby achieving high-efficiency modulation at extremely low power consumption. The blue-shift and the red-shift of each pair of micro-ring resonators arranged above the upper arm and the lower arm of the Mach-Zehnder interferometer can cause a phase difference change and thereby obtain a significant total phase difference. The execution order of the blue-shift and the red-shift, i.e. the red-shift and the blue-shift of each pair of micro-ring resonators, can also cause a phase difference change and thereby obtain a significant reverse total phase difference. In addition to the phase change caused by the broadband phase shifter, a large range of absolute phase difference changes can be finally achieved at extremely low power consumption. Moreover, when the resonance wavelength of one pair of micro-ring resonators is blue-shifted and red-shifted around a certain working center wavelength, the phase of the working center wavelength can be arbitrarily adjusted. When the resonance wavelengths of multiple pairs of micro-ring resonators are blue-shifted and red-shifted around multiple working center wavelengths, the phases of the multiple working center wavelengths can be arbitrarily adjusted. Due to the narrowband characteristics of the micro-ring resonator, even if multiple wavelengths are modulated at the same time, the micro-ring resonators will hardly affect each other. In summary, within one pre-design calculation unit, multiple wavelengths of the optical signal can be independently and arbitrarily adjusted. This method can be applied to various cascading manners such as the multi-port Mach-Zehnder interferometer and the multi-port Mach-Zehnder interferometer in cascade, the single-port Mach-Zehnder interferometer and the multi-port Mach-Zehnder interferometer in cascade, the single-port Mach-Zehnder interferometer and the single-port Mach-Zehnder interferometer in cascade, and the like.

[0078] In the embodiments of the present application, the first modulation parameter is determined from a preset micro-ring modulation parameter range. The preset micro-ring modulation parameter range is not particularly limited, for example, the preset micro-ring modulation parameter range can be [π / 4, 3π / 4]. The preset second modulation parameter is also not particularly limited, for example, the second modulation parameter can be preset as π / 2.

[0079] For example, when the pre-design calculation unit is working, first, the broadband phase shifter is used to make the phase delay of the light passing through the corresponding Mach-Zehnder interferometer reach π / 2, so that the entire Mach-Zehnder interferometer is placed at the quadrature operating point, then, the micro-ring resonators 1, 2, 3, …, N arranged on the upper arm of the Mach-Zehnder interferometer are gradually blue-shifted (i.e., moved in the short wave direction) to the resonance wavelength of the micro-ring resonators, so that the phase of the light passing through the lower arm of the corresponding Mach-Zehnder interferometer is advanced to π / 4 to 3π / 4, and the micro-ring resonators 1', 2', 3', …, N' arranged on the lower arm of the Mach-Zehnder interferometer are gradually red-shifted (i.e., moved in the long wave direction) to the resonance wavelength of the micro-ring resonators, so that the phase of the light passing through the upper arm of the corresponding Mach-Zehnder interferometer is delayed to π / 4 to 3π / 4. The phase change generated by the broadband phase shifter plus the phase change caused by the blue-shift and red-shift of each pair of micro-ring resonators 1 and 1', 2 and 2', 3 and 3', …, N and N' can obtain a total phase difference of π to 2π. Similarly, the phase change generated by the broadband phase shifter plus the phase change caused by the red-shift and blue-shift of each pair of micro-ring resonators 1 and 1', 2 and 2', 3 and 3', …, N and N' can obtain a total phase difference of -π to 0. Finally, the absolute phase difference change in the range of [0, 2π] can be obtained under extremely low power consumption.

[0080] It should be noted that in the embodiments of the present application, the first modulation parameter is determined for each pair of micro-ring resonators according to whether the matrix and matrix multiplication or the matrix and vector multiplication needs to be performed on the to-be-calculated optical signal. For example, for the first pair of micro-ring resonators 1 and 1', the second pair of micro-ring resonators 2 and 2', the third pair of micro-ring resonators 3 and 3', …, and the Nth pair of micro-ring resonators N and N' arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer, each pair of micro-ring resonators, i.e., each pair of oppositely arranged micro-ring resonators (for example, 1 and 1') has the same first modulation parameter. When the matrix and matrix multiplication is performed, the total N pairs of micro-ring resonators have the same first modulation parameter. When the matrix and vector multiplication is performed, the first modulation parameter between any two pairs of micro-ring resonators in the total N pairs of micro-ring resonators is different.

[0081] Suppose that the optical computing network structure has M input ports and M output ports, the light of each input port has a total of N wavelengths (M and N are positive integers), the Nth wavelength of the input light of the Mth input port is denoted as , the pre-design calculation unit The modulation factor of each micro-ring resonator is denoted as When each micro-ring resonator in each pre-computation unit modulates each wavelength uniformly, the input matrix of the input port encoding can be completed and the multiplication of the weight matrix encoded by all the pre-computation units, which means that a single task can be efficiently calculated, and the calculation scale is improved by using the two dimensions of spatial ports and wavelengths. When N pairs of micro-ring resonators in each pre-computation unit modulate each wavelength differently, the multiplication of M input vectors and N different weight matrices can be completed, which means that multiple tasks can be simultaneously calculated in parallel. If the optical computing network architecture is integrated on a chip, it is equivalent to integrating multiple cores that can simultaneously execute different tasks in the optical chip.

[0082] In the optical computing method based on the optical computing network structure provided in the embodiments of the present application, a micro-ring resonator is arranged as a resonant phase shifter on the Mach-Zehnder interferometer, the narrow-band and high-efficiency modulation characteristics of the micro-ring resonator are used to realize low-power phase adjustment of the Mach-Zehnder interferometer on different wavelength optical signals, and a wideband phase shifter is arranged to make the wideband phase shifter provide common-mode phase shift / phase bias and work with the micro-ring resonator to overcome the loss caused by resonance. In this way, the Mach-Zehnder interferometer, the micro-ring resonator and the wideband phase shifter are constructed as a basic computing unit, which can effectively improve the modulation efficiency of the Mach-Zehnder interferometer, thereby reducing the footprint and power consumption. Further, the optical computing network structure is constructed by using the basic computing unit, and according to the type to be calculated, a first modulation parameter is determined for each micro-ring resonator from a preset micro-ring modulation parameter range; in the case where the type to be calculated is matrix and matrix multiplication, each pair of micro-ring resonators arranged above the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computing unit has the same first modulation parameter; in the case where the type to be calculated is different matrix and vector multiplication, any two pairs of micro-ring resonators arranged above the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computing unit have different first modulation parameters; the determined first modulation parameters are respectively input to the corresponding micro-ring resonators, and a preset second modulation parameter is input to each wideband phase shifter, so that each wideband phase shifter respectively performs phase delay modulation on the light passing through the corresponding Mach-Zehnder interferometer according to the second modulation parameter, each micro-ring resonator performs phase modulation on the light passing through the corresponding Mach-Zehnder interferometer according to the first modulation parameter thereof, the optical signal to be calculated is input to the optical computing network structure, and the optical computing result output by the optical computing network structure is obtained. The programmable free switching of matrix and matrix multiplication and matrix and vector multiplication can also be realized, thereby improving the computing scale, the flexibility of processing diversified computing tasks and the applicable scenarios of optical computing.

[0083] In the embodiments of the present application, the optical computing network structure used can include cascaded input modulation modules, matrix multiplication modules and output detection modules. The number of pre-designed computing units of the input modulation modules and the matrix multiplication modules and the number of add-drop ring filters and balanced detectors of the output modules are related to the number of input ports of the optical computing network structure.

[0084] Correspondingly, in some embodiments, the optical computing network structure includes cascaded input modulation modules, matrix multiplication modules and output detection modules.

[0085] The input modulation module includes the pre-designed calculation units in a number consistent with the number of input ports of the optical computing network structure; and the matrix multiplication module includes the pre-designed calculation units in a number consistent with the square of the number of input ports of the optical computing network structure.

[0086] The output detection module includes the add-subtract microring filters and the balanced detectors in a number consistent with the number of input ports of the optical computing network structure, and the through end and the selected end of one add-subtract microring filter are connected to one balanced detector.

[0087] Supposing that the optical computing network structure is designed to have M input ports and M output ports (M is a positive integer), the input modulation module includes M pre-designed calculation units, the matrix multiplication module includes MxM pre-designed calculation units, and the output module includes M add-subtract microring filters and M balanced detectors. Based on this setting mode, different matrices and matrix multiplications can be programmed and freely switched, or different vectors and matrix multiplications can be programmed and freely switched.

[0088] As described above, when each pre-designed calculation unit modulates each wavelength differently, the multiplication of M input vectors and N different weight matrices can be completed, which means that multi-task computing can be simultaneously performed. If the optical computing network architecture is integrated on a chip, and compared with the multi-core architecture of an electronic chip processor, it is equivalent to integrating multiple cores that can simultaneously perform different tasks in the optical chip. Through the output detection module, the interaction between different cores can be further realized, that is, different output vectors represented by different wavelengths input from the same port to the output detection module are selected by a series of add-subtract microring filters or pass through the balanced detectors to complete the addition and subtraction operations of different output vectors, greatly improving the computing efficiency and flexibility.

[0089] Further, when coherent light needs to be detected, a pre-designed calculation unit can be cascaded before each add-subtract microring filter in the output module, and the type of the pre-designed calculation unit is a two-port real number calculation unit. It can be understood that when non-coherent light needs to be detected, a pre-designed calculation unit can not be cascaded before each add-subtract microring filter in the output module. Correspondingly, in some embodiments, a pre-designed calculation unit of a two-port real number calculation unit is cascaded before each add-subtract microring filter in the output module.

[0090] In addition to the defects mentioned above, the traditional optical computing method also has the defect that it is difficult to realize the on-chip integration of the all-optical nonlinear activation function, which also limits the application range. In this regard, the applicant proposes to cascade the pre-designed computing unit with the same structure between the matrix multiplication module and the output detection module. Based on the pre-designed computing unit with the same structure, the programmability of the all-optical nonlinear activation function can be further realized.

[0091] Correspondingly, in some embodiments, a nonlinear activation module is also cascaded between the matrix multiplication module and the output detection module, and the nonlinear activation module includes a plurality of pre-designed computing units with a number consistent with the number of input ports of the optical computing network structure.

[0092] The micro-ring resonator above the upper arm of each pre-designed computing unit of the nonlinear activation module and the micro-ring resonator above the lower arm of each pre-designed computing unit of the nonlinear activation module are respectively in one of the over-coupling state and the under-coupling state.

[0093] Each micro-ring resonator on each pre-designed computing unit of the nonlinear activation module works in different nonlinear working zones.

[0094] When the resonator is adjusted to be in a resonant state, a very strong light field intensity is generated on the resonator, thereby a nonlinear effect can be induced in the silicon or silicon nitride waveguide. In order to highlight the nonlinear effect and use it for all-optical nonlinear activation function, for a nonlinear activation unit of N input wavelengths, N pairs of microring resonators are needed. In the embodiments of the present application, as shown in FIG. 3, the pre-computation unit (the two sides of the Mach-Zehnder interferometer also include a directional coupler or a multimode interferometer, which is not operated in the embodiments of the present application and is only briefly mentioned) including a Mach-Zehnder interferometer, a broadband phase shifter arranged on the upper arm of the Mach-Zehnder interferometer, and multiple pairs of microring resonators arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer is used again. The multiple microring resonators 1, 2, 3, …, N arranged on the upper arm of the Mach-Zehnder interferometer are placed in an over-coupled state, and the multiple microring resonators 1', 2', 3', …, N' arranged on the lower arm of the Mach-Zehnder interferometer are placed in an under-coupled state (or vice versa), so that the light field phases on the two arms induced by the nonlinear effect change in opposite directions, thereby making the nonlinear effect more significant. At the same time, optical pumping is used to make each microring resonator work in a different nonlinear working area (this can be achieved by adjusting the resonant wavelength of the microring resonator, or by arranging optical bias ports 1 and 1', 2 and 2', 3 and 3', …, N and N' on each microring resonator to inject laser light at a resonant point of the microring resonator that is not at the working wavelength). Finally, the programmability of the all-optical nonlinear activation function can be achieved, and the nonlinear effect can be flexibly controlled, thereby achieving the best performance in different application scenarios.

[0095] It should be noted that when the nonlinear activation module is cascaded between the matrix multiplication module and the output detection module, for the pre-computation unit included in the nonlinear activation module, the first modulation parameter does not need to be determined and input to the microring resonator thereof, and the preset second modulation parameter does not need to be input to the broadband phase shifter thereof, that is, the microring resonator and the broadband phase shifter thereof do not need to modulate the light passing through the corresponding Mach-Zehnder interferometer according to the first modulation parameter and the preset second modulation parameter, respectively.

[0096] As described above, the Mach-Zehnder interferometer can be divided into a two-port Mach-Zehnder interferometer and a single-port Mach-Zehnder interferometer, and accordingly, different types of pre-computation units can be constructed based on the Mach-Zehnder interferometer, and different types of optical computing network structures can be constructed based on different types of pre-computation units.

[0097] Correspondingly, in some embodiments, the type of the optical computing network structure is any one of the following: singular value decomposition structure of complex number calculation, mapping structure of complex number calculation, singular value decomposition structure of real number calculation, mapping structure of real number calculation.

[0098] In the singular value decomposition structure of the complex number calculation, the type of the pre-designed calculation unit in the input modulation module comprises a double-port complex number calculation unit and / or a single-port complex number calculation unit, and the type of the pre-designed calculation unit in the matrix multiplication module comprises a double-port real number calculation unit and / or a single-port real number calculation unit and a double-port complex number calculation unit;

[0099] In the mapping structure of the complex number calculation, the type of the pre-designed calculation unit in the input modulation module comprises a double-port complex number calculation unit and / or a single-port complex number calculation unit, and the type of the pre-designed calculation unit in the matrix multiplication module comprises a single-port complex number calculation unit and / or a double-port complex number calculation unit;

[0100] In the singular value decomposition structure of the real number calculation, the type of the pre-designed calculation unit in the input modulation module comprises a double-port real number calculation unit and / or a single-port real number calculation unit, and the type of the pre-designed calculation unit in the matrix multiplication module comprises a double-port real number calculation unit and / or a single-port real number calculation unit and a double-port complex number calculation unit;

[0101] In the mapping structure of the real number calculation, the type of the pre-designed calculation unit in the input modulation module comprises a double-port real number calculation unit and / or a single-port real number calculation unit, and the type of the pre-designed calculation unit in the matrix multiplication module comprises a single-port real number calculation unit and / or a double-port real number calculation unit.

[0102] For the above four different types of optical computing network structures, the following refers to FIG. 4a, FIG. 4b, FIG. 4c and FIG. 4d, taking an example of an optical computing network structure including 4 input ports and 4 output ports: as shown in FIG. 4a, it is a schematic diagram of an embodiment of the singular value decomposition structure of complex number calculation provided by the present application; as shown in FIG. 4b, it is a schematic diagram of an embodiment of the mapping structure of complex number calculation provided by the present application; as shown in FIG. 4c, it is a schematic diagram of an embodiment of the singular value decomposition structure of real number calculation provided by the present application; as shown in FIG. 4d, it is a schematic diagram of an embodiment of the mapping structure of real number calculation provided by the present application. In FIG. 4a, FIG. 4b, FIG. 4c and FIG. 4d, each block marked with an Arabic numeral represents a pre-designed calculation unit, and different Arabic numerals represent different types of pre-designed calculation units, “1” represents a double-port real number calculation unit, “2” represents a double-port complex number calculation unit, “3” represents a single-port real number calculation unit, and “4” represents a single-port complex number calculation unit. It should be noted that, since the micro-ring resonator and the wideband phase shifter in the pre-designed calculation unit included in the nonlinear activation module do not need to modulate the light passing through the corresponding Mach-Zehnder interferometer according to the first modulation parameter and the preset second modulation parameter, in order to distinguish from the pre-designed calculation units in the input modulation module, the matrix multiplication module and the output detection module, the type of the pre-designed calculation unit included in the nonlinear activation module is represented by the Arabic numeral “5”, but in fact, the structure of the pre-designed calculation unit “5” included in the nonlinear activation module has the possibility of being consistent with the structure of the pre-designed calculation unit “3”.

[0103] As can be seen in FIG. 4a, FIG. 4b, FIG. 4c and FIG. 4d, after the optical signal to be calculated is input into the optical computing network structure, the input signal is first modulated by the input modulation module composed of 4 pre-designed calculation units, then the matrix multiplication is completed by the matrix multiplication module composed of 16 pre-designed calculation units, then the nonlinear activation is performed by the nonlinear activation module composed of 4 pre-designed calculation units, and finally the coherent light is detected by the output detection module composed of 4 pre-designed calculation units, 4 addition-subtraction micro-ring filters and 4 balanced detectors, and the optical computing result is finally obtained.

[0104] Since the optical computing network structure can independently modulate the light of different wavelengths from different input ports, this programmable modulation greatly increases the flexibility and computing scale of optical computing, and can arbitrarily switch between different matrix and matrix multiplication, different matrix and vector multiplication, and in addition, the nonlinear activation module can directly perform on-chip integrated all-optical nonlinear activation operation after linear operation, and can also realize all-optical deep learning inference function.

[0105] It should be noted that the four types of optical computing network structures shown in FIG. 4a, FIG. 4b, FIG. 4c and FIG. 4d are all exemplary illustrated by taking an example that the type of pre-designed computing unit in the input modulation module only includes one type, and the type of real number computing unit or complex number computing unit in the matrix multiplication module only includes one type. In fact, for the singular value decomposition structure of complex number calculation as shown in FIG. 4a, part or all of the “2” (i.e. double-port complex number computing unit) in the input modulation module can be replaced by “4” (i.e. single-port complex number computing unit), and part or all of the “1” (i.e. double-port real number computing unit) in the matrix multiplication module can be replaced by “3” (i.e. single-port real number computing unit), and the singular value decomposition structure of complex number calculation can still be obtained. For the mapping structure of complex number calculation as shown in FIG. 4b, part or all of the “4” (i.e. single-port complex number computing unit) in the input modulation module can be replaced by “2” (i.e. double-port complex number computing unit), and part or all of the “4” (i.e. single-port complex number computing unit) in the matrix multiplication module can be replaced by “2” (i.e. double-port complex number computing unit), and the mapping structure of complex number calculation can still be obtained. For the singular value decomposition structure of real number calculation as shown in FIG. 4c, part or all of the “1” (i.e. double-port real number computing unit) in the input modulation module can be replaced by “3” (i.e. single-port real number computing unit), and part or all of the “1” (i.e. double-port real number computing unit) in the matrix multiplication module can be replaced by “3” (i.e. single-port real number computing unit), and the singular value decomposition structure of real number calculation can still be obtained. For the mapping structure of real number calculation as shown in FIG. 4d, part or all of the “3” (i.e. single-port real number computing unit) in the input modulation module can be replaced by “1” (i.e. double-port real number computing unit), and part or all of the “3” (i.e. single-port real number computing unit) in the matrix multiplication module can be replaced by “1” (i.e. double-port real number computing unit), and the mapping structure of real number calculation can still be obtained.

[0106] Correspondingly, in some embodiments, the double-port real number computing unit has double input ports and double output ports;

[0107] The double-port complex number computing unit has double input ports and double output ports, and further comprises a plurality of pairs of micro-ring resonators oppositely arranged above the output waveguide of the Mach-Zehnder interferometer, the number of the plurality of pairs of micro-ring resonators is consistent compared with the number of the plurality of pairs of micro-ring resonators oppositely arranged above the upper arm and the lower arm of the same Mach-Zehnder interferometer, and in the case that the type of computation to be calculated is matrix-by-matrix multiplication, the plurality of pairs of micro-ring resonators have the same first modulation parameter, and in the case that the type of computation to be calculated is different matrix-by-vector multiplication, the first modulation parameter between any two pairs of micro-ring resonators in the plurality of pairs of micro-ring resonators is different;

[0108] The single-port real number calculation unit has a single input port and a single output port;

[0109] The single-port complex number calculation unit has a single input port and a single output port, and further comprises a plurality of micro-ring resonators arranged opposite to the output waveguide of the Mach-Zehnder interferometer, the plurality of micro-ring resonators are consistent in number compared with the plurality of micro-ring resonators arranged on the upper arm or the lower arm of the same Mach-Zehnder interferometer, and in the case of the type to be calculated being matrix and matrix multiplication, the plurality of micro-ring resonators have the same first modulation parameter, and in the case of the type to be calculated being different matrix and vector multiplication, any two micro-ring resonators of the plurality of micro-ring resonators have different first modulation parameters.

[0110] The following will be described in detail with reference to FIGS. 5a, 5b, 5c and 5d.

[0111] As shown in FIG. 5a, it is a schematic diagram of one embodiment of the dual-port real number calculation unit provided by the embodiments of the present application. In the dual-port real number calculation unit, for N different wavelengths, N pairs of micro-ring resonators and 1 wideband phase shifter are used to adjust the phase of light passing through the Mach-Zehnder interferometer between the directional couplers or the multimode interferometers on both sides of the Mach-Zehnder interferometer. Specifically, the N pairs of micro-ring resonators can be arranged opposite to the upper arm and the lower arm of the Mach-Zehnder interferometer, and the wideband phase shifter can be arranged on the upper arm of the Mach-Zehnder interferometer. Each pair of micro-ring resonators performs one of phase advance modulation and phase delay modulation on the light passing through the upper arm and the lower arm of the Mach-Zehnder interferometer, and the wideband phase shifter performs phase delay modulation on the light passing through the upper arm of the Mach-Zehnder interferometer.

[0112] As shown in FIG. 5b, it is a schematic diagram of one embodiment of the dual-port complex number calculation unit provided by the embodiments of the present application. In the dual-port complex number calculation unit, for N different wavelengths, N pairs of micro-ring resonators and 1 wideband phase shifter are used to adjust the phase of light passing through the Mach-Zehnder interferometer between the directional couplers or the multimode interferometers on both sides of the Mach-Zehnder interferometer. Specifically, the N pairs of micro-ring resonators can be arranged opposite to the upper arm and the lower arm of the Mach-Zehnder interferometer, and the wideband phase shifter can be arranged on the upper arm of the Mach-Zehnder interferometer. Each pair of micro-ring resonators performs one of phase advance modulation and phase delay modulation on the light passing through the upper arm and the lower arm of the Mach-Zehnder interferometer, and the wideband phase shifter performs phase delay modulation on the light passing through the upper arm of the Mach-Zehnder interferometer.

[0113] In addition, for the dual-port complex number calculation unit, N pairs of micro-ring resonators are also arranged on the output waveguide of the Mach-Zehnder interferometer, and each pair of micro-ring resonators performs one of the phase advance modulation and the phase delay modulation on the light passing through the Mach-Zehnder interferometer. It should be noted that the N pairs of micro-ring resonators arranged on the output waveguide of the Mach-Zehnder interferometer are consistent with the N pairs of micro-ring resonators arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer in terms of the number (both N pairs, that is, 2N micro-ring resonators) and the same first modulation parameters of the N pairs of micro-ring resonators in the case of the matrix-matrix multiplication type to be calculated, and different first modulation parameters between any two pairs of micro-ring resonators in the case of the different matrix-vector multiplication type to be calculated. The difference is that the first modulation parameters of the N pairs of micro-ring resonators arranged on the output waveguide of the Mach-Zehnder interferometer can be the same or different, for example, the micro-rings M and M'...P and P' shown in FIG. 5b, and the micro-rings 1 and 1'...N and N'. After the N pairs of micro-ring resonators arranged on the output waveguide of the Mach-Zehnder interferometer modulate the light signal, the absolute phase difference change in the range of [0, 2π] can also be achieved.

[0114] As shown in FIG. 5c, it is a schematic diagram of one embodiment of the single-port real number calculation unit provided by the embodiment of the application. In the single-port real number calculation unit, for N different wavelengths, N pairs of micro-ring resonators and one broadband phase shifter are used to perform phase adjustment on the light passing through the Mach-Zehnder interferometer between the directional couplers or the multi-mode interferometers on both sides of the Mach-Zehnder interferometer. Specifically, the N pairs of micro-ring resonators can be arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer, and the broadband phase shifter can be arranged on the upper arm of the Mach-Zehnder interferometer. Each pair of micro-ring resonators performs one of the phase advance modulation and the phase delay modulation on the light passing through the upper arm and the lower arm of the Mach-Zehnder interferometer, and the broadband phase shifter performs the phase delay modulation on the light passing through the upper arm of the Mach-Zehnder interferometer. The difference between the single-port real number calculation unit and the dual-port real number calculation unit is that the dual-port real number calculation unit has a dual-input port and a dual-output port, while the single-port real number calculation unit has a single-input port and a single-output port.

[0115] As shown in FIG. 5d, it is a schematic diagram of an embodiment of the single-port complex computing unit provided by the present application. In the single-port complex computing unit, for N different wavelengths, N pairs of micro-ring resonators and one wideband phase shifter are used between the directional couplers or multi-mode interferometers on both sides of the Mach-Zehnder interferometer to adjust the phase of the light passing through the Mach-Zehnder interferometer. Specifically, the N pairs of micro-ring resonators can be arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer, and the wideband phase shifter can be arranged on the upper arm of the Mach-Zehnder interferometer. Each pair of micro-ring resonators performs one of the phase advance modulation and the phase delay modulation on the light passing through the upper arm and the lower arm of the Mach-Zehnder interferometer, and the wideband phase shifter performs the phase delay modulation on the light passing through the upper arm of the Mach-Zehnder interferometer.

[0116] In addition, for the single-port complex computing unit, N micro-ring resonators are also arranged on the output waveguide of the Mach-Zehnder interferometer, each of which performs the phase advance modulation or the phase delay modulation on the light passing through the Mach-Zehnder interferometer. It should be noted that the N micro-ring resonators arranged on the output waveguide of the Mach-Zehnder interferometer are consistent with the N micro-ring resonators arranged on the upper arm or the lower arm of the same Mach-Zehnder interferometer in terms of the same number (both N) and the same first modulation parameter when the type of the calculation to be performed is the matrix-matrix multiplication, and different in that the first modulation parameter can be the same or different, for example, the micro-rings M…M' and P' shown in FIG. 5d, compared with the micro-rings 1…N, have the same or different first modulation parameter. After the N micro-ring resonators arranged on the output waveguide of the Mach-Zehnder interferometer modulate the light signal, the absolute phase difference change in the range of [0, π] can also be achieved.

[0117] It can be seen that for N different wavelengths, the double-port real computing unit includes 2N micro-ring resonators and one wideband phase shifter, the double-port complex computing unit includes 4N micro-ring resonators and one wideband phase shifter, the single-port real computing unit includes 2N micro-ring resonators and one wideband phase shifter, and the single-port complex computing unit includes 3N micro-ring resonators and one wideband phase shifter. In the pre-design computing unit, the total number of micro-ring resonators depends on the type of the pre-design computing unit and the number of wavelengths of the input light. In the optical computing network structure, the total number of micro-ring resonators depends on the type of the pre-design computing unit, the type of the optical computing network structure, and the number of wavelengths of the input light.

[0118] The dual-port real number computing unit and the single-port real number computing unit can modulate the amplitude of the input optical signal as a whole through the N pairs of micro-ring resonators arranged above the Mach-Zehnder interferometer. The dual-port complex number computing unit and the single-port complex number computing unit can modulate the amplitude of the input optical signal as a whole through the N pairs of micro-ring resonators arranged above the Mach-Zehnder interferometer; can modulate the phase of the input optical signal as a whole through the N pairs of micro-ring resonators or N micro-ring resonators arranged on the output waveguide; and thus can modulate the amplitude and the phase of the input optical signal as a whole. Therefore, the singular value decomposition structure constructed based on the dual-port complex number computing unit and / or the single-port complex number computing unit and the dual-port real number computing unit, and the mapping structure constructed based on the dual-port complex number computing unit and / or the single-port complex number computing unit, can modulate the amplitude and the phase of the optical signal at the same time, and the amplitude and the phase of the optical signal correspond to the amplitude and the phase angle of the complex number respectively, so that the complex number calculation can be performed. If only real number calculation is needed, the dual-port complex number computing unit and the single-port complex number computing unit in the input modulation module of the singular value decomposition structure for complex number calculation can be replaced by the dual-port real number computing unit and the single-port real number computing unit respectively, so that the singular value decomposition structure for real number calculation is obtained, or the dual-port complex number computing unit and the single-port complex number computing unit in the input modulation module and the matrix multiplication module of the mapping structure for complex number calculation can be replaced by the dual-port real number computing unit and the single-port real number computing unit respectively, so that the mapping structure for real number calculation is obtained.

[0119] As a second aspect of the present application, an electronic device is provided, as shown in FIG. 6, the electronic device comprises:

[0120] one or more processors 101;

[0121] a memory 102, on which one or more computer programs are stored, when the one or more computer programs are executed by the one or more processors 101, the one or more processors 101 implement the optical computing method based on the optical computing network structure provided in the first aspect of the present application.

[0122] The electronic device can further comprise one or more I / O interfaces 103 connected between the processor 101 and the memory 102, configured to realize the information interaction between the processor 101 and the memory 102.

[0123] The processor 101 is a device with data processing capability, including but not limited to a central processing unit (CPU) and the like; the memory 102 is a device with data storage capability, including but not limited to a random access memory (RAM, more specifically SDRAM, DDR, etc.), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), and a flash memory (FLASH); the I / O interface (read-write interface) is connected between the processor and the memory, and can realize information interaction between the processor and the memory, including but not limited to a data bus (Bus) and the like.

[0124] In some embodiments, the processor 101, the memory 102, and the I / O interface 103 are connected to each other through the bus 104, and further connected to other components of the computing device.

[0125] As a third aspect of the present application, an optical computing system is provided, wherein the optical computing system comprises an optical computing network structure and the electronic device provided in the second aspect of the present application,

[0126] The optical computing network structure comprises a plurality of pre-designed computing units, each of the pre-designed computing units comprising a Mach-Zehnder interferometer, a wideband phase shifter, and a plurality of microring resonators, the wideband phase shifter being arranged on an upper arm of the Mach-Zehnder interferometer, and at least part of the microring resonators being arranged on the upper arm and a lower arm of the Mach-Zehnder interferometer, respectively;

[0127] The wideband phase shifter is configured to perform phase delay modulation on light passing through the corresponding Mach-Zehnder interferometer according to a preset second modulation parameter; each of the microring resonators is configured to perform phase modulation on the light passing through the corresponding Mach-Zehnder interferometer according to a first modulation parameter of the microring resonator; and the optical computing is performed on the to-be-computed optical signal; in a case where the to-be-computed type is a matrix-matrix multiplication, each pair of microring resonators arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computing unit has the same first modulation parameter; and in a case where the to-be-computed type is a different matrix-vector multiplication, any two pairs of microring resonators arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computing unit have different first modulation parameters.

[0128] The structure and working principle of the optical computing network structure have been described in detail above, and will not be repeated here.

[0129] In the optical computing system provided in the embodiments of the present application, a micro-ring resonator is arranged as a resonant phase shifter on the Mach-Zehnder interferometer, the narrow-band and high-efficiency modulation characteristics of the micro-ring resonator are used to realize low-power phase adjustment of the Mach-Zehnder interferometer on different wavelength optical signals, and a wide-band phase shifter is arranged to make the wide-band phase shifter provide common-mode phase shift / phase bias and work with the micro-ring resonator to overcome the loss caused by resonance. In this way, the Mach-Zehnder interferometer, the micro-ring resonator and the wide-band phase shifter are constructed as a basic computing unit, which can effectively improve the modulation efficiency of the Mach-Zehnder interferometer, thereby reducing the footprint and power consumption. Further, the optical computing network structure is constructed by using the basic computing unit, and according to the type to be calculated, a first modulation parameter is determined for each micro-ring resonator from a preset micro-ring modulation parameter range; in the case that the type to be calculated is matrix and matrix multiplication, each pair of micro-ring resonators arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computing unit has the same first modulation parameter; in the case that the type to be calculated is different matrix and vector multiplication, any two pairs of micro-ring resonators arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computing unit have different first modulation parameters; the determined first modulation parameters are respectively input to the corresponding micro-ring resonators, and a preset second modulation parameter is input to each wide-band phase shifter, so that each wide-band phase shifter respectively performs phase delay modulation on the light passing through the corresponding Mach-Zehnder interferometer according to the second modulation parameter, each micro-ring resonator performs phase modulation on the light passing through the corresponding Mach-Zehnder interferometer according to the first modulation parameter thereof, the optical signal to be calculated is input to the optical computing network structure, and the optical computing result output by the optical computing network structure is obtained. The programmable free switching of matrix and matrix multiplication and matrix and vector multiplication can also be realized, thereby improving the computing scale, the flexibility of processing diversified computing tasks and the applicable scenarios of optical computing.

[0130] In the embodiments of the present application, the phase adjustment mode of the wide-band phase shifter and the micro-ring resonator is not specially limited, for example, electrical tuning, thermal tuning, optical tuning and the like can be used.

[0131] In the embodiments of the present application, the optical integrated platform on which the optical computing network structure is implemented is not specially limited, for example, the optical computing network structure can be implemented on a silicon-based (Silicon) optical platform (pure silicon or silicon germanium platform), a silicon nitride (SiN) optical platform, a group III-V optical platform such as indium phosphide (InP) or indium arsenide (InAs), a lithium niobate (LN) optical platform and the like.

[0132] In some embodiments, the optical computing network structure comprises a cascaded input modulation module, a matrix multiplication module, a nonlinear activation module, and an output detection module;

[0133] The input modulation module comprises the pre-designed computing units in a number consistent with the number of input ports of the optical computing network structure;

[0134] The matrix multiplication module comprises the pre-designed computing units in a number consistent with the square of the number of input ports of the optical computing network structure;

[0135] The nonlinear activation module comprises the pre-designed computing units in a number consistent with the number of input ports of the optical computing network structure; the microring resonators above the upper arms of each of the pre-designed computing units of the nonlinear activation module and the microring resonators above the lower arms of each of the pre-designed computing units of the nonlinear activation module are in one of the over-coupled state and the under-coupled state and the other, respectively; each of the microring resonators on each of the pre-designed computing units of the nonlinear activation module operates in a different nonlinear operating region;

[0136] The output detection module comprises the add-subtract microring filters and the balanced detectors in a number consistent with the number of input ports of the optical computing network structure, respectively; the through end and the selected end of one of the add-subtract microring filters are connected to one of the balanced detectors simultaneously.

[0137] In some embodiments, the type of the optical computing network structure is any one of the following: a singular value decomposition structure for complex number computation, a mapping structure for complex number computation, a singular value decomposition structure for real number computation, a mapping structure for real number computation;

[0138] In the singular value decomposition structure for complex number computation, the type of the pre-designed computing units in the input modulation module comprises a two-port complex number computation unit and / or a single-port complex number computation unit, the type of the pre-designed computing units in the matrix multiplication module comprises a two-port real number computation unit and / or a single-port real number computation unit, and a two-port complex number computation unit;

[0139] In the mapping structure for complex number computation, the type of the pre-designed computing units in the input modulation module comprises a two-port complex number computation unit and / or a single-port complex number computation unit, the type of the pre-designed computing units in the matrix multiplication module comprises a single-port complex number computation unit and / or a two-port complex number computation unit;

[0140] In the singular value decomposition structure for real number computation, the type of the pre-designed computing units in the input modulation module comprises a two-port real number computation unit and / or a single-port real number computation unit, the type of the pre-designed computing units in the matrix multiplication module comprises a two-port real number computation unit and / or a single-port real number computation unit, and a two-port complex number computation unit;

[0141] In the mapping structure of the real number calculation, the type of the pre-designed calculation unit in the input modulation module comprises a two-port real number calculation unit and / or a single-port real number calculation unit, and the type of the pre-designed calculation unit in the matrix multiplication module comprises a single-port real number calculation unit and / or a two-port real number calculation unit.

[0142] The two-port real number calculation unit has a two-input port and a two-output port.

[0143] The two-port complex number calculation unit has a two-input port and a two-output port, and further comprises a plurality of pairs of micro-ring resonators arranged opposite to the output waveguide of the Mach-Zehnder interferometer, the number of the plurality of pairs of micro-ring resonators is consistent compared with the number of the plurality of pairs of micro-ring resonators arranged opposite to the upper arm and the lower arm of the same Mach-Zehnder interferometer, and in the case of the type of the calculation being matrix and matrix multiplication, the plurality of pairs of micro-ring resonators have the same first modulation parameter, and in the case of the type of the calculation being different matrix and vector multiplication, the first modulation parameter between any two pairs of micro-ring resonators in the plurality of pairs of micro-ring resonators is different.

[0144] The single-port real number calculation unit has a single-input port and a single-output port.

[0145] The single-port complex number calculation unit has a single-input port and a single-output port, and further comprises a plurality of micro-ring resonators arranged opposite to the output waveguide of the Mach-Zehnder interferometer, the number of the plurality of micro-ring resonators is consistent compared with the number of the plurality of micro-ring resonators arranged on the upper arm or the lower arm of the same Mach-Zehnder interferometer, and in the case of the type of the calculation being matrix and matrix multiplication, the plurality of micro-ring resonators have the same first modulation parameter, and in the case of the type of the calculation being different matrix and vector multiplication, the first modulation parameter between any two micro-ring resonators in the plurality of micro-ring resonators is different.

[0146] It can be understood that the connection mode between two pre-designed calculation units can be two-port interconnection or single-port connection.

[0147] As a fourth aspect of the embodiments of the present application, as shown in FIG. 7, a computer readable medium is provided, and the computer readable medium stores a computer program, wherein the computer program is executed by a processor to implement the optical calculation method based on the optical calculation network structure provided in the first aspect of the present application.

[0148] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. Accordingly, the computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the method of any one of the above embodiments can be implemented. In the embodiments provided by the present application, any reference to memory, storage, database or other medium can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM) and memory bus dynamic RAM (RDRAM), etc.

[0149] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Those skilled in the art should understand that the present application includes but is not limited to the contents described in the above specific embodiments and the accompanying drawings. Any modification that does not deviate from the functional and structural principles of the present application will be included in the scope of the claims.

Claims

1. An optical computing method based on an optical computing network structure, characterized by, The optical computing network structure comprises a plurality of pre-designed computing units, the pre-designed computing unit comprises a Mach-Zehnder interferometer, a broadband phase shifter and a plurality of microring resonators, the broadband phase shifter is arranged on the upper arm of the Mach-Zehnder interferometer, and at least part of the microring resonators are oppositely arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer, and the method comprises: According to the type to be calculated, a first modulation parameter is determined for each microring resonator from a pre-set microring modulation parameter range; wherein, in the case that the type to be calculated is matrix and matrix multiplication, each pair of microring resonators oppositely arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computing unit has the same first modulation parameter; in the case that the type to be calculated is different matrix and vector multiplication, any two pairs of microring resonators oppositely arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computing unit have different first modulation parameters; The determined first modulation parameters are respectively input into the corresponding microring resonators, and a pre-set second modulation parameter is input into each broadband phase shifter, so that each broadband phase shifter respectively performs phase delay modulation on the light passing through the corresponding Mach-Zehnder interferometer according to the second modulation parameter, and each microring resonator performs phase modulation on the light passing through the corresponding Mach-Zehnder interferometer according to the first modulation parameter of the microring resonator itself; The optical computing network structure is input into the optical computing network structure, and an optical computing result output by the optical computing network structure is obtained.

2. The method of claim 1, wherein, The optical computing network structure comprises a cascaded input modulation module, a matrix multiplication module and an output detection module; The input modulation module comprises a plurality of pre-designed computing units corresponding to the number of input ports of the optical computing network structure; The matrix multiplication module comprises a plurality of pre-designed computing units corresponding to the square of the number of input ports of the optical computing network structure; The output detection module comprises a plurality of addition and subtraction microring filters and balanced detectors corresponding to the number of input ports of the optical computing network structure, and the through end and the selected end of one addition and subtraction microring filter are connected into one balanced detector.

3. The method of claim 2, wherein, A nonlinear activation module is further cascaded between the matrix multiplication module and the output detection module, and the nonlinear activation module comprises a plurality of pre-designed computing units corresponding to the number of input ports of the optical computing network structure; The microring resonators on the upper arm of each pre-designed computing unit of the nonlinear activation module and the microring resonators on the lower arm of each pre-designed computing unit of the nonlinear activation module are respectively in one of the over-coupling state and the under-coupling state and the other; Each microring resonator on each pre-designed computing unit of the nonlinear activation module works in different nonlinear working zones.

4. The method of claim 2, wherein, The type of the optical computing network structure is any one of the following: singular value decomposition structure of complex number calculation, mapping structure of complex number calculation, singular value decomposition structure of real number calculation, mapping structure of real number calculation; In the singular value decomposition structure of the complex number calculation, the type of the pre-designed calculation unit in the input modulation module includes a double-port complex number calculation unit and / or a single-port complex number calculation unit, and the type of the pre-designed calculation unit in the matrix multiplication module includes a double-port real number calculation unit and / or a single-port real number calculation unit, and a double-port complex number calculation unit; In the mapping structure of the complex number calculation, the type of the pre-designed calculation unit in the input modulation module includes a double-port complex number calculation unit and / or a single-port complex number calculation unit, and the type of the pre-designed calculation unit in the matrix multiplication module includes a single-port complex number calculation unit and / or a double-port complex number calculation unit; In the singular value decomposition structure of the real number calculation, the type of the pre-designed calculation unit in the input modulation module includes a double-port real number calculation unit and / or a single-port real number calculation unit, and the type of the pre-designed calculation unit in the matrix multiplication module includes a double-port real number calculation unit and / or a single-port real number calculation unit, and a double-port complex number calculation unit; In the mapping structure of the real number calculation, the type of the pre-designed calculation unit in the input modulation module includes a double-port real number calculation unit and / or a single-port real number calculation unit, and the type of the pre-designed calculation unit in the matrix multiplication module includes a single-port real number calculation unit and / or a double-port real number calculation unit.

5. The method of claim 4, wherein, The double-port real number calculation unit has a double-input port and a double-output port; The double-port complex number calculation unit has a double-input port and a double-output port, and further includes a plurality of pairs of micro-ring resonators arranged opposite to the output waveguide of the Mach-Zehnder interferometer, the number of the plurality of pairs of micro-ring resonators is consistent compared with the number of the plurality of pairs of micro-ring resonators arranged opposite to the upper arm and the lower arm of the same Mach-Zehnder interferometer, and in the case of the type of the calculation to be a matrix and matrix multiplication, the plurality of pairs of micro-ring resonators have the same first modulation parameter, and in the case of the type of the calculation to be different matrix and vector multiplication, any two pairs of micro-ring resonators in the plurality of pairs of micro-ring resonators have different first modulation parameters; The single-port real number calculation unit has a single-input port and a single-output port; The single-port complex number calculation unit has a single-input port and a single-output port, and further includes a plurality of micro-ring resonators arranged opposite to the output waveguide of the Mach-Zehnder interferometer, the number of the plurality of micro-ring resonators is consistent compared with the number of the plurality of micro-ring resonators arranged on the upper arm or the lower arm of the same Mach-Zehnder interferometer, and in the case of the type of the calculation to be a matrix and matrix multiplication, the plurality of micro-ring resonators have the same first modulation parameter, and in the case of the type of the calculation to be different matrix and vector multiplication, any two micro-ring resonators in the plurality of micro-ring resonators have different first modulation parameters.

6. An electronic device, comprising: The electronic device comprises: one or more processors; a memory having one or more computer programs stored thereon, when the one or more computer programs are executed by the one or more processors, the one or more processors implement an optical computing method based on an optical computing network structure according to any one of claims 1-5.

7. An optical computing system, characterized by, The optical computing system comprises the optical computing network structure and the electronic device according to claim 6, the optical computing network structure comprises a plurality of pre-designed computing units, the pre-designed computing unit comprises a Mach-Zehnder interferometer, a broadband phase shifter and a plurality of microring resonators, the broadband phase shifter is arranged on the upper arm of the Mach-Zehnder interferometer, and at least part of the microring resonators are oppositely arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer; The broadband phase shifter is used for performing phase delay modulation on the light passing through the corresponding Mach-Zehnder interferometer according to a preset second modulation parameter; Each of the microring resonators is used for performing phase modulation on the light passing through the corresponding Mach-Zehnder interferometer according to a first modulation parameter of the microring resonator itself; The optical computing is performed on the to-be-computed optical signal; in the case that the to-be-computed type is matrix and matrix multiplication, each pair of microring resonators oppositely arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computing unit has the same first modulation parameter; In the case that the to-be-computed type is different matrix and vector multiplication, any two pairs of microring resonators oppositely arranged on the upper arm and the lower arm of the Mach-Zehnder interferometer in the pre-designed computing unit have different first modulation parameters.

8. The photonic computing system of claim 7, wherein, The optical computing network structure comprises a cascaded input modulation module, a matrix multiplication module, a nonlinear activation module and an output detection module; The input modulation module comprises a number of pre-designed computing units consistent with the number of input ports of the optical computing network structure; The matrix multiplication module comprises a number of pre-designed computing units consistent with the square of the number of input ports of the optical computing network structure; The nonlinear activation module comprises a number of pre-designed computing units consistent with the number of input ports of the optical computing network structure; The microring resonators on the upper arm of each pre-designed computing unit of the nonlinear activation module and the microring resonators on the lower arm of each pre-designed computing unit of the nonlinear activation module are respectively in one of the over-coupling state and the under-coupling state and the other; Each microring resonator on each pre-designed computing unit of the nonlinear activation module works in different nonlinear working zones; The output detection module comprises a number of addition-subtraction microring filters and balanced detectors, the through end and the selected end of one addition-subtraction microring filter are connected to one balanced detector at the same time.

9. The photonic computing system of claim 8, wherein, The type of the optical computing network structure is any one of the following: singular value decomposition structure of complex number calculation, mapping structure of complex number calculation, singular value decomposition structure of real number calculation, mapping structure of real number calculation; In the singular value decomposition structure of complex number calculation, the type of the pre-designed computing unit in the input modulation module comprises a two-port complex number calculation unit and / or a single-port complex number calculation unit, and the type of the pre-designed computing unit in the matrix multiplication module comprises a two-port real number calculation unit and / or a single-port real number calculation unit and a two-port complex number calculation unit. In the complex number calculation mapping structure, the pre-designed calculation unit in the input modulation module includes a double-port complex number calculation unit and / or a single-port complex number calculation unit, and the pre-designed calculation unit in the matrix multiplication module includes a single-port complex number calculation unit and / or a double-port complex number calculation unit. In the singular value decomposition structure of the real number calculation, the pre-designed calculation unit in the input modulation module includes a double-port real number calculation unit and / or a single-port real number calculation unit, and the pre-designed calculation unit in the matrix multiplication module includes a double-port real number calculation unit and / or a single-port real number calculation unit, and a double-port complex number calculation unit. In the real number calculation mapping structure, the pre-designed calculation unit in the input modulation module includes a double-port real number calculation unit and / or a single-port real number calculation unit, and the pre-designed calculation unit in the matrix multiplication module includes a single-port real number calculation unit and / or a double-port real number calculation unit. The double-port real number calculation unit has double input ports and double output ports. The double-port complex number calculation unit has double input ports and double output ports, and further includes a plurality of pairs of micro-ring resonators arranged opposite to the output waveguide of the Mach-Zehnder interferometer, the number of the plurality of pairs of micro-ring resonators is consistent compared with the number of the plurality of pairs of micro-ring resonators arranged opposite to the upper arm and the lower arm of the same Mach-Zehnder interferometer, and in the case of matrix and matrix multiplication, the plurality of pairs of micro-ring resonators have the same first modulation parameter, and in the case of different matrix and vector multiplication, any two pairs of micro-ring resonators in the plurality of pairs of micro-ring resonators have different first modulation parameters. The single-port real number calculation unit has a single input port and a single output port. The single-port complex number calculation unit has a single input port and a single output port, and further includes a plurality of micro-ring resonators arranged opposite to the output waveguide of the Mach-Zehnder interferometer, the number of the plurality of micro-ring resonators is consistent compared with the number of the plurality of micro-ring resonators arranged on the upper arm or the lower arm of the same Mach-Zehnder interferometer, and in the case of matrix and matrix multiplication, the plurality of micro-ring resonators have the same first modulation parameter, and in the case of different matrix and vector multiplication, any two micro-ring resonators in the plurality of micro-ring resonators have different first modulation parameters.

10. A computer readable medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the optical calculation method based on the optical calculation network structure according to any one of claims 1-5.

Images