Nonlinear processing method and apparatus based on all-optical computing architecture

Abstract

The embodiment of the application provides a nonlinear processing method and device based on a full-optical computing architecture. The method can solve the problems of high hardware resource occupancy, low calculation efficiency and high power consumption caused by complex nonlinear function calculation in the deep learning inference and training process. In addition, by replacing electronic logic units with micro-ring resonators, Mach-Zehnder interferometers and phase modulation, using wavelength division multiplexing and multi-ring resonance, supporting full-optical parallel processing of matrix blocks, reducing the number of optoelectronic conversion for nonlinear function processing in optoelectronic hybrid computing, significantly reducing the delay and improving the overall performance.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of optical computing, and in particular to a nonlinear processing method and device based on a full-optical computing architecture. BACKGROUND

[0002] With the rapid development of artificial intelligence technology, deep learning models have been widely applied in many fields. However, in the model inference and training process, a large number of complex nonlinear functions and mathematical operations are involved, such as multiplication, division, exponential operation and activation function calculation. These operations usually require a large amount of hardware resources and computing time, especially in an optoelectronic hybrid computing architecture, because optical operations cannot directly complete nonlinear function calculations, and frequent optoelectronic conversion increases the computing delay and hardware complexity. SUMMARY

[0003] Therefore, embodiments of the present application provide a nonlinear processing method and device based on a full-optical computing architecture, which can reduce the number of optoelectronic conversions for nonlinear function processing in optoelectronic hybrid computing, significantly reduce the delay, and improve the overall performance.

[0004] In a first aspect, a nonlinear processing method based on a full-optical computing architecture is provided, the full-optical computing architecture comprising a micro ring resonator MRR and a Mach-Zehnder interferometer MZI, the nonlinear processing method comprising: inputting an optical signal matrix block X into the micro ring resonator MRR, outputting a first optical signal, the first optical signal being a maximum optical signal in the optical signal matrix block X; inputting the optical signal matrix block X and the first optical signal into a cascaded Mach-Zehnder interferometer MZI and a series of micro ring resonators MRR in turn, obtaining a second optical signal corresponding to any optical signal in the optical signal matrix block X, the second optical signal being an optical signal corresponding to an exponential power t with a base of 2, wherein t = (x i -x max )·log2e, x i being a value of the any optical signal, x max being a value of the maximum optical signal; inputting the second optical signal into a low wavelength channel and a high wavelength channel in turn, outputting a third optical signal corresponding to the any optical signal, the third optical signal being a local exponential value 2 t corresponding to the any optical signal, wherein, The low wavelength channel comprises a Mach-Zehnder interferometer MZI and a micro ring resonator MRR, and is used to output a 2 r corresponding optical signal, r being a decimal part in t, the high wavelength channel comprising q cascaded Mach-Zehnder interferometers MZI, and being used to output a 2 qcorresponding to the optical signal matrix block X, q is an integer part in t; inputting the third optical signal corresponding to each optical signal in the optical signal matrix block X into a plurality of micro ring resonators MRR in series, and outputting a fourth optical signal, the fourth optical signal being a sum of the third signals corresponding to all optical signals in the optical signal matrix block X, and the fourth optical signal being an optical signal corresponding to a local index value of the optical signal matrix block X; modulating the third signal corresponding to any optical signal as an optical signal corresponding to a first wavelength, and modulating the fourth optical signal as an optical signal corresponding to a second wavelength, the optical signal corresponding to the first wavelength being corresponding to the optical signal, the optical signal corresponding to the second wavelength being corresponding to the optical signal, k1 and v1 being parameters of the first wavelength, k2 and v2 being parameters of the second wavelength, and v1 and v2 being constants less than 1; based on the sizes of v1 and v2, inputting the optical signal corresponding to the first wavelength and the optical signal corresponding to the second wavelength into a micro ring resonator MRR array and a cascaded Mach-Zehnder interferometer MZI in turn, and outputting a fifth optical signal corresponding to any optical signal, the fifth optical signal being an optical signal corresponding to a ratio of N1 and N2, wherein, if v1-v2 is greater than or equal to 0, the fifth optical signal is corresponding to the optical signal; if v1-v2 is less than 0, the fifth signal is corresponding to the optical signal.

[0005] In a possible implementation, the nonlinear processing method further includes: storing the value of the first optical signal corresponding to each optical signal matrix block X in a plurality of optical signal matrix blocks X in an optical buffer ring OBR, and outputting a sixth signal, the sixth signal being a maximum value x' in the values of the first optical signals corresponding to the plurality of optical signal matrix blocks X max corresponding to the optical signal; inputting the sixth signal and the first signal corresponding to any optical signal matrix block X into a cascaded Mach-Zehnder interferometer MZI and a plurality of micro ring resonators MRR in series in turn, to obtain a seventh signal corresponding to any optical signal matrix block X, the seventh signal being an optical signal corresponding to an exponential power t' with a base of 2, wherein t' = x max -x' max ; inputting the seventh signal into the low wavelength channel and the high wavelength channel in turn, and outputting an eighth optical signal corresponding to any optical signal matrix block X, the eighth optical signal being an optical signal corresponding to 2 t′ , wherein, the low wavelength channel is used to output an optical signal corresponding to 2 r′ , r' being a decimal part in t', the high wavelength channel including q' cascaded Mach-Zehnder interferometers MZI, and being used to output an optical signal corresponding to 2 q′corresponding to the optical signal pair, the fifth optical signal corresponding to any optical signal in the optical signal matrix block X is modulated into an optical signal corresponding to a third wavelength, and the eighth optical signal corresponding to the optical signal matrix block X is modulated into an optical signal corresponding to a fourth wavelength, the optical signal corresponding to the third wavelength is corresponding to the optical signal, the optical signal corresponding to the fourth wavelength is corresponding to the optical signal, k3 and v3 are parameters of the third wavelength, k4 and v4 are parameters of the fourth wavelength, v3 and v4 are constants less than 1; based on the size of v3 and v4, the optical signal corresponding to the third wavelength and the optical signal corresponding to the fourth wavelength are sequentially input into the micro ring resonator MRR array and the cascaded Mach-Zehnder interferometer MZI, and the ninth optical signal corresponding to any optical signal in the optical signal matrix block X is output, the ninth optical signal is an optical signal corresponding to the product of N3 and N4, and the ninth optical signal is an optical signal corresponding to the global index value of any optical signal in the optical signal matrix block X, wherein if v3+v4 is less than 1, the ninth optical signal is corresponding to the optical signal; if v3+v4 is greater than or equal to 1, the ninth optical signal is corresponding to the optical signal.

[0006] The second aspect provides a nonlinear processing device based on a full-optical computing architecture, the full-optical computing architecture comprising a micro ring resonator MRR and a Mach-Zehnder interferometer MZI, the nonlinear processing device comprising: a maximum light selection module for inputting an optical signal matrix block X into the micro ring resonator MRR, and outputting a first optical signal, the first optical signal being a maximum optical signal in the optical signal matrix block X; a light phase shift module for sequentially inputting the optical signal matrix block X and the first optical signal into a cascaded Mach-Zehnder interferometer MZI and a plurality of micro ring resonators MRR in series, to obtain a second optical signal corresponding to any optical signal in the optical signal matrix block X, the second optical signal being an optical signal corresponding to an exponential power t with a base of 2, wherein t=(x i -x max )·log2e,x i being a value of the any optical signal, x max being a value of the maximum optical signal; a light approximate index module for sequentially inputting the second optical signal into a low wavelength channel and a high wavelength channel, and outputting a third optical signal corresponding to the any optical signal, the third optical signal being an optical signal corresponding to a local index value 2 t of the any optical signal, wherein The low wavelength channel comprises a Mach-Zehnder interferometer MZI and a micro ring resonator MRR, and is used for outputting 2 rThe corresponding optical signal, where r is the fractional part of t, includes q cascaded Mach-Zehnder interferometers (MZIs) for outputting 2 q The corresponding optical signal, where q is the integer part of t; the optical wavelength division multiplexing accumulation module is used to input the third optical signal corresponding to each optical signal in the optical signal matrix block X to multiple micro-ring resonators (MRRs) connected in series, and output a fourth optical signal, which is the sum of the third signals corresponding to all optical signals in the optical signal matrix block X, and the fourth optical signal is the optical signal corresponding to the local exponent value of the optical signal matrix block X; the optical division shift module is used to modulate the third signal corresponding to any optical signal into an optical signal corresponding to a first wavelength and to modulate the fourth optical signal into an optical signal corresponding to a second wavelength, where the optical signal corresponding to the first wavelength is... The corresponding optical signal, the optical signal corresponding to the second wavelength is The corresponding optical signals, k1 and v1 are parameters of the first wavelength, k2 and v2 are parameters of the second wavelength, and v1 and v2 are constants less than 1; and based on the magnitudes of v1 and v2, the optical signals corresponding to the first wavelength and the second wavelength are sequentially input into a microring resonator (MRR) array and a cascaded Mach-Zehnder interferometer (MZI), outputting a fifth optical signal corresponding to any of the optical signals, wherein the fifth optical signal is the optical signal corresponding to the ratio of N1 to N2, wherein if v1-v2 is greater than or equal to 0, the fifth optical signal is... The corresponding optical signal; if v1-v2 is less than 0, the fifth signal is The corresponding optical signal.

[0007] In one possible implementation, the maximum optical selection module is further configured to: store the value of the first optical signal corresponding to each of the plurality of optical signal matrix blocks X in an optical buffer ring (OBR), and output a sixth signal, wherein the sixth signal is the maximum value x' among the values ​​of the first optical signals corresponding to the plurality of optical signal matrix blocks X. max The corresponding optical signal; the optical phase shift module is further configured to: sequentially input the sixth signal and the first signal corresponding to any of the optical signal matrix blocks X into a cascaded Mach-Zehnder interferometer (MZI) and a series of multiple microring resonators (MRRs) to obtain a seventh signal corresponding to any of the optical signal matrix blocks X, wherein the seventh signal is an optical signal corresponding to a power t' to the base 2, where t' = x max -x' max The optical approximation index module is further configured to: input the seventh signal sequentially to the low-wavelength channel and the high-wavelength channel, and output the eighth optical signal corresponding to any optical signal matrix block X, wherein the eighth optical signal is 2. t′ The corresponding optical signal, of which, The low-wavelength channel is used to output 2. r′ The corresponding optical signal, r' being the fractional part of t', the high-wavelength channel comprising q' cascaded Mach-Zehnder interferometers (MZIs) for outputting 2 q′ The corresponding optical signal, q' is the integer part of t'; the nonlinear processing device further includes: an optical multiplication and shift module, used to: modulate the fifth optical signal corresponding to any optical signal in the optical signal matrix block X into an optical signal corresponding to the third wavelength, and modulate the eighth optical signal corresponding to the optical signal matrix block X into an optical signal corresponding to the fourth wavelength, wherein the optical signal corresponding to the third wavelength is The corresponding optical signal, the optical signal corresponding to the fourth wavelength is The corresponding optical signals, k3 and v3 are parameters of the third wavelength, k4 and v4 are parameters of the fourth wavelength, and v3 and v4 are constants less than 1; and based on the magnitudes of v3 and v4, the optical signals corresponding to the third wavelength and the fourth wavelength are sequentially input into the microring resonator (MRR) array and the cascaded Mach-Zehnder interferometer (MZI), outputting the ninth optical signal corresponding to any optical signal in the optical signal matrix block X, wherein the ninth optical signal is the optical signal corresponding to the product of N3 and N4, and the ninth optical signal is the optical signal corresponding to the global exponent value of any optical signal in the optical signal matrix block X, wherein if v3 + v4 is less than 1, the ninth optical signal is The corresponding optical signal; if v3 + v4 is greater than or equal to 1, the ninth optical signal is The corresponding optical signal.

[0008] Thirdly, a nonlinear processing device based on an all-optical computing architecture is provided, comprising: a processor, and a memory communicatively connected to the processor, wherein the memory stores program instructions executable by the processor, and the processor can execute the nonlinear processing method based on the all-optical computing architecture according to the first aspect and any possible implementation thereof by calling the program instructions.

[0009] Fourthly, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform a nonlinear processing method based on an all-optical computing architecture, as described in the first aspect and any possible implementation thereof.

[0010] Based on the above technical solutions, the problems of high hardware resource consumption, low computational efficiency, and high power consumption caused by complex nonlinear function calculations (such as exponential operations and activation function calculations) during deep learning inference and training can be solved. Furthermore, by replacing electronic logic units with microring resonators (MRRs), Mach-Zehnder interferometers (MZIs), and phase modulation, and utilizing wavelength division multiplexing and multiring resonance, all-optical parallel processing of matrix blocks can be supported. This reduces the number of photoelectric conversions for nonlinear function processing in optoelectronic hybrid computing, significantly reducing latency and improving overall performance. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on the drawings without creative effort.

[0012] Figure 1 This is a schematic block diagram illustrating a nonlinear processing method based on an all-optical computing architecture provided in an embodiment of this application.

[0013] Figure 2 Another schematic block diagram of the nonlinear processing method based on an all-optical computing architecture provided in the embodiments of this application.

[0014] Figure 3 A schematic block diagram of a nonlinear processing device based on an all-optical computing architecture provided in an embodiment of this application. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0016] With the rapid development of artificial intelligence technology, deep learning models have been widely used in many fields. However, the model inference and training process involves a large number of complex nonlinear functions and mathematical operations, such as multiplication, division, exponentiation, and activation function calculations. These operations typically require significant hardware resources and computation time, especially in optoelectronic hybrid computing architectures, where optical operations cannot directly perform nonlinear function calculations and frequent optoelectronic conversions are necessary, increasing computational latency and hardware complexity.

[0017] In related technologies, most are based on Digital Signal Processing (DSP) hardware or Application-Specific Integrated Circuit (ASIC) design, and attempt to improve computational efficiency through linear approximation, quantization, or pipelined architecture optimization. However, these methods still suffer from high hardware overhead, high power consumption, and long computational latency in high-precision computation, making it difficult to meet the low latency and high bandwidth requirements of optical computing architectures. Furthermore, in optoelectronic hybrid systems, frequent photoelectric conversions lead to signal loss and computational latency. Especially in the field of deep learning, many common activation functions, such as Softmax and Sigmoid, can be abstractly modeled using the exponent e. How to design a hardware-friendly optimization method to further reduce power consumption and hardware complexity while ensuring computational accuracy has become a current research challenge and focus.

[0018] In view of this, embodiments of this application provide a nonlinear processing method and apparatus based on an all-optical computing architecture, which can solve the problems of high hardware resource consumption, low computational efficiency, and high power consumption caused by complex nonlinear function calculations (such as exponential operations and activation function calculations) during deep learning inference and training. Furthermore, by replacing electronic logic units with microring resonators (MRRs), Mach-Zehnder interferometers (MZIs), and phase modulation, and utilizing wavelength division multiplexing and multi-ring resonance, all-optical parallel processing of matrix blocks is supported, reducing the number of photoelectric conversions for nonlinear function processing in optoelectronic hybrid computing, significantly reducing latency, and improving overall performance.

[0019] 1. All-optical computing architecture is a computing system based on optical devices and optical signal processing. Its core idea is to utilize the characteristics of light (such as high-speed transmission, parallel processing, and low power consumption) to perform computing tasks, rather than relying on traditional electronic computing technologies. The goal of all-optical computing architecture is to improve computing efficiency, reduce power consumption, and reduce latency by reducing or eliminating photoelectric conversion processes and directly completing data transmission, storage, and processing in the optical domain.

[0020] The following section will mainly introduce several key optical devices involved in the embodiments of this application.

[0021] Microring resonators (MRRs) typically consist of one or more ring waveguides coupled to a straight waveguide. The ring waveguide is the core component, generally having a micrometer-scale or even smaller size, and can be made of optical materials such as silicon, silicon nitride, or silicon dioxide. The straight waveguide is used for input and output optical signals and is connected to the ring waveguide through a specific coupling region. The design of this coupling region is crucial to the transmission efficiency of the optical signal between the straight and ring waveguides. In some complex microring resonator structures, multiple ring waveguides may be coupled together, or integrated with other optical components such as gratings and modulators to achieve more complex optical functions. MRRs are commonly used in various fields. For example, in optical communications, MRRs can be used as wavelength division multiplexing systems and optical switches. In optical computing, MRRs can be used for logic operations and optical storage.

[0022] A Mach-Zehnder interferometer (MZI) mainly consists of two beam splitters (BS) and two mirrors. The beam splitters are typically 50:50, meaning they split the incident light into two beams of equal intensity. Beam splitters usually employ optical coating technology to create specific thin-film structures on the surface of glass or other optical materials to achieve their beam-splitting function. The mirrors change the direction of light propagation, allowing the two beams to travel along different paths and then rejoin. Mirrors typically have high reflectivity to minimize energy loss during reflection. The two optical paths formed by the two beam splitters and mirrors constitute interference arms. The lengths of the two interference arms can differ; usually, one interference arm has a fixed length, while the length of the other can be adjusted to introduce an optical path difference. According to the principle of light interference, when two beams satisfy a certain phase relationship, constructive or destructive interference will occur. If the optical path difference between two beams is an integer multiple of the wavelength, the two beams will undergo constructive interference, resulting in a strong optical signal at the output. If the optical path difference is a half-integer multiple of the wavelength, destructive interference will occur, and the optical signal at the output will be weakened or even zero. By adjusting the length of one of the interference arms or the properties of the medium, the optical path difference between the two beams can be changed, thereby controlling the movement of the interference fringes and the intensity of the interference output.

[0023] Figure 1A schematic block diagram of a nonlinear processing method based on an all-optical computing architecture, according to an embodiment of this application, is shown. This nonlinear processing method is implemented on an all-optical computing architecture, meaning that each step in the nonlinear processing method 100 is achieved by a computer controlling the input of optical signals to various optical devices. For example, the nonlinear processing method is achieved by a processor in the computer controlling the input of optical signals to MRR and MZI. Optionally, as... Figure 1 As shown, the nonlinear processing method 100 includes some or all of the following contents.

[0024] S110, input the optical signal matrix block X into the micro-ring resonator MRR, and output the first optical signal, which is the maximum optical signal in the optical signal matrix block X.

[0025] S120, the optical signal matrix block X and the first optical signal are sequentially input into a cascaded Mach-Zehnder interferometer (MZI) and multiple series microring resonators (MRRs) to obtain a second optical signal corresponding to any optical signal in the optical signal matrix block X. The second optical signal is an optical signal corresponding to an exponent t with a base 2 power, where t = (x i -x max )·log2e,x i Let x be the value of any of the optical signals. max The value of the maximum optical signal.

[0026] S130, the second optical signal is sequentially input into the low-wavelength channel and the high-wavelength channel, and a third optical signal corresponding to any one of the optical signals is output, wherein the third optical signal is the local exponent value 2 of any one of the optical signals. t The corresponding optical signal, of which, The low-wavelength channel includes a Mach-Zehnder interferometer (MZI) and a microring resonator (MRR) for outputting 2 r The corresponding optical signal, where r is the fractional part of t, includes q cascaded Mach-Zehnder interferometers (MZIs) for outputting 2 q The corresponding optical signal, where q is the integer part of t.

[0027] S140, the third optical signal corresponding to each optical signal in the optical signal matrix block X is input to a plurality of micro-ring resonators (MRRs) connected in series, and a fourth optical signal is output. The fourth optical signal is the sum of the third signals corresponding to all optical signals in the optical signal matrix block X, and the fourth optical signal is the optical signal corresponding to the local exponent value of the optical signal matrix block X.

[0028] S150, modulate the third signal corresponding to any one of the optical signals into an optical signal corresponding to a first wavelength and modulate the fourth optical signal into an optical signal corresponding to a second wavelength, wherein the optical signal corresponding to the first wavelength is... The corresponding optical signal, the optical signal corresponding to the second wavelength is The corresponding optical signals, k1 and v1 are parameters of the first wavelength, k2 and v2 are parameters of the second wavelength, and v1 and v2 are constants less than 1.

[0029] S160, based on the magnitudes of v1 and v2, the optical signals corresponding to the first wavelength and the second wavelength are sequentially input into the microring resonator (MRR) array and the cascaded Mach-Zehnder interferometer (MZI), and a fifth optical signal corresponding to any of the optical signals is output. The fifth optical signal is the optical signal corresponding to the ratio of N1 to N2, wherein if v1 - v2 is greater than or equal to 0, the fifth optical signal is... The corresponding optical signal; if v1-v2 is less than 0, the fifth signal is The corresponding optical signal.

[0030] It should be noted that, Figure 1 The illustrated embodiment uses the calculation of the Softmax function on the original optical signal as an example. However, those skilled in the art will understand that the embodiments of this application can also calculate other nonlinear functions on the original optical signal, such as the Sigmoid function or the Tanh function. Only adjustments are needed. Figure 1 The order of the steps and the combination of various optical devices in the illustrated embodiments are as described. This application does not limit these aspects.

[0031] The following will use the Softmax function, i.e. For example, a detailed description Figure 1 The technical solution shown is as follows. First, based on mathematical principles, the exponential function with base e in the Softmax function can be converted into an equivalent operation that optical devices can perform through the following transformation.

[0032] First, use the base-change formula to change the exponent with the natural constant e to an exponent with the base 2:

[0033]

[0034] Where, x i Let x be the value of any optical signal in the optical signal matrix block X. max Let be the value of the maximum optical signal in the optical signal matrix block X, and t be the power of 2 corresponding to any optical signal, i.e., t = (x i -x max )·log2e.

[0035] Secondly, convert t to:

[0036] t=I·log2e≈I+(I>>1)+(I>>3)-(I>>4) (2);

[0037] Where I = x i -x max Therefore, after obtaining the value of each optical signal in the optical signal matrix block X and the value of the maximum optical signal, the exponent t corresponding to each optical signal can be calculated by shifting and addition / subtraction.

[0038] After obtaining the exponent t, further calculation of 2 is required. t Considering that t may be a non-integer, 2 t This might not be achievable entirely through bit shifting; in such cases, t can be decomposed into an integer part r and a fractional part q. Therefore, 2 t This can be converted into the following formula:

[0039] 2 t =2 q+r =2 q ·2 r =(qr)2 (3);

[0040] Where (qr)2 is 2 q ·2 r The binary bitstream representation.

[0041] According to Taylor's expansion on 2 r Making a further numerical approximation, we can obtain:

[0042] 2 r ≈1+(r>>1) (4);

[0043] Therefore, 2 t This can be converted into the following formula:

[0044] 2 t ≈(1+(r>>1))<<q (5).

[0045] Thus, through the derivation of formulas (1) to (5) above, the exponential function with base e can be transformed into shift operations and simple addition and subtraction operations, which can be completely performed by optical devices, as follows:

[0046] In step S110, the optical signal matrix block X can be input to the micro-ring resonator (MRR) to filter out the maximum optical signal in the optical signal matrix block X. It should be noted that step S110 is not a mandatory step, but can be optionally executed based on the actual situation. This filtering of the maximum optical signal can be used to avoid numerical overflow by subtracting the value of the maximum optical signal in subsequent steps.

[0047] After obtaining the maximum optical signal, the exponent t needs to be calculated. That is, in step S120, the maximum optical signal and the optical signal matrix block X can be sequentially input into multiple MZI cascaded and series-connected MRRs. The MZI cascade is used to implement the shift operation in the above formula (2), while the series-connected multiple MRRs are used to implement the addition and subtraction operations in formula (2). Finally, the optical signal corresponding to each optical signal in the optical signal matrix block X is converted into the optical signal corresponding to the exponent t with base 2. That is, each optical signal in the optical signal matrix block X corresponds to a second optical signal, and the second optical signal is the optical signal corresponding to the exponent t in formula 2. Through this method, the calculation of the exponent is simplified to addition and shift operations, which greatly reduces the hardware resource consumption and computational complexity.

[0048] Further, in step S130, the second optical signal corresponding to each optical signal output in step S120 is first sequentially input into the low-wavelength channel and the high-wavelength channel using wavelength solution. The low-wavelength channel processes the second optical signal through MZI phase modulation and MRR resonant frequency adjustment to obtain the optical signal corresponding to formula (4). That is, each optical signal in the optical signal matrix block X after passing through the low-wavelength channel can output a signal corresponding to 2... r The corresponding optical signal. Further, the optical signal passing through the low-wavelength channel can be input to the high-wavelength channel, which utilizes q cascaded MZIs to achieve fast left-shift processing, corresponding to the optical signal in formula (5). That is, each optical signal in the optical signal matrix block X after sequentially passing through the low-wavelength channel and the high-wavelength channel can output a corresponding 2... t The corresponding optical signal, that is, the output of step S130 is the optical signal obtained after performing an exponential function operation of e on each optical signal in the optical signal matrix block X. In step S130, the integer part can be achieved by cascading q MZIs. q The fast left shift calculation, combined with the approximation of the fractional part, significantly reduces computational complexity and the need for complex hardware computing units in traditional methods.

[0049] To complete the calculation of the sofmax function, it is necessary to further calculate the sum of all optical signals in the optical signal matrix block X after performing an exponential function operation with respect to e. That is, in step S140, the optical signals obtained in step S130 after performing an exponential function operation with respect to e can be allocated to different wavelengths λ1, λ2, λ3, ..., λ... n In multiple MRRs connected in series via fiber combiners, the resonant wavelength is set to λ. i When the input optical signal matches the resonance condition, the MRR couples the light to the ring cavity and accumulates it. The final output optical signal is the sum of the optical signals obtained after performing an exponential function operation of e on all optical signals in the optical signal matrix block X in step S130.

[0050] Next, combining the concepts of logarithms and antilogarithms, and calculating the intensity ratio based on the resonant wavelength selectivity of the MRR and the interference modulation of the MZI, exponential division can be achieved using a simple shift operation. Specifically, we modulate the third optical signal corresponding to each optical signal obtained in step S130 into an optical signal with a first wavelength λ1, and modulate the fourth optical signal corresponding to the optical signal matrix block X obtained in step S140 into an optical signal with a second wavelength λ2. The two modulated optical signals are input into the MRR array, and then a shift operation is achieved through MZI cascading, resulting in the optical signal output from any optical signal in the optical signal matrix block X after passing through the softmax function.

[0051] The specific derivation process is as follows:

[0052] First, the numerator term in the sofmax function can be expressed as: The denominator term can be expressed as: Where k1 and v1 are parameters of λ1, k2 and v2 are parameters of λ2, and v1 and v2 are constants less than 1. That is, once λ1 and λ2 are known, k1, v1, k2, and v2 can be obtained. The value Q of the optical signal output after any optical signal in the optical signal matrix block X passes through the softmax function is:

[0053]

[0054] Take the logarithm of Q:

[0055]

[0056] Combining this with Taylor expansion, we have:

[0057] log2Q≈(k1-k2-1)+(1+v1-v2) (8);

[0058] Then perform the antilogic operation:

[0059]

[0060] Finally, based on the size of v1-v2, Q can be simplified to the following two cases:

[0061]

[0062] First, v1-v2 can be calculated using the interference effect of MZI. Then, the resonant wavelength of the MRR will shift according to the exponential part of the input light intensity. By detecting the shift of the resonant wavelength, the exponential difference of k1-k2 can be obtained. In addition, the addition and subtraction operations in formula (10) can be calculated using the MRR array. Finally, the corresponding number of MZIs are concatenated to the exponential term to shift the corresponding number of bits to the left. In this way, division can be achieved through a simple shift operation of the binary bit stream, thereby efficiently updating the value of the nonlinear function in the optical computing architecture and improving the overall performance of the architecture.

[0063] Typically, in some high-speed optical communication systems, to improve signal processing speed and efficiency, the optical signal is split into multiple parallel processing channels. Each channel processes a portion of the signal, and the processed signals are then combined. This fully utilizes the parallel processing capabilities of multiple processing units, improving the overall system processing speed. Steps S110 to S150 described above only describe the softmax function operation for the optical signal matrix block X in one channel. In reality, each channel can be implemented using steps S110 to S150. To further optimize the effect of the softmax function on each optical signal, the aforementioned x... max Replace with the value of the largest optical signal among all optical signals, x' max That is, the global maximum value. For example... Figure 2 As shown, the nonlinear processing method 100 may further include the following steps:

[0064] S165, the value of the first optical signal corresponding to each of the plurality of optical signal matrix blocks X is stored in an optical buffer ring (OBR), and a sixth signal is output, wherein the sixth signal is the maximum value x' among the values ​​of the first optical signals corresponding to the plurality of optical signal matrix blocks X. max The corresponding optical signal stores the values ​​of the first optical signals corresponding to the plurality of optical signal matrix blocks X in an optical buffer ring (OBR) and outputs a sixth signal, wherein the sixth signal is the maximum value x' among the values ​​of the first optical signals corresponding to the plurality of optical signal matrix blocks X. max The corresponding optical signal;

[0065] S170, the sixth signal and the first signal corresponding to any of the optical signal matrix blocks X are sequentially input into a cascaded Mach-Zehnder interferometer (MZI) and a series of multiple microring resonators (MRRs) to obtain a seventh signal corresponding to any of the optical signal matrix blocks X. The seventh signal is an optical signal corresponding to a power t' to the base 2, where t' = x max -x' max ;

[0066] S175, the seventh signal is input sequentially to the low-wavelength channel and the high-wavelength channel, and the eighth optical signal corresponding to any optical signal matrix block X is output, wherein the eighth optical signal is 2. t′ The corresponding optical signal, of which, The low-wavelength channel is used to output 2. r′ The corresponding optical signal, r' being the fractional part of t', the high-wavelength channel comprising q' cascaded Mach-Zehnder interferometers (MZIs) for outputting 2 q′ The corresponding optical signal, q' is the integer part of t';

[0067] S180, modulate the fifth optical signal corresponding to any optical signal in the optical signal matrix block X into an optical signal corresponding to the third wavelength, and modulate the eighth optical signal corresponding to the optical signal matrix block X into an optical signal corresponding to the fourth wavelength, wherein the optical signal corresponding to the third wavelength is... The corresponding optical signal, the optical signal corresponding to the fourth wavelength is The corresponding optical signal, k3 and v3 are parameters of the third wavelength, k4 and v4 are parameters of the fourth wavelength, and v3 and v4 are constants less than 1;

[0068] S185, based on the values ​​of v3 and v4, the optical signal corresponding to the third wavelength and the optical signal corresponding to the fourth wavelength are sequentially input into the microring resonator (MRR) array and the cascaded Mach-Zehnder interferometer (MZI), and the ninth optical signal corresponding to any optical signal in the optical signal matrix block X is output. The ninth optical signal is the optical signal corresponding to the product of N3 and N4, and the ninth optical signal is the optical signal corresponding to the global exponent value of any optical signal in the optical signal matrix block X. Wherein, if v3 + v4 is less than 1, the ninth optical signal is... The corresponding optical signal; if v3 + v4 is greater than or equal to 1, the ninth optical signal is The corresponding optical signal.

[0069] Specifically, in step S110, the value x of the maximum optical signal in each channel's optical signal matrix block X is... maxThe data is stored in an optical storage device, such as an optical buffer ring (OBR). In step S165, the OBR dynamically adjusts the resonant wavelength through thermo-optical tuning, thereby dynamically updating the current global maximum value x'. max This means selecting the global maximum optical signal from the maximum optical signals output from multiple channels, thereby supporting the synchronization and consistency of multi-module operations.

[0070] Each optical signal corresponds to Converted to the following formula:

[0071]

[0072] In calculation Time, and calculation Similarly, first convert it to a base-2 exponent:

[0073]

[0074] Where, x max x' is the value of the maximum optical signal in the optical signal matrix block X. max Let t' be the value of the maximum optical signal among multiple optical signal matrix blocks X, i.e., the global maximum value, and let t' be the base-2 exponent corresponding to any optical signal matrix block X, i.e., t' = (x max -x′ max )·log2e.

[0075] Secondly, convert t' to:

[0076] t=I'·log2e≈I'+(I'>>1)+(I'>>3)-(I'>>4) (13);

[0077] Where, I' = x max -x' max Therefore, after obtaining the value of the maximum optical signal in optical signal matrix block X and the value of the maximum optical signal in multiple optical signal matrix blocks X, the exponent t' corresponding to each optical signal can be calculated by shifting and addition / subtraction.

[0078] After obtaining the exponent t', further calculation of 2 is required. t′ Considering that t' may be a non-integer, 2 t′ This might not be achievable entirely through bit shifting. In such cases, t' can be decomposed into an integer part r' and a fractional part q'. Therefore, t' can be transformed into the following formula:

[0079] 2 t′ =2 q′+r′ =2 q′ ·2 r′ =(q'.r')2 (14);

[0080] Where (q'.r')2 is 2 q′ ·2 r′ The binary bitstream representation.

[0081] According to Taylor's expansion on 2 r′ Making a further numerical approximation, we can obtain:

[0082] 2 r′ ≈1+(r'>>1) (15);

[0083] Therefore, 2 t This can be converted into the following formula:

[0084] 2 t′ ≈(1+(r'>>1))<<q' (16).

[0085] Thus, through the derivation of formulas (12) to (16) above, the exponential function with base e can be transformed into shift operations and simple addition and subtraction operations, which can be completely performed by optical devices, as follows:

[0086] In step S170, the maximum optical signal in any optical signal matrix block X and the maximum optical signals in multiple optical signal matrix blocks X can be sequentially input into multiple MZI cascaded and series-connected MRRs. The MZI cascade is used to implement the shift operation in the above formula (13), while the series-connected multiple MRRs are used to implement the addition and subtraction operations in formula (13). Finally, the optical signal corresponding to each optical signal matrix block X is output as the optical signal corresponding to the power of 2 t', that is, each optical signal matrix block X corresponds to a seventh optical signal, and the seventh optical signal is the optical signal corresponding to the power of t' in formula (13). Through this method, the calculation of the power is simplified to addition and shift operations, which greatly reduces the hardware resource consumption and computational complexity.

[0087] Further, in step S175, the seventh optical signal corresponding to each optical signal matrix block X output in step S170 is first sequentially input into the low-wavelength channel and the high-wavelength channel using wavelength solution. The low-wavelength channel processes the seventh optical signal through MZI phase modulation and MRR resonant frequency adjustment to obtain the optical signal corresponding to formula (15), that is, after passing through the low-wavelength channel, any optical signal matrix block X outputs a signal corresponding to 2... r′ The corresponding optical signal. Further, the optical signal passing through the low-wavelength channel can be input to the high-wavelength channel, which utilizes q' cascaded MZIs to achieve fast left-shift processing, corresponding to the optical signal in formula (16). That is, each optical signal matrix block X after sequentially passing through the low-wavelength channel and the high-wavelength channel can output a corresponding 2t′ The corresponding optical signal, that is, the output of step S175 is the optical signal obtained after performing an exponential function operation of e on the maximum optical signal in each optical signal matrix block X. In step S175, the integer part can be achieved by cascading q' MZIs. q′ The fast left shift calculation, combined with the approximation of the fractional part, significantly reduces computational complexity and the need for complex hardware computing units in traditional methods.

[0088] Next, combining the concepts of logarithms and antilogarithms, and based on the resonant wavelength selectivity of the MRR and the interference modulation of the MZI, the optical intensity multiplication calculation can be performed using a simple shift operation to achieve exponential multiplication. Specifically, we modulate the third optical signal corresponding to any optical signal in each optical signal matrix block X obtained in step S140 into a third wavelength λ3 optical signal, and modulate the eighth optical signal corresponding to each optical signal matrix block X obtained in step S175 into a fourth wavelength λ4 optical signal. Inputting the two modulated optical signals into the MRR array, and then performing a shift operation through MZI cascading, we can obtain the ninth optical signal corresponding to any optical signal in the optical signal matrix block X, i.e. The corresponding optical signal.

[0089] The specific derivation process is as follows:

[0090] First, the optical signal of the third wavelength is represented as The fourth wavelength optical signal is represented as Where k3 and v3 are parameters of λ3, k4 and v4 are parameters of λ4, and v3 and v4 are constants less than 1. That is, once λ3 and λ4 are known, k3, v3, k4, and v4 can be obtained. Then, the value M of the ninth optical signal corresponding to any optical signal in the optical signal matrix block X is:

[0091]

[0092] Similarly, by taking the logarithm of M, approximating it with a Taylor expansion, and then using the antilogarithm, we can obtain:

[0093]

[0094] Then, v3+v4 is calculated using the interference effect of MZI. Based on the case of v3+v4, it can be divided into two types:

[0095]

[0096] The principles and calculation processes of exponential multiplication and division are similar. For example, v3+v4 can be calculated using the interference effect of MZI. Then, the resonant wavelength of the MRR will shift according to the exponential part of the input light intensity. By detecting the shift of the resonant wavelength, the exponential difference of k3+k4 can be obtained. In addition, the addition and subtraction operations in formula (19) can be calculated using the MRR array. Finally, the corresponding number of MZIs are concatenated to the exponential term to shift the corresponding number of bits to the left. In this way, division can be achieved through a simple shift operation of the binary bit stream, thereby efficiently updating the value of the nonlinear function in the optical computing architecture and improving the overall performance of the architecture.

[0097] In Updated to After that, it is also necessary to Updated to

[0098] Where sum' is the sum of the global exponent values ​​of each optical signal matrix block X, and sum is the sum of the local exponent values ​​of each optical signal matrix block X.

[0099] Since the sum operation has already been performed in step S140 above, therefore, after calculating... Then, the sum of the local exponent values ​​in the aforementioned optical signal matrix block X can be multiplied by this sum to obtain the sum of the global exponent values ​​for each optical signal matrix block X.

[0100] Furthermore, after obtaining The corresponding optical signal and After obtaining the corresponding optical signal, steps S150 and S160 can be executed again. In this way, after considering the global maximum optical signal, the optical signal after performing the softmax function operation on each optical signal in the optical signal matrix block X can be output, thereby enhancing the stability of numerical calculation and avoiding numerical overflow problems caused by excessively large input values ​​in exponential operations.

[0101] In some embodiments, after obtaining the exponent t corresponding to each optical signal in the optical signal matrix block X, storing it in an optical storage device can bypass repetitive exponentiation operations, thereby greatly reducing computational complexity and hardware resource consumption.

[0102] In other embodiments, after obtaining the local exponent value of each optical signal in the optical signal matrix block X, it can also be stored in an optical storage device. After obtaining the global exponent value of each optical signal in the optical signal matrix block X, the local exponent value stored in the optical storage device can be replaced with the global exponent value.

[0103] In some other embodiments, after obtaining the local softmax function value of each optical signal matrix block X, it can also be stored in an optical storage device. After obtaining the global softmax function value of each optical signal matrix block X, the local softmax function value stored in the optical storage device can be replaced with the global softmax function value.

[0104] Figure 3 A schematic block diagram of a nonlinear processing device based on an all-optical computing architecture provided in an embodiment of this application is shown. Figure 3 As shown, the nonlinear processing device 200 includes some or all of the following components.

[0105] The maximum light selection module 210 is used to input the optical signal matrix block X into the micro-ring resonator MRR and output a first optical signal, wherein the first optical signal is the maximum optical signal in the optical signal matrix block X.

[0106] The optical phase shift module 220 is used to sequentially input the optical signal matrix block X and the first optical signal into a cascaded Mach-Zehnder interferometer (MZI) and multiple series microring resonators (MRRs) to obtain a second optical signal corresponding to any optical signal in the optical signal matrix block X. The second optical signal is an optical signal corresponding to a power t to the base 2, where t = (x... i -x max )·log2e,x i Let x be the value of any of the optical signals. max The value of the maximum optical signal.

[0107] The optical approximation index module 230 is used to sequentially input the second optical signal into the low-wavelength channel and the high-wavelength channel, and output a third optical signal corresponding to any one of the optical signals, wherein the third optical signal is the local index value 2 of the any one optical signal. t The corresponding optical signal, of which, The low-wavelength channel includes a Mach-Zehnder interferometer (MZI) and a microring resonator (MRR) for outputting 2 r The corresponding optical signal, where r is the fractional part of t, includes q cascaded Mach-Zehnder interferometers (MZIs) for outputting 2 q The corresponding optical signal, where q is the integer part of t.

[0108] The optical wavelength division multiplexing accumulation module 240 is used to input the third optical signal corresponding to each optical signal in the optical signal matrix block X into a series of multiple micro-ring resonators (MRRs) and output a fourth optical signal. The fourth optical signal is the sum of the third signals corresponding to all optical signals in the optical signal matrix block X, and the fourth optical signal is the optical signal corresponding to the local exponent value of the optical signal matrix block X.

[0109] The optical division shift module 250 is used to modulate the third signal corresponding to any one of the optical signals into an optical signal corresponding to a first wavelength and to modulate the fourth optical signal into an optical signal corresponding to a second wavelength, wherein the optical signal corresponding to the first wavelength is... The corresponding optical signal, the optical signal corresponding to the second wavelength is The corresponding optical signals, k1 and v1 are parameters of the first wavelength, k2 and v2 are parameters of the second wavelength, and v1 and v2 are constants less than 1; and based on the magnitudes of v1 and v2, the optical signals corresponding to the first wavelength and the second wavelength are sequentially input into a microring resonator (MRR) array and a cascaded Mach-Zehnder interferometer (MZI), outputting a fifth optical signal corresponding to any of the optical signals, wherein the fifth optical signal is the optical signal corresponding to the ratio of N1 to N2, wherein if v1-v2 is greater than or equal to 0, the fifth optical signal is... The corresponding optical signal; if v1-v2 is less than 0, the fifth signal is The corresponding optical signal.

[0110] In one possible embodiment, the maximum optical selection module 210 is further configured to: store the value of the first optical signal corresponding to each of the plurality of optical signal matrix blocks X in an optical buffer ring (OBR), and output a sixth signal, wherein the sixth signal is the maximum value x' among the values ​​of the first optical signals corresponding to the plurality of optical signal matrix blocks X. max The corresponding optical signal.

[0111] The optical phase shift module 220 is further configured to: sequentially input the sixth signal and the first signal corresponding to any of the optical signal matrix blocks X into a cascaded Mach-Zehnder interferometer (MZI) and a series of multiple microring resonators (MRRs) to obtain a seventh signal corresponding to any of the optical signal matrix blocks X, wherein the seventh signal is an optical signal corresponding to a power t' to the base 2, where t' = x max -x' max .

[0112] The optical approximation index module 230 is further configured to: input the seventh signal sequentially to the low-wavelength channel and the high-wavelength channel, and output the eighth optical signal corresponding to any optical signal matrix block X, wherein the eighth optical signal is 2. t′ The corresponding optical signal, of which, The low-wavelength channel is used to output 2. r′ The corresponding optical signal, r' being the fractional part of t', the high-wavelength channel comprising q' cascaded Mach-Zehnder interferometers (MZIs) for outputting 2 q′The corresponding optical signal, q' is the integer part of t'.

[0113] The nonlinear processing device 200 further includes: an optical multiplication and shift module 260, used to modulate the fifth optical signal corresponding to any optical signal in the optical signal matrix block X into an optical signal corresponding to a third wavelength, and to modulate the eighth optical signal corresponding to the optical signal matrix block X into an optical signal corresponding to a fourth wavelength, wherein the optical signal corresponding to the third wavelength is... The corresponding optical signal, the optical signal corresponding to the fourth wavelength is The corresponding optical signals, k3 and v3 are parameters of the third wavelength, k4 and v4 are parameters of the fourth wavelength, and v3 and v4 are constants less than 1; and based on the magnitudes of v3 and v4, the optical signals corresponding to the third wavelength and the fourth wavelength are sequentially input into the microring resonator (MRR) array and the cascaded Mach-Zehnder interferometer (MZI), outputting the ninth optical signal corresponding to any optical signal in the optical signal matrix block X, wherein the ninth optical signal is the optical signal corresponding to the product of N3 and N4, and the ninth optical signal is the optical signal corresponding to the global exponent value of any optical signal in the optical signal matrix block X, wherein if v3 + v4 is less than 1, the ninth optical signal is The corresponding optical signal; if v3 + v4 is greater than or equal to 1, the ninth optical signal is The corresponding optical signal.

[0114] Based on the same idea, this application also provides another nonlinear processing device based on an all-optical computing architecture. This nonlinear processing device includes a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the program instructions are executed by the processor, the nonlinear processing device performs the various method embodiments described above.

[0115] It should be noted that the details of the device-side embodiment can be found in the method-side embodiment, and for the sake of brevity, they will not be repeated here.

[0116] Based on the same idea, embodiments of this application also provide a computer-readable storage medium, which includes a stored program, wherein the program, when running, controls the device where the computer-readable storage medium is located to execute the various method embodiments described above.

[0117] Although this application has been described with reference to preferred embodiments, various modifications can be made thereto and components can be replaced with equivalents without departing from the scope of this application. In particular, the technical features mentioned in the various embodiments can be combined in any manner, provided there is no structural conflict. This application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A nonlinear processing method based on an all-optical computing architecture, characterized in that, The all-optical computing architecture includes a microring resonator (MRR) and a Mach-Zehnder interferometer (MZI), and the nonlinear processing method includes: The optical signal matrix block X is input into the micro-ring resonator MRR, and the first optical signal is output. The first optical signal is the maximum optical signal in the optical signal matrix block X. The optical signal matrix block X and the first optical signal are sequentially input into a cascaded Mach-Zehnder interferometer (MZI) and multiple series-connected microring resonators (MRRs) to obtain a second optical signal corresponding to any optical signal in the optical signal matrix block X. The second optical signal is an optical signal corresponding to an exponentiation t to the base 2. e, The value of any of the optical signals, The value of the maximum optical signal; The second optical signal is sequentially input into the low-wavelength channel and the high-wavelength channel, and a third optical signal corresponding to any one of the optical signals is output. The third optical signal is the local exponential value of any one of the optical signals. The corresponding optical signal, of which, The low-wavelength channel includes a Mach-Zehnder interferometer (MZI) and a microring resonator (MRR) for output. The corresponding optical signal, where r is the fractional part of t, includes q cascaded Mach-Zehnder interferometers (MZIs) for output. The corresponding optical signal, where q is the integer part of t; The third optical signal corresponding to each optical signal in the optical signal matrix block X is input to a series of multiple micro-ring resonators (MRRs) to output a fourth optical signal. The fourth optical signal is the sum of the third optical signals corresponding to all optical signals in the optical signal matrix block X, and the fourth optical signal is the optical signal corresponding to the local exponent value of the optical signal matrix block X. The third optical signal corresponding to any one of the optical signals is modulated into an optical signal corresponding to a first wavelength, and the fourth optical signal is modulated into an optical signal corresponding to a second wavelength, wherein the optical signal corresponding to the first wavelength is... The corresponding optical signal, the optical signal corresponding to the second wavelength is The corresponding optical signal, and The parameters of the first wavelength, and For the parameters of the second wavelength, and A constant less than 1; based on and The size of the optical signal is determined by sequentially inputting the optical signal corresponding to the first wavelength and the optical signal corresponding to the second wavelength into a microring resonator (MRR) array and a cascaded Mach-Zehnder interferometer (MZI), and outputting a fifth optical signal corresponding to any of the optical signals. The fifth optical signal is... and The ratio of the optical signals corresponding to the ratio, where if - The fifth optical signal is greater than or equal to 0. The corresponding optical signal; if - Less than 0, the fifth signal is The corresponding optical signal.

2. The nonlinear processing method according to claim 1, characterized in that, The nonlinear processing method further includes: The value of the first optical signal corresponding to each of the plurality of optical signal matrix blocks X is stored in the optical buffer ring (OBR), and a sixth signal is output, wherein the sixth signal is the maximum value among the values ​​of the first optical signals corresponding to the plurality of optical signal matrix blocks X. The corresponding optical signal; The sixth signal and the first optical signal corresponding to any of the optical signal matrix blocks X are sequentially input into a cascaded Mach-Zehnder interferometer (MZI) and multiple series microring resonators (MRRs) to obtain the seventh signal corresponding to any of the optical signal matrix blocks X. The seventh signal is a power of 2. The corresponding optical signal, of which, ; The seventh signal is input sequentially to the low-wavelength channel and the high-wavelength channel, and the eighth optical signal corresponding to any optical signal matrix block X is output. The eighth optical signal is... The corresponding optical signal, of which, The low-wavelength channel is used for output. The corresponding optical signal, for The fractional part of the high-wavelength channel includes A cascaded Mach-Zehnder interferometer (MZI) is used for output. The corresponding optical signal, for The integer part; The fifth optical signal corresponding to any optical signal in the optical signal matrix block X is modulated into an optical signal corresponding to a third wavelength, and the eighth optical signal corresponding to the optical signal matrix block X is modulated into an optical signal corresponding to a fourth wavelength. The optical signal corresponding to the third wavelength is... The corresponding optical signal, the optical signal corresponding to the fourth wavelength is The corresponding optical signal, and For the parameters of the third wavelength, and The parameters for the fourth wavelength are... and A constant less than 1; based on and The size of the optical signal is determined by sequentially inputting the optical signal corresponding to the third wavelength and the optical signal corresponding to the fourth wavelength into a microring resonator (MRR) array and a cascaded Mach-Zehnder interferometer (MZI), and outputting the ninth optical signal corresponding to any optical signal in the optical signal matrix block X. The ninth optical signal is... and The product of the products corresponds to the optical signal, and the ninth optical signal is the optical signal corresponding to the global exponent value of any optical signal in the optical signal matrix block X, wherein, if + Less than 1, the ninth optical signal is The corresponding optical signal; if + The ninth optical signal is greater than or equal to 1. The corresponding optical signal.

3. A nonlinear processing device based on an all-optical computing architecture, characterized in that, The all-optical computing architecture includes a microring resonator (MRR) and a Mach-Zehnder interferometer (MZI), and the nonlinear processing device includes: The maximum optical selection module is used to input the optical signal matrix block X into the micro-ring resonator MRR and output a first optical signal, wherein the first optical signal is the maximum optical signal in the optical signal matrix block X. An optical phase shift module is used to sequentially input the optical signal matrix block X and the first optical signal into a cascaded Mach-Zehnder interferometer (MZI) and multiple series-connected microring resonators (MRRs) to obtain a second optical signal corresponding to any optical signal in the optical signal matrix block X. The second optical signal is an optical signal corresponding to a power t to the base 2. e, The value of any of the optical signals, The value of the maximum optical signal; The optical approximation exponent module is used to sequentially input the second optical signal into the low-wavelength channel and the high-wavelength channel, and output a third optical signal corresponding to any one of the optical signals, wherein the third optical signal is the local exponent value of any one of the optical signals. The corresponding optical signal, of which, The low-wavelength channel includes a Mach-Zehnder interferometer (MZI) and a microring resonator (MRR) for output. The corresponding optical signal, where r is the fractional part of t, includes q cascaded Mach-Zehnder interferometers (MZIs) for output. The corresponding optical signal, where q is the integer part of t; The optical wavelength division multiplexing accumulation module is used to input the third optical signal corresponding to each optical signal in the optical signal matrix block X into multiple micro-ring resonators (MRRs) connected in series, and output a fourth optical signal. The fourth optical signal is the sum of the third optical signals corresponding to all optical signals in the optical signal matrix block X, and the fourth optical signal is the optical signal corresponding to the local exponent value of the optical signal matrix block X. The optical division shift module is used to modulate the third optical signal corresponding to any one of the optical signals into an optical signal corresponding to a first wavelength, and to modulate the fourth optical signal into an optical signal corresponding to a second wavelength, wherein the optical signal corresponding to the first wavelength is... The corresponding optical signal, the optical signal corresponding to the second wavelength is The corresponding optical signal, and The parameters of the first wavelength, and For the parameters of the second wavelength, and A constant less than 1; and based on and The size of the optical signal is determined by sequentially inputting the optical signal corresponding to the first wavelength and the optical signal corresponding to the second wavelength into a microring resonator (MRR) array and a cascaded Mach-Zehnder interferometer (MZI), and outputting a fifth optical signal corresponding to any of the optical signals. The fifth optical signal is... and The ratio of the optical signals corresponding to the ratio, where if - The fifth optical signal is greater than or equal to 0. The corresponding optical signal; if - Less than 0, the fifth signal is The corresponding optical signal.

4. The nonlinear processing apparatus according to claim 3, characterized in that, The maximum light selection module is also used for: The value of the first optical signal corresponding to each of the plurality of optical signal matrix blocks X is stored in the optical buffer ring (OBR), and a sixth signal is output, wherein the sixth signal is the maximum value among the values ​​of the first optical signals corresponding to the plurality of optical signal matrix blocks X. The corresponding optical signal; The optical phase shift module is also used for: The sixth signal and the first optical signal corresponding to any of the optical signal matrix blocks X are sequentially input into a cascaded Mach-Zehnder interferometer (MZI) and multiple series microring resonators (MRRs) to obtain the seventh signal corresponding to any of the optical signal matrix blocks X. The seventh signal is a power of 2. The corresponding optical signal, of which, ; The optical approximation index module is also used for: The seventh signal is input sequentially to the low-wavelength channel and the high-wavelength channel, and the eighth optical signal corresponding to any optical signal matrix block X is output. The eighth optical signal is... The corresponding optical signal, of which, The low-wavelength channel is used for output. The corresponding optical signal, for The fractional part of the high-wavelength channel includes A cascaded Mach-Zehnder interferometer (MZI) is used for output. The corresponding optical signal, for The integer part; The nonlinear processing device further includes: Optical multiplication and shift module, used for: The fifth optical signal corresponding to any optical signal in the optical signal matrix block X is modulated into an optical signal corresponding to a third wavelength, and the eighth optical signal corresponding to the optical signal matrix block X is modulated into an optical signal corresponding to a fourth wavelength. The optical signal corresponding to the third wavelength is... The corresponding optical signal, the optical signal corresponding to the fourth wavelength is The corresponding optical signal, and For the parameters of the third wavelength, and The parameters for the fourth wavelength are... and A constant less than 1; and based on and The size of the optical signal is determined by sequentially inputting the optical signal corresponding to the third wavelength and the optical signal corresponding to the fourth wavelength into a microring resonator (MRR) array and a cascaded Mach-Zehnder interferometer (MZI), and outputting the ninth optical signal corresponding to any optical signal in the optical signal matrix block X. The ninth optical signal is... and The product of the products corresponds to the optical signal, and the ninth optical signal is the optical signal corresponding to the global exponent value of any optical signal in the optical signal matrix block X, wherein, if + Less than 1, the ninth optical signal is The corresponding optical signal; if + The ninth optical signal is greater than or equal to 1. The corresponding optical signal.

5. A nonlinear processing device based on an all-optical computing architecture, characterized in that, include: A processor, and a memory communicatively connected to the processor, wherein the memory stores program instructions executable by the processor, and the processor invokes the program instructions to execute the nonlinear processing method based on the all-optical computing architecture as described in claim 1.

Images