WO2023039858A1

WO2023039858A1 - Optical computing system, optical computing method and control apparatus

Info

Publication number: WO2023039858A1
Application number: PCT/CN2021/119163
Authority: WO
Inventors: 周雷; 董晓文; 王戎
Original assignee: 华为技术有限公司
Priority date: 2021-09-17
Filing date: 2021-09-17
Publication date: 2023-03-23
Also published as: CN117980920A

Abstract

The present application provides an optical computing system, an optical computing method and a control apparatus, and belongs to the technical field of optical communications. The system comprises an optical signal providing unit, an optical distribution unit and an optical computing network unit, wherein the optical distribution unit comprises an optical steering element and an optical convergence element; the optical signal providing unit is used for providing a plurality of first optical signals; the optical steering element is used for steering the plurality of first optical signals; the optical convergence element is used for converging the plurality of steered first optical signals into at least one second optical signal, and for transmitting the at least one second optical signal to the optical computing network unit; and the optical computing network unit is used for receiving the at least one second optical signal, and for modulating the at least one second optical signal.

Description

Optical computing system, optical computing method and control device

technical field

The present application relates to the technical field of optical communication, and in particular to an optical computing system, an optical computing method and a control device.

Background technique

An optical computing system refers to a system that uses photons as a medium for computing. Compared with electrons, photons have the advantages of high speed, high bandwidth, and low power consumption, so optical computing systems are more suitable for processing large-scale convolution operations.

The optical computing system includes an optical signal providing unit and an optical computing network unit. The optical signal providing unit provides multiple optical signals, and the multiple optical signals respectively carry multiple input information. The optical computing network unit has multiple computing nodes, and the multiple computing nodes communicate with The multi-channel optical signals correspond one-to-one, and the calculation node performs amplitude modulation on the corresponding optical signal according to the preset convolution kernel coefficient, so as to perform the multiplication operation of the convolution kernel coefficient and the input information. The convolution operation of the input matrix composed of a plurality of input information and the convolution kernel matrix composed of a plurality of convolution kernel coefficients is realized in the above manner.

In related technologies, there is a one-to-one correspondence between the multi-channel optical signals of the modulation unit and the computing nodes, and the corresponding relationship between the multiplied convolution kernel coefficients and the input information is relatively fixed, which makes it difficult for the optical computing system to meet the volume requirements in some scenarios. Cumulative computing needs.

Contents of the invention

The present application provides an optical computing system, an optical computing method and a control device, and the technical solution is as follows:

In one aspect, an optical computing system is provided, including an optical signal providing unit, an optical distribution unit, and an optical computing network unit, where the optical distribution unit includes an optical steering element and an optical converging element.

Wherein, the optical signal providing unit is used to provide multiple channels of first optical signals; the optical diverting element is used to divert multiple channels of first optical signals; the optical converging element is used to converge the diverted multiple channels of first optical signals into at least one channel The second optical signal transmits at least one second optical signal to the optical computing network unit; the optical computing network unit is configured to receive at least one second optical signal and modulate at least one second optical signal.

In the optical computing system provided in this application, the optical signal providing unit provides multiple optical signals, and the steering and convergence of the corresponding relationship of the optical signals can be controlled through the optical steering element and the optical converging element, thereby controlling the input information carried by the optical signal to be transmitted to the optical Calculate which computing node of the network unit, and then realize the flexible configuration of the corresponding relationship between the input information and the convolution kernel coefficient, so that the system can control the coefficient corresponding to each input information when performing convolution calculation, so as to ensure the realization of the convolution operation The flexible configuration, such as using different convolution kernel coefficient matrices to convolute images, can meet the convolution calculation requirements in various scenarios.

In some possible implementation manners, the optical signal providing unit includes a light source unit and a modulation unit, and the optical computing network unit includes at least one computing node and at least one demultiplexer, and usually there are multiple computing nodes and demultiplexers, Taking multiple as an example, multiple computing nodes are divided into multiple groups.

Wherein, the light source unit is used to provide multiple channels of third optical signals; the modulation unit is used to receive multiple channels of third optical signals and modulate the multiple channels of third optical signals to obtain multiple channels of first optical signals, and the first optical signals carry Input information; the light steering element is used to control the direction of multiple first optical signals; the light converging element is used to converge the multiple first optical signals passing through the light steering element into at least one second optical signal; the demultiplexer receives at least one A second optical signal, and at least one second optical signal is decoupled, and the decoupled first optical signal is output to a computing node, and the computing node modulates the amplitude of the received first optical signal.

In the optical computing system provided in this application, the light source unit provides multiple optical signals, and the modulation unit performs the first modulation on the multiple optical signals, and the amplitude of the optical signal after the first modulation is used to represent the input information. That is, the input information is modulated onto the optical signal to prepare for the calculation of the input information; after the first step of modulation, the first modulated optical signal is distributed to each computing node in the optical computing network through the optical distribution unit , that is, by steering and converging the optical signal, it is sent to the optical computing network unit. The optical computing network unit includes a demultiplexer and a computing node. For the received optical signal, it is demultiplexed by the demultiplexer first. coupling, so that the optical signals converged by the optical converging element are separated, so that each computing node is distributed to the optical signal carrying the corresponding input information, and these multiple computing nodes complete the second modulation of these optical signals. Secondary modulation is also amplitude modulation, and the amplitude of the optical signal after the second modulation represents the magnitude of the input information multiplied by a coefficient. The above two-step calculation completes the multiplication part in the convolution operation. In the above process, the corresponding relationship between the input information and the computing node can be controlled through the light steering element and the light converging element, so as to control which computing node the input information is transmitted to, and then the flexible configuration of the corresponding relationship between the input information and the convolution kernel coefficient can be realized .

Taking the image convolution operation as an example, the convolution kernel coefficients of each computing node in the optical computing network form a convolution kernel coefficient matrix, and the input information carried by multiple first optical signals forms an input pixel amplitude matrix. Through the light steering element and the light converging element, the pixel amplitude matrix can be input to the set calculation node. The scale and position of the set calculation node are variable, that is, the convolution kernel coefficient matrix of different sizes is used to perform image processing. Convolution processing.

In some possible implementation manners, the number of channels of the first optical signal provided by the optical signal providing unit is the same as the number of computing nodes in the optical computing network unit.

Each path of the first optical signal can be allocated to each computing node in a one-to-one correspondence through the light steering element and the light converging element. It is also possible to divert and converge only a part of the first optical signals, so that part of the first optical signals are distributed to some computing nodes.

In some other possible implementation manners, the number of paths of the first optical signal provided by the optical signal providing unit is different from the number of computing nodes in the optical computing network unit, for example, the number of paths of the first optical signal is greater than the number of computing nodes, or the number of paths of the first optical signal is greater than the number of computing nodes, or The number of paths of an optical signal is less than the number of computing nodes.

All or part of the first optical signals can be diverted and converged by the light diverting element and the light converging element, and then distributed to all or part of the computing nodes.

In the above-mentioned optical computing process, there are cases where multiple first optical signals are gathered together, and then the required first optical signals are decomposed in the optical computing network unit. In order to ensure that the aggregated first optical signals can be re-decomposed, it is necessary to ensure that the wavelengths or modes of the first optical signals are different from each other.

In some embodiments of the present application, the multiple first optical signals are different in wavelength and mode at the same time, that is, at least one of the wavelength and mode of the two first optical signals is different.

Exemplarily, the multiple first optical signals include a first optical signal, a second optical signal, and a third optical signal, the modes of the first optical signal and the second optical signal are different, and the second optical signal and The wavelengths of the third optical signals are different. In this way, different optical signals are provided from two dimensions of wavelength and mode, so that the optical computing network can support larger-scale convolution operations, making the optical computing system more widely used.

In another implementation manner of the present application, the multiple first optical signals differ only in one dimension of wavelength or mode, which will not be repeated here.

It should be noted that, in the present application, different light modes refer to different transverse mode modes, such as LP00, LP01, LP10, and the like. In the field of optics, the wavelength of light corresponds to the longitudinal mode of light.

The implementation of the light source unit is given below when multiple channels of first optical signals are simultaneously different in two dimensions: wavelength and mode.

In some possible implementations, the light source unit includes a multi-wavelength light source and multiple mode converters.

Wherein, the multi-wavelength light source is used to provide multiple channels of fourth optical signals with different wavelengths; each mode converter is used to receive one of the multiple channels of fourth optical signals with different wavelengths, and is used to convert the received channel of the fourth optical signal to The four optical signals are converted into multiple third optical signals with different modes, thereby obtaining multiple third optical signals, and any two of the third optical signals have different wavelengths, or different modes, or both different wavelengths and modes.

In some other possible implementation manners, the light source unit includes a multi-mode light source and multiple mode converters.

Wherein, the multi-mode light source is used to provide multiple fourth optical signals with different modes; each mode converter is used to receive one fourth optical signal among the multiple fourth optical signals with different modes, and is used to convert the received one fourth optical signal to The four optical signals are converted into multiple channels of third optical signals with different wavelengths. It is worth noting that, in the former implementation of the light source unit, the function of the mode converter is to convert the transverse mode of light, so as to convert one beam of light into multiple beams of light with the same wavelength and different modes. However, in this implementation of the light source unit, the function of the mode converter is to convert the longitudinal mode of light, so as to convert one beam of light into multiple beams of light with the same mode and different wavelengths.

Both of the above two implementation methods can provide multiple optical signals with different wavelengths and modes at the same time, so that the optical computing network can support larger-scale convolution operations, making the optical computing system more widely used.

In some possible implementation manners, the modulating unit includes a plurality of modulators, and each modulator is used to modulate a third optical signal, for example, modulate a corresponding third optical signal with an electrical signal, so that the electrical signal carries The input information is modulated onto the third optical signal to obtain the first optical signal. Here, each channel of the first optical signal only includes a component modulated by the channel of the third optical signal. The modulator is an electro-optical conversion device. The bias voltage of the modulator is controlled by an electrical signal, and the amplitude of the input optical signal is modulated so that it can represent the input information carried by the electrical signal.

Exemplarily, the modulator and the third optical signal may be set one-to-one.

Exemplarily, the modulators and the third optical signals may not be arranged one-to-one, for example, the number of modulators is less than that of the third optical signals, or the number of modulators is greater than that of the third optical signals.

In some other possible implementation manners, the modulation unit includes multiple couplers and multiple modulators, and each coupler may correspond to one modulator.

Multiple couplers are used to couple multiple channels of third optical signals into multiple channels of fifth optical signals; multiple modulators are used to use multiple channels of first electrical signals to respectively modulate multiple channels of fifth optical signals to obtain multiple channels of first optical signals. For the optical signal, each first electrical signal is obtained by coupling multiple second electrical signals with different frequencies. Here, each first optical signal includes components modulated by multiple third optical signals. In this case, the calculation node in the subsequent optical processing network unit modulates the first optical signal for one of the first optical signals. portion.

Exemplarily, the multiple third optical signals corresponding to each coupler may be optical signals carrying input information corresponding to one or several rows of computing nodes in the optical computing network unit, and the multiple third optical signals coupled by one coupler There may be no special requirements on wavelength or mode. Of course, multiple third optical signals of the same wavelength or mode can also be used to carry the input information corresponding to one or several rows of computing nodes in the optical computing network unit, and then be coupled through a coupler.

In this implementation manner, since the third optical signal is coupled first, and then the coupled optical signal is modulated, the number of required modulators is reduced, which is beneficial to reducing the size and cost of the entire optical computing system.

The generation process of the first electrical signal is described below:

First, the digital circuit provides multiple digital frequency signals with the same frequency, and each digital frequency signal has already carried input information; multiple multipliers are used to convert the frequency of the multiple digital frequency signals to obtain multiple channels with different frequencies. Digital frequency signal (second electrical signal); then use the radio frequency source to couple multiple digital frequency signals to obtain multiple coupled electrical signals; use a digital-to-analog converter to convert the digital signal into an analog signal to obtain multiple first electric signal. In this way, the digital-to-analog conversion is performed after the electrical signal is coupled, which reduces the number of digital-to-analog converters and is beneficial to reduce the size and cost of the entire optical computing system.

Wherein, the digital circuit that provides the digital frequency signal may be hardware that needs to perform convolution calculation, such as a graphics card, a processor, and the like.

It is worth noting that in the above-mentioned process of frequency conversion using a multiplier, all digital frequency signals can be converted into different frequencies, or the digital frequency signals that need to be coupled into one electrical signal can be converted into frequency according to the subsequent coupling relationship. Different frequencies, and the frequencies of the digital frequency signals coupled into different circuit electrical signals after frequency conversion may be the same or different.

When the above coupled electrical signal is used to modulate the optical signal, each modulated first optical signal is actually modulated by multiple second electrical signals at the same time, but when actually representing input information, each The first optical signal only needs to correspond to one second electrical signal, so after passing through the optical computing network unit, the optical signal needs to be filtered to filter out components corresponding to redundant electrical signals during modulation.

Exemplarily, the optical computing system further includes at least one filter.

Each filter is used to filter the optical signal output by at least one computing node, and retain a component corresponding to one of the multiple second electrical signals with different frequencies. Here, filtering is actually filtering out components of the optical signal. When modulated by the modulation unit, the optical signal is modulated by multiple second electrical signals at the same time, so that the amplitude of the optical signal is the sum of the modulation results of the multiple second electrical signals. The component modulated by the electrical signal other than the signal is filtered out.

In some other possible implementations, the optical signal providing unit includes a multi-wavelength light source and multiple mode converters, or the optical signal providing unit includes a multi-mode light source and multiple mode converters. For details, refer to the previous description about the light source unit. . It is worth noting that multiple mode converters convert the optical signals output by the multi-wavelength light source or multi-mode light source to obtain multiple optical signals with different modes, or multiple optical signals with different wavelengths. The signal or multiple optical signals with different wavelengths are used to generate multiple first optical signals.

Exemplarily, the optical signal providing unit may further include a modulating unit, configured to modulate the optical signals output by the multiple mode converters, so as to generate multiple channels of first optical signals.

Exemplarily, the optical signal supply unit may not include a modulation unit, and the optical signals output by multiple mode converters do not need to be modulated by the modulation unit, and the input information has been carried on the optical signal, that is, the output is a multi-channel first light signal.

Exemplarily, the light redirecting element comprises a liquid crystal on silicon cell. The light diverting element may include a plurality of liquid crystal on silicon units, and each liquid crystal on silicon unit receives an optical signal output by a modulation unit, and controls the propagation direction of the optical signal, for example, controls the steering of the optical signal, or does not change the direction of the optical signal. The liquid crystal on silicon cell is a voltage-driven element. By controlling the voltage of the liquid crystal cell and changing its electric field, the deflection direction of the liquid crystal is changed, thereby changing the propagation path of light passing through the liquid crystal cell.

The number of liquid crystal on silicon cells in the light steering element can be equal to the number of light signals output by the modulation unit, and the number of liquid crystal on silicon cells in the light steering element can also be greater than the number of light signals output by the modulation unit.

Exemplarily, the light converging element includes a diffraction grating, and the diffraction grating converges the light incident in a specific direction into a beam, and then transmits it to the optical computing network unit. Using the diffraction grating as the light converging element makes the structure of the light converging element relatively simple, thereby The volume of the entire optical computing system can be reduced.

In some possible implementation manners, by changing the voltage applied to the liquid crystal on silicon cell, it is possible to flexibly adjust the propagation direction of the optical signal.

Exemplarily, the optical computing system further includes a control device.

The control device is used to control the liquid crystal on silicon cell to realize at least one of the following controls:

Controlling the composition of the first optical signal converged into a beam of light (one second optical signal) in the light converging element, that is, controlling which first optical signals will be converged together after arriving at the light converging element;

Control the first optical signal sent to the first computing node in the optical computing network unit through the optical converging element, the first computing node is any computing node, that is, control the first optical signal to be finally emitted to the optical computing network unit which location. Since the position of each optical signal converged by the light converging element is fixed to the optical computing network unit, it is controlled where the second optical signal is diverted through the liquid crystal on silicon unit to achieve converging at the optical converging element.

The steering control of the optical signal is described below in combination with the structure of the optical computing network unit:

As mentioned above, the optical computing network unit includes multiple computing nodes arranged in an array, that is, divided into multiple rows and columns, wherein one row or one column of computing nodes constitutes a group in the above. The following takes a column of compute nodes as a group as an example.

In some possible implementations, the first optical signal corresponding to a row of computing nodes is transmitted to a specific position of the light converging element through the light steering element, converged into a beam of light through the light converging element, and then transmitted to the optical computing device through the light converging element. A row in a network element. A row of computing nodes is connected through an optical transmission medium, so that a row of computing nodes can obtain corresponding optical signals.

For example, the optical converging element converges the diverted multiple first optical signals into N beams (N second optical signals), and the N beams of light correspond to the optical transmissions that are sent to the N rows of computing nodes connected to the optical computing network unit. Medium, N is a positive integer.

In some other possible implementation manners, the first optical signals corresponding to multiple rows of computing nodes are transmitted to a specific position of the light converging element through the light diverting element, converged into a beam of light through the light converging element, and then transmitted to the Multiple rows in optical computing network units. Multiple rows of computing nodes are connected through an optical transmission medium, so that multiple rows or columns of computing nodes can obtain corresponding optical signals.

Here, after the aggregated optical signal is sent to the optical computing network unit, the demultiplexer performs decoupling, and outputs the decoupled first optical signal to the corresponding computing node.

Taking a row of computing nodes corresponding to a bundle of converged light as an example, each computing node is connected to the optical transmission medium of the row through a demultiplexer, which can receive the optical signal and decouple it to obtain the corresponding first light signal.

Exemplarily, the optical computing network unit may be one processing unit, or may be composed of multiple processing units. For example, the optical computing network may be composed of multiple optical processing units, and each optical processing unit is a processing unit. Each optical processing unit can contain multiple rows or columns of computing nodes and multiple demultiplexers. When performing light distribution, each optical processing unit can correspond to a beam of converged light, or one optical processing unit can Converging beams of light.

That is, the optical computing network unit includes at least one optical processing unit, and transmits at least one second optical signal to the optical computing network unit, including:

According to the processing capability of the at least one optical processing unit, at least one second optical signal is sent to the at least one optical processing unit.

Wherein, the processing capability of the optical processing unit is usually determined by the number of computing nodes it contains, and according to the processing capability of each optical processing unit, it is determined whether at least one second optical signal is sent to one or more optical processing units, so that the optical The computing system allocates the optical processing unit to process the optical signal according to needs, so as to avoid excess or insufficient processing capacity of the provided optical processing unit.

Exemplarily, a computing node includes a modulator. Alternatively, the compute node includes a phase change material for modulating the amplitude of the received optical signal.

Regardless of the computing node implemented in that way, the computing node performs amplitude modulation on the corresponding first optical signal according to the convolution kernel coefficient to perform multiplication of the convolution kernel coefficient and the input information on the corresponding first optical signal Operation, multiple computing nodes correspond to multiple convolution kernel coefficients one by one.

Optionally, the optical computing system further includes: a converting unit, configured to add and convert the optical signals output by the optical computing network unit into electrical signals. The same group of multiplication calculation results are added through the conversion unit, and converted into an electrical signal output, so that the calculation result can be applied to the calculation of the subsequent electrical processing unit.

Exemplarily, the conversion unit includes a plurality of photoelectric converters, each photoelectric converter corresponds to a group of computing nodes, the output of this group of computing nodes is connected to the input of the same photoelectric converter, and the output of the photoelectric converter to this group of computing nodes The optical signal is converted to photoelectricity to realize the addition and conversion of the optical signal mentioned above.

Exemplarily, the optical computing system is used for computing in image processing, or the optical computing system is used for computing in digital signal processing, or the optical computing system is used for computing in a processor. The optical computing system provided by this application can provide faster computing speed and larger computing scale, and can be applied in the above-mentioned fields to further improve the speed and scale of image processing and digital signal processing.

In another aspect, an optical computing system is provided, including a light source unit, a modulation unit, and an optical computing network unit. The modulation unit includes a coupler and a modulator.

Wherein, the light source unit is used to provide multiple channels of first optical signals; the coupler is used to couple multiple channels of first optical signals into at least one channel of second optical signals; The signal is modulated to obtain a modulated second optical signal, and each first electrical signal in at least one first electrical signal is obtained by coupling multiple second electrical signals with different frequencies; the optical computing network unit is used for At least one second optical signal is received, and at least one second optical signal is modulated.

It is worth noting that the above-mentioned first optical signal and second optical signal are only used to represent the change process of the optical signal in this implementation mode, and the first optical signal here and the first optical signal defined in the previous aspect may be different optical signals. Signal.

In the optical computing system provided in this application, the light source unit provides multiple optical signals, and the modulation unit performs the first modulation on the multiple optical signals, and the amplitude of the optical signal after the first modulation is used to represent the input information. That is to say, the input information is modulated onto the optical signal; after the first step of modulation, the first modulated optical signal is distributed to the optical computing network unit through the optical distribution unit, and the optical computing network unit completes the first step of these optical signals Secondary modulation, the second modulation is also amplitude modulation, and the amplitude of the optical signal after the second modulation represents the magnitude of the input information multiplied by a coefficient. The above two-step calculation completes the multiplication part in the convolution operation. In the above process, the optical signals are coupled together by a coupler and then modulated, so that the number of required modulators is greatly reduced, which is beneficial to reduce the size and cost of the entire optical computing system.

In the above optical computing process, multiple first optical signals are coupled together, then modulated, and then decomposed into required second optical signals in the optical computing network unit. In order to ensure that the coupled optical signals can be decomposed again, it is necessary to ensure that the wavelengths or modes of the first optical signals of each channel are different.

In some embodiments of the present application, the wavelengths and modes of the multiple first optical signals are different at the same time, that is, at least one of the wavelengths and modes of the two first optical signals is different.

Exemplarily, the multiple first optical signals include a first optical signal, a second optical signal, and a third optical signal, the modes of the first optical signal and the second optical signal are different, and the second optical signal and The wavelengths of the third optical signals are different.

Wherein, the multi-wavelength light source is used to provide multiple channels of third optical signals with different wavelengths; each mode converter is used to receive one of the multiple channels of third optical signals with different wavelengths, and is used to convert the received channel of the third optical signal to The three optical signals are converted into multiple first optical signals with different modes, thereby obtaining multiple first optical signals, and any two of them have different wavelengths, or different modes, or both different wavelengths and modes.

Among them, the multi-mode light source is used to provide multiple third optical signals with different modes; each mode converter is used to receive one of the third optical signals among the multiple third optical signals with different modes, and is used to convert the received third optical signal The three optical signals are converted into multiple channels of first optical signals with different wavelengths. It is worth noting that, in the former implementation of the light source unit, the function of the mode converter is to convert the transverse mode of light, so as to convert one beam of light into multiple beams of light with the same wavelength and different modes. However, in this implementation of the light source unit, the function of the mode converter is to convert the longitudinal mode of light, so as to convert one beam of light into multiple beams of light with the same mode and different wavelengths.

As mentioned above, the first electrical signal is obtained by coupling multiple second electrical signals with different frequencies. The generation process of the first electrical signal is described below:

First, the digital circuit provides multiple digital frequency signals with the same frequency, and each digital frequency signal has already carried input information; multiple multipliers are used to convert the frequency of the multiple digital frequency signals to obtain multiple channels with different frequencies. A digital frequency signal (second electrical signal); then use a radio frequency source to couple multiple digital frequency signals to obtain a coupled electrical signal; use a digital-to-analog converter to convert the digital signal into an analog signal to obtain the first electrical signal. In this way, the digital-to-analog conversion is performed after the electrical signal is coupled, which reduces the number of digital-to-analog converters and is beneficial to reduce the size and cost of the entire optical computing system.

It is worth noting that, in the above-mentioned process of frequency conversion by using a multiplier, all digital frequency signals are converted into different frequencies.

Exemplarily, the optical computing network unit includes at least one demultiplexer, and the at least one demultiplexer is configured to receive at least one second optical signal and decouple at least one second optical signal.

For example, each demultiplexer is used to receive the optical signal output by the modulator, decouple the optical signal output by the modulator, output the decoupled optical signal to the corresponding computing node, and separate the coupled optical signals , so that each computing node is assigned an optical signal carrying corresponding input information.

When the above coupled electrical signal is used to modulate the optical signal, each modulated optical signal is actually modulated by multiple second electrical signals at the same time, but when actually representing input information, each optical signal It only needs to correspond to one second electrical signal, so after passing through the optical computing network unit, the optical signal needs to be filtered to filter out the component corresponding to the redundant electrical signal during modulation.

Exemplarily, the optical computing network unit includes at least one computing node, an optical computing system, and a plurality of filters.

Optionally, the optical computing system further includes an optical distribution unit. The optical distribution unit is responsible for sending the optical signal output by the modulation unit to the optical computing network unit.

In some possible embodiments, the light distribution unit comprises a light redirecting element.

The light diverting element is used to divert at least one second optical signal, so that the at least one second optical signal is sent to at least one optical processing unit.

In this way, when the light turning element is controlled, at least one second optical signal can be transmitted to at least one optical processing unit according to the processing capability of at least one optical processing unit. Wherein, the processing capability of the optical processing unit is usually determined by the number of computing nodes it contains, and according to the processing capability of each optical processing unit, it is determined whether at least one second optical signal is sent to one or more optical processing units, so that the optical The computing system allocates the optical processing unit to process the optical signal according to needs, so as to avoid excess or insufficient processing capacity of the provided optical processing unit.

Exemplarily, the light redirecting element comprises a liquid crystal on silicon cell. By changing the voltage applied to the silicon-based liquid crystal unit, it is possible to flexibly adjust the propagation direction of the optical signal.

When the optical computing network unit has only one input optical signal, in order to ensure that each computing node can obtain one second optical signal in the input optical signal. Each computing node in the optical computing network unit is connected through an optical transmission medium. Alternatively, the optical distribution unit sends one optical signal to computing nodes in multiple rows or columns at the same time.

Exemplarily, a computing node includes a modulator. Alternatively, the computing node includes a phase change material for modulating the amplitude of the received second optical signal.

In another aspect, a light calculation method is provided, the method comprising:

providing multiple first optical signals;

Steering the multiple first optical signals;

Converge the diverted multiple first optical signals into at least one second optical signal, and transmit at least one second optical signal to the optical computing network unit;

The at least one second optical signal is modulated by the optical computing network unit.

Optionally, the method also includes:

Before modulating the at least one second optical signal, at least one second optical signal is decoupled.

Exemplarily, the optical computing network unit includes at least one optical processing unit, and transmits at least one second optical signal to the optical computing network unit, including:

providing multiple first optical signals;

coupling multiple first optical signals into at least one second optical signal;

At least one first electrical signal is used to modulate at least one second optical signal to obtain a modulated second optical signal, and each first electrical signal in the at least one first electrical signal is obtained by pairing multiple first electrical signals with different frequencies. It is obtained by coupling two electrical signals;

Optionally, the method also includes:

Optionally, the optical computing network unit includes at least one computing node, and the method further includes:

The optical signal output by at least one computing node is filtered, and a component corresponding to one of the multiple second electrical signals with different frequencies is retained.

In another aspect, a control device is provided, including a processor and a memory; program codes are stored in the memory, and when the program codes are executed by the processor, the method described in any one of the preceding items is implemented.

Description of drawings

FIG. 1 is a schematic structural diagram of an optical computing system provided by some embodiments of the present application;

FIG. 2 is a schematic structural diagram of an optical computing system provided by some embodiments of the present application;

Fig. 3 is a schematic structural diagram of a light source unit provided by some embodiments of the present application;

Fig. 4 is a schematic structural diagram of a light source unit provided by some embodiments of the present application;

Fig. 5 is a partial structural schematic diagram of an optical computing system provided by some embodiments of the present application;

Fig. 6 is a partial structural schematic diagram of an optical computing system provided by some embodiments of the present application;

Fig. 7 is a partial structural schematic diagram of an optical computing system provided by some embodiments of the present application;

Fig. 8 is a partial structural schematic diagram of an optical computing system provided by some embodiments of the present application;

Fig. 9 is a partial structural schematic diagram of an optical computing system provided by some embodiments of the present application;

FIG. 10 is a schematic structural diagram of an electrical signal generating circuit provided by some embodiments of the present application;

Fig. 11 is a partial structural schematic diagram of an optical computing system provided by some embodiments of the present application;

Fig. 12 is a schematic structural diagram of an electrical signal generating circuit provided by some embodiments of the present application;

Fig. 13 is a flowchart of an optical calculation method provided by some embodiments of the present application;

Fig. 14 is a flowchart of an optical calculation method provided by some embodiments of the present application;

Fig. 15 is a schematic structural diagram of a control device provided by some embodiments of the present application.

Detailed ways

Fig. 1 is a schematic structural diagram of an optical computing system provided by some embodiments of the present application. Referring to FIG. 1 , the optical computing system includes: an optical signal providing unit 10 , an optical distribution unit 20 and an optical computing network unit 30 .

The output end of the optical signal supply unit 10 is connected to the input end of the optical distribution unit 20 , and the output end of the optical distribution unit 20 is connected to the input end of the optical computing network unit 30 .

The optical signal providing unit 10 is used to provide multiple first optical signals; the optical distribution unit 20 is used to distribute the multiple first optical signals to the optical computing network unit 30; the optical computing network unit 30 is used to receive the multiple first optical signals , and modulate the multiple first optical signals.

Exemplarily, the optical distribution unit may be implemented in the following manner: First, multiple channels of first optical signals are aggregated into multiple channels of second optical signals, and then sent to the optical computing network unit 30 .

Fig. 2 is a schematic structural diagram of an optical computing system provided by some embodiments of the present application. Referring to FIG. 2 , the optical signal providing unit 10 may include a light source unit 11 and a modulation unit 12 . The light source unit 11 is used to provide multiple third optical signals; the modulation unit 12 is used to receive multiple third optical signals, modulate the multiple third optical signals, and obtain multiple first optical signals, and the first optical signals carry Enter information.

Exemplarily, the optical computing network unit 30 may have at least one computing node 31, and the number of computing nodes 31 is usually multiple. Taking multiple computing nodes as an example, multiple computing nodes 31 are divided into multiple groups, and the computing nodes 31 are used for receiving The amplitude of the received first optical signal is modulated.

Optionally, the optical computing system may further include a conversion unit 50, the input end of the conversion unit 50 is connected to the output end of the optical computing network unit 30, and the conversion unit 50 is used to add and convert optical signals output by computing nodes belonging to the same group for electrical signals.

As shown in FIG. 2 , the number of third optical signals provided by the light source unit 11 is equal to the number of computing nodes in the optical computing network. In other possible implementation manners, the number of third optical signals provided by the light source unit 11 and the number of computing nodes in the optical computing network may also be unequal.

Referring to Fig. 2, the light source unit 11 provides multiple channels of third optical signals, such as λ11, λ12...λ1n, λ21, λ22...λ2n...λm1, λm2...λmn in the figure, a total of m×n channels, m and n are both positive integer. Correspondingly, there are also m×n computing nodes in the optical computing network unit 30, and the multi-channel first optical signals modulated by the multiple third optical signals are respectively transmitted to multiple computing nodes, and each computing node completes the input information and the multiplication operation of the convolution kernel coefficients (a ₁₁ ~a _mn ), after the multiplication operation is completed, the outputs of the calculation nodes belonging to a column are added to obtain the result of the convolution calculation (y1y2...yn).

Taking the image convolution operation as an example, the convolution kernel coefficients of each computing node in the optical computing network in Figure 2 constitute a convolution kernel coefficient matrix, and the input information carried by multiple first optical signals constitutes an input pixel amplitude matrix. The multiplication of information and convolution kernel coefficients, and the addition of the output results of the calculation nodes in the same column, thereby obtaining the result matrix y1y2...yn.

In the optical computing system provided in this application, multiple optical signals are provided, and the first step of modulation is performed on the multiple optical signals by the modulation unit, and the amplitude of the optical signal after the first modulation is used to represent the input information; After the first step of modulation, the optical signal after the first modulation is distributed to each computing node in the optical computing network through the optical distribution unit, and the second modulation of these optical signals is completed by these computing nodes. Secondary modulation is also amplitude modulation, and the amplitude of the optical signal after the second modulation represents the magnitude of the input information multiplied by a coefficient. The above two-step calculation completes the multiplication part in the convolution operation. Then, the conversion unit performs the addition of the multiplication calculation results of the same column, and converts it into an electrical signal output, so that the calculation results can be applied to the calculation of the subsequent electrical processing unit. In this solution, each computing node uses a separate first optical signal for calculation, that is, different optical signals participate in the calculation of each computing node, so that the power loss of the optical signal is small.

In the above optical computing system, at least one of the modes and wavelengths of any two third optical signals in the multiple third optical signals provided by the light source unit is different, that is, the present application provides different optical signals from two dimensions of wavelength and mode. signal, so that the optical computing network can support larger-scale convolution operations.

Fig. 3 is a schematic structural diagram of a light source unit provided by some embodiments of the present application. Referring to FIG. 3 , the light source unit 11 includes a multi-wavelength light source 111 and a plurality of mode converters 112 . The output terminal of the multi-wavelength light source 111 is connected to the input terminals of a plurality of mode converters 112 .

The multi-wavelength light source 111 is used to provide multiple channels of fourth optical signals with different wavelengths, such as λ1, λ2...λm in FIG. 3 . Usually, the wavelengths of the two optical signals can be separated by a certain value to ensure a certain degree of discrimination between the optical signals.

Multiple mode converters 112 correspond to multiple fourth optical signals with different wavelengths; the mode converters 112 are used to convert the corresponding fourth optical signals into multiple third optical signals with different modes, as shown in Figure 3. The optical signal λ1 is converted into multiple third optical signals λ11, λ12...λ1n, the fourth optical signal λ2 is converted into multiple third optical signals λ21, λ22...λ2n...λm1, and the fourth optical signal λm is converted into multiple The third optical signals λm1, λm2...λmn.

Here, the mode converter 112 converts the optical signal to obtain multiple third optical signals with different modes, that is, the converted third optical signal has multiple modes, where the multiple modes can be multiple transverse modes, such as transverse modes Die LP00, LP01, LP10, etc.

Exemplarily, the multi-wavelength light source 111 may be an optical frequency comb, a multi-wavelength laser array, and the like.

Fig. 4 is a schematic structural diagram of a light source unit provided by some embodiments of the present application. Referring to FIG. 4 , the light source unit 11 includes a multi-mode light source 113 and a plurality of mode converters 112 . The output of the multi-mode light source 113 is connected to the input of the plurality of mode converters 112 .

The multi-mode light source 113 is used to provide multiple fourth optical signals with different modes, such as λ1 , λ2 . . . λm in FIG. 4 .

Multiple mode converters 112 correspond to multiple fourth optical signals with different wavelengths; the mode converters 112 are used to convert a corresponding fourth optical signal into multiple third optical signals with different wavelengths, as shown in Figure 4. The four optical signals λ1 are converted into multiple third optical signals λ11, λ12...λ1n, the fourth optical signal λ2 is converted into multiple third optical signals λ21, λ22...λ2n...λm1, and the fourth optical signal λm is converted into multiple A third optical signal λm1, λm2...λmn.

In some possible implementations, the computing node 31 includes a modulator, such as a Mach Zehnder modulator (MZM). The modulator is an electro-optical conversion device, and by controlling the bias voltage of the optical modulation, it can adjust For the convolution kernel coefficient corresponding to the modulator, for example, the convolution kernel coefficient a ₁₁ =0.5 is required, and then the bias voltage of the calculation node is controlled to be 1V. The modulator modulates the amplitude of the input optical signal, so that the amplitude is attenuated or enhanced according to the ratio corresponding to the coefficient of the convolution kernel.

In some other possible implementation manners, the computing node 31 includes a phase-change material (or called a phase-sensitive material), and the phase-change material in this application is a photoinduced phase-change material. The phase change material is used to change the transmittance under the control of light, so that the amplitude of the light signal passing through the light signal changes according to the coefficient of the convolution kernel.

Exemplarily, the phase-change material is GST (Ge ₂ Sb ₂ Te ₅ ) material. GST material exhibits different states when the intensity of light is controlled differently. ), so that the transmittance of the material itself changes, and the modulation function of the input light is realized.

Exemplarily, the converting single light 50 includes: a plurality of photoelectric converters. A group of computing nodes 31 in the optical computing network unit 30 corresponds to a photoelectric converter, the input terminals of the photoelectric converter are respectively connected to the output terminals of a plurality of computing nodes 31 of the corresponding group, and the output terminals of the photoelectric converter output electrical signals.

Exemplarily, the photoelectric converter is a photodiode, such as an avalanche photodiode (avalanche photodiodes, APD), a photodiode (photodiodes, PD) and the like.

Fig. 5 is a partial structural diagram of an optical computing system provided by some embodiments of the present application. It mainly shows the detailed structure of the modulation unit 12, the optical computing network unit 30, and the optical distribution unit 20, see FIG. 5:

The modulation unit 12 includes a plurality of modulators 122, each modulator 122 is responsible for processing a third optical signal, and the modulator 122 is used to modulate the corresponding third optical signal with an electrical signal, so as to modulate the input information carried by the electrical signal to On the third optical signal, the first optical signal is obtained.

In the structure shown in Fig. 5, each modulator modulates one third optical signal. The modulator uses an electrical signal to modulate the optical signal, and the amplitude of the electrical signal corresponds to the bias voltage of the modulator. Therefore, the input information is represented by the magnitude of the electrical signal.

In the embodiment of the present application, the electrical signal is obtained by converting the digital signal through a digital-to-analog converter, each electrical signal corresponds to a digital-to-analog converter, and the output end of the digital-to-analog converter is connected to the modulator 122 .

Referring again to FIG. 5 , the multiple computing nodes 31 of the optical computing network unit 30 are arranged in an array, that is, divided into multiple rows and multiple columns, wherein one column of computing nodes 31 constitutes a group as mentioned above. Computing nodes 31 located in the same row can be connected through an optical transmission medium 33 . The optical transmission medium 33 may be an optical waveguide (such as an optical fiber), used to realize the transmission of optical signals. In the arrangement of computing nodes 31 shown in FIG. 5 , a row of computing nodes 31 may also be regarded as a group, and at this time, a row of computing nodes 31 may be connected through an optical transmission medium 33 .

In other possible implementation manners, the multiple computing nodes 31 may also be arranged in other manners, which will not be repeated here.

In some possible implementations, the optical distribution unit 20 includes a light turning element 21 and an optical converging element 22; the optical computing network unit 30 also includes at least one demultiplexer 32, and the number of demultiplexers 32 is usually multiple, Below are several examples. The input end of the light diverting element 21 is connected to the output ends of the plurality of modulators 122, the output end of the light diverting element 21 is opposite to the input end of the light converging element 22, the output end of the light converging element 22 is connected to the optical transmission of the computing node 31 The medium 33 is opposite. A plurality of demultiplexers 32 correspond to a plurality of computing nodes 31, and the demultiplexers 32 are connected to the optical waveguide, and a plurality of demultiplexers 32 corresponding to a row of computing nodes 31 are connected in series to the corresponding optical transmission on medium 33.

The light diverting element 21 is used to control the diversion of the multiple first optical signals output by the multiple modulators 122, that is, to divert the multiple first optical signals; the light converging element 22 is used to divert the diverted multiple first optical signals Converge into at least one second optical signal, transmit at least one second optical signal to the optical computing network unit; demultiplexer 32 is used to receive at least one second optical signal, and decouple at least one second optical signal.

Exemplarily, the at least one second optical signal is N second optical signals, and the N second optical signals are corresponding to the optical waveguides connected to N rows of computing nodes 31 transmitted to the optical computing network unit 30, and the optical computing network unit 30 calculates The number of rows of the node 31 is greater than or equal to N, where N is a positive integer. The demultiplexer 32 corresponds to the computing node 31, and the demultiplexer 32 decouples the second optical signal transmitted by the optical transmission medium 33 corresponding to the row where the computing node 31 is located, and outputs the decoupled second optical signal to the corresponding computing node 31. first light signal.

In this implementation, through the light diverting element and the light converging element, it is possible to control the position where the optical signal is finally transmitted to the optical network (such as the row or column position), so as to realize flexible scheduling and configuration of the optical signal. The demultiplexer can decompose the optical signal corresponding to the calculation needs of the calculation node from the coupled composite optical signal.

Exemplarily, the light redirecting element 21 includes liquid crystal on silicon, and the liquid crystal on silicon includes a plurality of liquid crystal on silicon units 210, and the plurality of liquid crystal on silicon units 210 correspond to multiple channels of first optical signals. By controlling the electric field where the liquid crystal unit is located, the deflection direction of the liquid crystal is changed, thereby changing the propagation path of the light passing through the liquid crystal unit. The propagating direction of each first optical signal is controllable by providing a liquid crystal unit for each first optical signal.

Exemplarily, the light converging element 22 includes a diffraction grating, and the diffraction grating converges light incident from a specific direction into a beam, and then transmits it to a corresponding row of the optical computing network unit 30 .

Fig. 6 is a partial structural diagram of an optical computing system provided by some embodiments of the present application. Compared with FIG. 5 , the optical computing system further includes a control device 150 . The control device 150 is connected with the light turning element 21 and is used for controlling the voltage output to the liquid crystal on silicon unit 210 in the light turning element 21 .

The control device 150 is used to control the liquid crystal on silicon unit to realize at least one of the following controls:

Controlling the composition of the first optical signal converged into a second optical signal in the optical converging element, that is, controlling which first optical signals will be converged together after arriving at the optical converging element;

Control the first optical signal sent to the first computing node in the optical computing network unit through the optical converging element, the first computing node is any computing node, that is, control the first optical signal to be finally emitted to the optical computing network unit which location. Since the position of each optical signal converged by the light converging element is fixed to the optical computing network unit, it means that the first optical signal is controlled to be converged at the light converging element after being steered by the liquid crystal on silicon unit.

The corresponding relationship between input information and computing nodes can be controlled through the control device, light steering element and light converging element, so as to control which computing node the input information is transmitted to, and then the flexible configuration of the corresponding relationship between input information and computing nodes can be realized, making the system When performing convolution calculations, the coefficients corresponding to each input information can be controlled to ensure the flexible configuration of convolution operations, such as using different convolution kernel coefficient matrices to perform convolution processing on images.

In some embodiments, in addition to controlling the light diverting element 21 , the control device 150 can also control the generation and modulation process of the light signal.

In the structure shown in FIG. 5, the demultiplexer 32 is connected to the optical transmission medium 33 arranged in the row direction. As shown in FIG. A multiplexer 32 , an input terminal and an output terminal of the demultiplexer 32 are connected to the optical transmission medium 33 , and the other output terminal of the demultiplexer 32 is connected to the computing node 31 . The working process of the demultiplexer 32 is described below with the demultiplexer of the first row in Fig. 3 as an example:

The optical signals λ11, λ12...λ1n are output from the optical converging element 22 to the optical transmission medium 33 where the first demultiplexer 32 is located from left to right (following will be described in this order), and the first demultiplexer 32 The components λ11 are decoupled and output to the corresponding computing node a ₁₁ , and the remaining parts λ12...λ1n are output to the second demultiplexer 32 in the same row. The second demultiplexer 32 decouples the component λ12 and outputs it to the corresponding computing node a ₁₂ , and outputs the remaining parts λ13...λ1n to the third demultiplexer 32 in the same row. By analogy, when reaching the last demultiplexer 32, only the component λ1n remains, and the last demultiplexer 32 outputs λ1n to the corresponding computing node a _1n .

In the structure shown in FIG. 6, the demultiplexer 32 is connected to the optical transmission medium 33 arranged in the row direction. As shown in FIG. Multiplexer 32 , an input terminal of demultiplexer 32 , and an output terminal of demultiplexer 32 is connected to computing node 31 . The difference between Fig. 6 and Fig. 5 mainly lies in that the inputs received by a plurality of demultiplexers 32 in a row in Fig. 5 are usually different, while in Fig. 6 the inputs received by a plurality of demultiplexers 32 in a row are The input is the same.

Exemplarily, the optical computing network unit 30 may be one processing unit, or may be composed of multiple processing units. For example, an optical computing network may be composed of multiple optical processing units (optical processing units, OPUs), and each optical processing unit is a processing unit. Each optical processing unit may contain multiple rows or columns of computing nodes. When performing light distribution, each optical processing unit may correspond to one beam of converged light, or one optical processing unit may correspond to multiple beams of converged light.

That is, the optical computing network unit 30 includes at least one optical processing unit, and transmits at least one second optical signal to the optical computing network unit 30, including:

Wherein, the processing capability of the optical processing unit is generally determined by the number of computing nodes it contains, and according to the processing capability of each optical processing unit, it is determined whether at least one second optical signal is sent to one or more optical processing units.

Fig. 7 is a partial structural diagram of an optical computing system provided by some embodiments of the present application. Referring to FIG. 7, the optical computing network unit 30 in the optical computing system includes a plurality of OPUs 300, and the multiple second optical signals converged by the optical converging elements are sent to the multiple OPUs 300, and each OPU 300 receives one second optical signal. light signal.

Fig. 8 is a partial structural diagram of an optical computing system provided by some embodiments of the present application. Referring to FIG. 8 , the optical computing network unit 30 in the optical computing system includes multiple OPUs, and the multiple second optical signals converged by the optical converging element are sent to one OPU, and the OPU receives the multiple second optical signals.

Figure 7 and Figure 8 show two possible implementations, the specific scheduling of the second optical signal can be controlled according to needs, for example, the control is converged into several second optical signals, and each second optical signal is transmitted to the optical computing network Where does the unit wait.

It should be noted that, in the structures shown in FIG. 7 and FIG. 8 , for simplicity, each OPU only shows one row of computing nodes 31 , but actually each OPU may include multiple rows of computing nodes 31 .

The optical computing system provided by other embodiments of the present disclosure includes an optical signal providing unit, an optical distribution unit, and an optical computing network unit. In these embodiments, the optical distribution unit is an optional component, for example, the optical signal output by the optical signal providing unit through an optical fiber, etc. It is transmitted to the optical computing network unit.

Wherein, the optical signal providing unit includes a light source unit and a modulation unit. The structure of the optical computing system provided by these embodiments will be described below in conjunction with the accompanying drawings. It should be noted that the first optical signal and the second optical signal used in the following embodiments are only used to represent the optical signals in these embodiments. In a change process, the first optical signal in the following embodiments and the first optical signal defined in the embodiments provided above may be different optical signals.

Fig. 9 is a partial structural diagram of an optical computing system provided by some embodiments of the present application. It mainly shows the detailed structure of the modulation unit 12, the optical distribution unit 20, and the optical computing network unit 30, see FIG. 9:

The modulation unit 12 includes a plurality of couplers 121 and a plurality of modulators 122, the input end of the coupler 121 is connected to the output end of the light source unit 11, the output end of the coupler 121 is connected to the input end of the modulator 122, and the input end of the modulator 122 The output end is connected to the optical transmission medium 33 connected to the computing node 31 through the optical distribution unit 20 . Wherein, the structure of the optical calculation unit 40 can be referred to FIG. 7 and FIG. 8 , which will not be repeated here.

Multiple couplers 121 are used to couple multiple channels of first optical signals into multiple channels of second optical signals; multiplexers 122 are used to adopt multiple channels of first electrical signals to respectively modulate multiple channels of second optical signals to obtain For the modulated second optical signal, each first electrical signal is obtained by coupling multiple second electrical signals with different frequencies.

In this implementation manner, since the first optical signal is coupled first, and then the coupled optical signal is modulated, the number of required modulators is reduced, which is beneficial to reducing the size and cost of the entire optical computing system.

In some possible implementations of the present application, the first electrical signal may be generated by an electrical signal generating circuit. The generation process of the first electrical signal will be described below in conjunction with FIG. 10 :

Fig. 10 is a schematic structural diagram of an electrical signal generating circuit provided by some embodiments of the present application. Referring to FIG. 10 , the electric signal generating circuit includes a digital circuit 91 , multiple multipliers 92 , multiple radio frequency sources 93 and multiple digital to analog converters (digital to analog converter, DAC) 94.

The digital circuit 91 provides multiple channels of digital frequency signals with the same frequency, and each channel of digital frequency signals has carried input information; multiple multipliers 92 are used to convert the frequencies of these multiple channels of digital frequency signals, thereby obtaining multiple channels of different frequencies. Digital frequency signal (second electrical signal); then utilize multiple radio frequency sources 93 to couple multiple digital frequency signals to obtain multiple coupled electrical signals; utilize multiple digital-to-analog converters 94 to convert the digital signal into an analog signal , to obtain multiple first electrical signals. In this way, the digital-to-analog conversion is performed after the electrical signal is coupled, which reduces the number of digital-to-analog converters and is beneficial to reduce the size and cost of the entire optical computing system.

When the above coupled electrical signal is used to modulate the optical signal, each modulated second optical signal is actually modulated by multiple second electrical signals at the same time, but when actually representing input information, each channel The second optical signal only needs to correspond to one second electrical signal, so before passing through the conversion unit, the optical signal needs to be filtered to filter out components corresponding to redundant electrical signals during modulation.

Referring again to FIG. 9 , the optical computing system may further include at least one filter 40 , usually there are multiple filters 40 , and the following descriptions will be made with multiple filters as an example. Multiple filters 40 correspond to multiple computing nodes 31 , the input terminals of the filters 40 are connected to the corresponding output terminals of the computing nodes 31 , and the output terminals of the filters 40 are connected to the conversion unit 50 . As shown in FIG. 9 , the output terminals of the filters 40 corresponding to the computing nodes 31 in the same column are connected to the same conversion unit 50 .

Each filter 40 is configured to filter an optical signal output by a computing node, and retain a corresponding component of one of the multiple second electrical signals with different frequencies. Here, filtering is actually filtering out components of the optical signal. When modulated by the modulation unit, the optical signal is modulated by multiple second electrical signals at the same time, so that the amplitude of the optical signal is the sum of the modulation results of the multiple second electrical signals. The component modulated by the electrical signal other than the signal is filtered out.

In some possible implementation manners, frequencies of electrical signals corresponding to any two computing nodes 31 are different. In some other possible implementation manners, the frequencies of the electrical signals corresponding to some computing nodes 31 may be the same, for example, the wavelengths of the second optical signals corresponding to these computing nodes 31 are different.

In this implementation manner, the electrical signal modulates the optical signal, and what is modulated is the amplitude of the optical signal. Multiple electrical signals modulate one optical signal at the same time, and the change in the amplitude of the optical signal is the sum of the amplitude changes of the multiple electrical signals. However, since the frequencies of the multiple electrical signals are different, it is still possible to obtain The component modulated by the corresponding electrical signal is filtered out from the optical signal. That is to say, in the embodiment corresponding to FIG. 9 , although multiple electrical signals are used for simultaneous modulation during modulation, the correspondence between the input information carried by the electrical signals used for modulation and the computing nodes still exists. Here, the filter It is also performed according to the corresponding relationship, that is, except for components obtained by modulation of electrical signals used to carry input information corresponding to computing nodes, the rest are filtered out. Let's illustrate with an example:

For example, m×n electrical signals are used, and the frequencies are f11, f12, and fmn respectively. During modulation, each first optical signal is modulated by the m×n electrical signals. Before adding, filter through the filter as follows:

For the calculation node a ₁₁ , other components other than the component modulated by f11 are filtered out; for the calculation node a ₂₁ , other components other than the component modulated by f21 are filtered out, and so on, the filtered optical signal is converted into electric signal.

For the optical computing system shown in FIG. 9 , in some possible implementations, the light distribution unit 20 may include a light diverting element and a light converging element.

In some other possible implementations, the light distribution unit 20 may only include light diverting elements. The light turning element is used to control the direction of the light signal output by the modulator.

Exemplarily, the optical computing network unit 30 includes at least one optical processing unit, and the optical steering element 21 is configured to divert at least one second optical signal, so that the at least one second optical signal is sent to the at least one optical processing unit.

Fig. 11 is a partial structural diagram of an optical computing system provided by some embodiments of the present application. It mainly shows the detailed structure of the modulation unit 12, the optical distribution unit 20, and the optical computing network unit 30, see FIG. 9:

The modulation unit 12 includes a coupler 121 and a modulator 122, the input end of the coupler 121 is connected to the output end of the light source unit 11, the output end of the coupler 121 is connected to the input end of the modulator 122, and the output end of the modulator 122 The optical distribution unit 20 is connected to the optical transmission medium 33 connected to the computing node 31 .

Wherein, the coupler 121 is used for coupling multiple first optical signals; the modulator 122 is used for using the first electrical signal to modulate the coupled multiple first optical signals, so that the multiple first optical signals are Modulate multiple channels of second optical signals, output the modulated optical signals, and obtain the first electrical signal by coupling multiple channels of second electrical signals with different frequencies.

In some possible implementations of the present application, the first electrical signal may be generated by an electrical signal generating circuit. The generation process of the first electrical signal will be described below in conjunction with FIG. 12 :

Fig. 12 is a schematic structural diagram of an electrical signal generating circuit provided by some embodiments of the present application. Referring to FIG. 10 , the electric signal generating circuit includes a digital circuit 91 , a plurality of multipliers 92 , a radio frequency source 93 and a digital-to-analog converter 94 .

The digital circuit 91 provides multiple channels of digital frequency signals with the same frequency, and each channel of digital frequency signals has carried input information; multiple multipliers 92 are used to convert the frequencies of these multiple channels of digital frequency signals, thereby obtaining multiple channels of different frequencies. digital frequency signal (second electrical signal); then utilize radio frequency source 93 to couple multiple digital frequency signals to obtain a coupled electrical signal; utilize a digital-to-analog converter 94 to convert the digital signal into an analog signal to obtain a first electrical signal an electrical signal. In this way, the digital-to-analog conversion is performed after the electrical signal is coupled, which reduces the number of digital-to-analog converters and is beneficial to reduce the size and cost of the entire optical computing system.

In the structure shown in FIG. 11 , the optical computing system also includes a filter 40 , and the description of FIG. 9 may be referred to.

In the structure shown in FIG. 11, the light distributing unit 20 may only include a light redirecting element.

In the above embodiments, Fig. 5 to Fig. 8 have introduced that the light distribution unit 20 can be composed of a light diverting element and a light converging element, and Fig. 9 to Fig. 12 have introduced that the modulation unit 12 can be composed of a coupler 121 and a modulator 122, It should be noted that, the above implementation manner of the optical distribution unit 20 and the implementation manner of the modulation unit 12 may be combined and applied in the same optical computing system.

Fig. 13 is a flow chart of an optical calculation method provided by some embodiments of the present application. Referring to Figure 13, the method includes:

131: Provide multiple channels of first optical signals.

The multiple channels of first optical signals are provided by the aforementioned optical signal providing unit 10 , and the detailed process of this step can refer to the aforementioned description about the optical signal providing unit 10 .

132: Steer the multiple channels of first optical signals.

Step 132 can be performed by the aforementioned light redirection element 21 , and for the detailed process of this step, reference can be made to the foregoing description about the light redirection element 21 .

133: Converge the diverted multiple first optical signals into at least one second optical signal, and transmit at least one second optical signal to the optical computing network unit.

Step 133 can be performed by the aforementioned light converging element 22 , and for the detailed process of this step, reference can be made to the foregoing description of the light converging element 22 .

134: Use the optical computing network unit to modulate at least one second optical signal.

Step 134 may be performed by the aforementioned optical computing network unit 30 , and for the detailed process of this step, reference may be made to the foregoing description about the optical computing network unit 30 .

Optionally, the method also includes:

It is worth noting that the optical calculation method provided in Figure 13 and the optical calculation system provided by any one of Figures 5 to 8 above adopt the same inventive concept, therefore, the details of the optical calculation method can refer to the previous Figures 5 to 8 correspond to the text descriptions and the drawings themselves.

Fig. 14 is a flow chart of an optical calculation method provided by some embodiments of the present application. Referring to Figure 14, the method includes:

141: Provide multiple channels of first optical signals.

The multiple channels of first light signals are provided by the light source unit 11 mentioned above, and the detailed process of this step can refer to the description about the light source unit 11 above.

142: Coupling multiple first optical signals into at least one second optical signal.

Step 142 can be provided by the aforementioned coupler 121 , and for the detailed process of this step, reference can be made to the aforementioned description about the coupler 121 .

143: At least one first electrical signal is used to modulate at least one second optical signal to obtain a modulated second optical signal, and each first electrical signal in at least one first electrical signal is obtained by adjusting multiple channels with different frequencies. obtained by coupling the second electrical signal.

Step 143 may be provided by the aforementioned modulator 122 , and for the detailed process of this step, reference may be made to the aforementioned description about the modulator 122 .

144: Use the optical computing network unit to modulate at least one second optical signal.

Step 144 may be performed by the aforementioned optical computing network unit 30 , and for the detailed process of this step, reference may be made to the foregoing description about the optical computing network unit 30 .

Optionally, the method also includes:

It is worth noting that the optical calculation method provided in Figure 14 uses the same inventive concept as the optical calculation system provided in any one of Figures 9 to 12 above. Therefore, the details of the optical calculation method can refer to the previous Figures 9 to 12 correspond to the text descriptions and the drawings themselves.

Fig. 15 shows a schematic structural diagram of a control device 150 provided by an exemplary embodiment of the present application. The control device 150 shown in FIG. 15 is used to execute the operations involved in the light calculation method shown in FIG. 13 or 14 above. The control device 150 can be realized by a general bus architecture.

As shown in FIG. 15 , the control device 150 includes at least one processor 151 , a memory 153 and at least one communication interface 154 .

The processor 151 is, for example, a general-purpose central processing unit (central processing unit, CPU), a digital signal processor (digital signal processor, DSP), a network processor (network processor, NP), a graphics processing unit (Graphics Processing Unit, GPU), A neural network processor (neural-network processing units, NPU), a data processing unit (Data Processing Unit, DPU), a microprocessor, or one or more integrated circuits for implementing the solution of this application. For example, the processor 151 includes an application-specific integrated circuit (application-specific integrated circuit, ASIC), a programmable logic device (programmable logic device, PLD) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. The PLD is, for example, a complex programmable logic device (complex programmable logic device, CPLD), a field-programmable gate array (field-programmable gate array, FPGA), a general array logic (generic array logic, GAL) or any combination thereof. It can realize or execute various logical blocks, modules and circuits described in conjunction with the disclosure of the embodiments of the present invention. The processor may also be a combination that implements computing functions, for example, a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.

Optionally, the control device 150 further includes a bus. The bus is used to transfer information between the various components of the control device 150 . The bus may be a peripheral component interconnect standard (PCI for short) bus or an extended industry standard architecture (EISA for short) bus or the like. The bus can be divided into address bus, data bus, control bus and so on. For ease of representation, only one thick line is used in FIG. 15 , but it does not mean that there is only one bus or one type of bus.

The memory 153 is, for example, a read-only memory (read-only memory, ROM) or other types of static storage devices that can store static information and instructions, or a random access memory (random access memory, RAM) or a memory that can store information and instructions. Other types of dynamic storage devices, such as electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc Storage (including Compact Disc, Laser Disc, Optical Disc, Digital Versatile Disc, Blu-ray Disc, etc.), magnetic disk storage medium, or other magnetic storage device, or is capable of carrying or storing desired program code in the form of instructions or data structures and capable of Any other medium accessed by a computer, but not limited to. The memory 153 exists independently, for example, and is connected to the processor 151 via a bus. The memory 153 can also be integrated with the processor 151 .

The communication interface 154 uses any device such as a transceiver for communicating with other devices or a communication network. The communication network can be Ethernet, radio access network (RAN) or wireless local area network (wireless local area networks, WLAN). The communication interface 154 may include a wired communication interface, and may also include a wireless communication interface. Specifically, the communication interface 154 can be an Ethernet (Ethernet) interface, a Fast Ethernet (Fast Ethernet, FE) interface, a Gigabit Ethernet (Gigabit Ethernet, GE) interface, an Asynchronous Transfer Mode (Asynchronous Transfer Mode, ATM) interface, a wireless local area network ( wireless local area networks, WLAN) interface, cellular network communication interface or a combination thereof. The Ethernet interface can be an optical interface, an electrical interface or a combination thereof. In the embodiment of the present application, the communication interface 154 may be used for the control device 150 to communicate with other devices.

In a specific implementation, as an embodiment, the processor 151 may include one or more CPUs, such as CPU0 and CPU1 shown in FIG. 15 . Each of these processors may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. A processor herein may refer to one or more devices, circuits, and/or processing cores for processing data (eg, computer program instructions).

In a specific implementation, as an embodiment, the control device 150 may include multiple processors, such as a processor 151 and a processor 155 as shown in FIG. 15 . Each of these processors can be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). A processor herein may refer to one or more devices, circuits, and/or processing cores for processing data such as computer program instructions.

In a specific implementation, as an example, the control device 150 may further include an output device and an input device. Output devices are in communication with processor 151 and can display information in a variety of ways. For example, the output device may be a liquid crystal display (liquid crystal display, LCD), a light emitting diode (light emitting diode, LED) display device, a cathode ray tube (cathode ray tube, CRT) display device, or a projector (projector). The input device is in communication with the processor 151 and can receive user input in a variety of ways. For example, the input device may be a mouse, a keyboard, a touch screen device, or a sensing device, among others.

In some embodiments, the memory 153 is used to store the program code 1510 for implementing the solution of the present application, and the processor 151 can execute the program code 1510 stored in the memory 153 . That is, the control device 150 can implement the optical computing method provided by the method embodiment through the processor 151 and the program code 1510 in the memory 153 . One or more software modules may be included in the program code 1510 . Optionally, the processor 151 itself may also store program codes or instructions for executing the solutions of the present application.

In a specific embodiment, the control device 150 in the embodiment of the present application may correspond to the control device in each method embodiment above, and the processor 151 in the control device 150 reads the instructions in the memory 153, so that the control shown in Fig. 15 The device 150 is capable of performing all or part of the operations performed by the control device.

Specifically, the processor 151 is configured to provide multiple channels of first optical signals; divert the multiple channels of first optical signals; converge the diverted multiple channels of first optical signals into at least one channel of second optical signals, and transmit The at least one second optical signal is sent to the optical computing network unit; the at least one second optical signal is modulated by the optical computing network unit.

Alternatively, providing multiple first optical signals; coupling the multiple first optical signals into at least one second optical signal; using at least one first electrical signal to modulate the at least one second optical signal to obtain modulation After the second optical signal, each first electrical signal in the at least one first electrical signal is obtained by coupling multiple second electrical signals with different frequencies; At least one second optical signal is modulated.

For other optional implementation manners, for the sake of brevity, details are not repeated here.

Wherein, each step of the optical calculation method shown in FIG. 13 or 14 is completed by an integrated logic circuit of hardware in the processor of the control device 150 or instructions in the form of software. The steps of the method disclosed in conjunction with the embodiments of the present application can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules in the processor. The software module can be located in a mature storage medium in the field such as random access memory, flash memory, read-only memory, programmable read-only memory or electrically erasable programmable memory, register. The storage medium is located in the memory, and the processor reads the information in the memory, and completes the steps of the above method in combination with its hardware. To avoid repetition, no detailed description is given here.

The embodiment of the present application also provides a chip, including: an input interface, an output interface, a processor, and a memory, the input interface, the output interface, the processor, and the memory are connected through an internal connection path, and the processor is used to execute the code in the memory , when the code is executed, the processor is configured to execute any one of the above optical calculation methods.

It should be understood that the above-mentioned processor may be a CPU, or other general-purpose processors, DSP, ASIC, FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the like. A general purpose processor may be a microprocessor or any conventional processor or the like. It should be noted that the processor may be a processor supporting the ARM architecture.

Further, in an optional embodiment, there are one or more processors and one or more memories. Optionally, the memory may be integrated with the processor, or the memory may be separated from the processor. The above-mentioned memory may include read-only memory and random-access memory, and provides instructions and data to the processor. Memory may also include non-volatile random access memory. For example, the memory may also store reference blocks and target blocks.

The memory can be either volatile memory or nonvolatile memory, or can include both volatile and nonvolatile memory. Wherein, the non-volatile memory may be ROM, PROM, EPROM, EEPROM or flash memory. Volatile memory can be RAM, which acts as external cache memory. By way of illustration and not limitation, many forms of RAM are available. For example, SRAM, DRAM, SDRAM, DDR SDRAM, ESDRAM, SLDRAM, and DR RAM.

In the embodiment of the present application, a computer-readable storage medium is also provided. The computer-readable storage medium stores computer instructions. When the computer instructions stored in the computer-readable storage medium are executed by a computer device, the computer device executes the above-mentioned The light calculation method provided.

In the embodiment of the present application, a computer program product including instructions is also provided, which, when running on a computer device, causes the computer device to execute the optical calculation method provided above.

In the above embodiments, all or part of them may be implemented by software, hardware, firmware or any combination thereof. When implemented using software, it may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, the processes or functions according to the present application will be generated in whole or in part. The computer can be a general purpose computer, a special purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website, computer, server or data center Transmission to another website site, computer, server, or data center by wired (eg, coaxial cable, optical fiber, DSL) or wireless (eg, infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that can be accessed by a computer, or a data storage device such as a server or a data center integrated with one or more available media. The available medium may be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, DVD), or a semiconductor medium (for example, a Solid State Disk).

Those of ordinary skill in the art can understand that all or part of the steps for implementing the above embodiments can be completed by hardware, and can also be completed by instructing related hardware through a program. The program can be stored in a computer-readable storage medium. The above-mentioned The storage medium mentioned may be a read-only memory, a magnetic disk or an optical disk, and the like.

The above descriptions are only optional embodiments of the present application, but the scope of protection of the present application is not limited thereto. Anyone skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present application. All should be covered within the scope of protection of this application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Unless otherwise defined, the technical terms or scientific terms used herein shall have the usual meanings understood by those having ordinary skill in the art to which the present disclosure belongs. "First", "second", "third" and similar words used in the specification and claims of this disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components . Likewise, words like "a" or "one" do not denote a limitation in quantity, but indicate that there is at least one. Words such as "comprises" or "comprising" and similar terms mean that the elements or items listed before "comprising" or "comprising" include the elements or items listed after "comprising" or "comprising" and their equivalents, and do not exclude other component or object.

The above is only an embodiment of the present application, and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included within the protection scope of the present application.

Claims

An optical computing system, characterized in that it comprises:

An optical signal providing unit (10), configured to provide multiple channels of first optical signals;

a light diverting element (21), configured to divert the multiple channels of first optical signals;

An optical converging element (22), configured to aggregate the diverted multiple first optical signals into at least one second optical signal, and transmit the at least one second optical signal to an optical computing network unit (30);

The optical computing network unit (30) is configured to receive the at least one second optical signal and modulate the at least one second optical signal.
The optical computing system according to claim 1, wherein the optical computing network unit (30) comprises:

At least one demultiplexer (32), the at least one demultiplexer (32) is used to receive the at least one second optical signal and decouple the at least one second optical signal.
The optical computing system according to claim 1, wherein the optical computing network unit (30) comprises at least one optical processing unit (300), and the transmitting of the at least one second optical signal to the optical computing network unit (30), including:

Transmitting the at least one second optical signal to the at least one optical processing unit (300) according to the processing capability of the at least one optical processing unit (300).
The optical computing system according to any one of claims 1 to 3, wherein the multiple first optical signals include a first optical signal, a second optical signal, and a third optical signal, and the first The modes of one optical signal and the second optical signal are different, and the wavelengths of the second optical signal and the third optical signal are different.
The optical computing system according to claim 4, wherein the optical signal providing unit (10) comprises:

A multi-wavelength light source (111), configured to provide multiple channels of fourth optical signals with different wavelengths;

A plurality of mode converters (112), configured to convert the received fourth optical signal into multiple optical signals with different modes, and the multiple optical signals with different modes are used to generate the multiple first light signal;

Alternatively, the optical signal providing unit (10) includes:

A multi-mode light source (113), configured to provide multiple fourth optical signals with different modes;

A plurality of wavelength converters (112), configured to convert the received fourth optical signal into multiple optical signals with different wavelengths, and the multiple optical signals with different wavelengths are used to generate the multiple first light signal.
The optical computing system according to any one of claims 1 to 4, wherein the optical signal providing unit (10) includes:

A light source unit (11), configured to provide multiple third optical signals;

A modulation unit (12), configured to receive the multiple third optical signals, and modulate the multiple third optical signals to obtain multiple first optical signals, the first optical signals carrying input information.
The optical computing system according to claim 6, wherein the modulation unit (12) comprises:

A plurality of couplers (121), configured to couple the multiple third optical signals into multiple fifth optical signals;

A multiplexer (122), configured to use multiple first electrical signals to modulate the multiple fifth optical signals respectively to obtain multiple first optical signals, each of the first electrical signals is obtained by pairing It is obtained by coupling multiple second electrical signals with different frequencies.
The optical computing system according to claim 7, wherein the optical computing network unit (30) comprises at least one computing node (31), and the optical computing system further comprises:

At least one filter (40), each of the filters (40) is used to filter the optical signal output by the at least one computing node (31), and retain the multiple second electrical signals with different frequencies The corresponding weight along the way.
The optical computing system according to any one of claims 6 to 8, wherein the light source unit (11) includes:

A multi-wavelength light source (111), configured to provide multiple channels of fourth optical signals with different wavelengths;

A plurality of mode converters (112), configured to convert one of the received fourth optical signals into multiple third optical signals with different modes;

Alternatively, the light source unit (11) includes:

A multi-mode light source (113), configured to provide multiple fourth optical signals with different modes;

A plurality of wavelength converters (112), configured to convert one path of the received fourth optical signal into multiple paths of third optical signals with different wavelengths.
The optical computing system according to any one of claims 1 to 9, characterized in that the light redirecting element (21) comprises a liquid crystal on silicon cell (210).
An optical computing system according to any one of claims 1 to 10, characterized in that the light concentrating element (22) comprises a diffraction grating.
The optical computing system according to any one of claims 1 to 11, wherein the optical computing system further comprises:

A conversion unit (50), configured to add and convert the optical signals output by the optical computing network unit (30) into electrical signals.
The optical calculation system according to any one of claims 1 to 12, wherein the optical calculation system is used for calculation in image processing, or the optical calculation system is used for calculation in digital signal processing, Or the optical computing system is used to perform calculations in a processor.
An optical computing system, characterized in that it comprises:

A light source unit (11), configured to provide multiple channels of first optical signals;

A coupler (121), configured to couple the multiple first optical signals into at least one second optical signal;

A modulator (122), configured to use at least one first electrical signal to modulate the at least one second optical signal to obtain the modulated second optical signal, and each of the at least one first electrical signal The first electrical signal is obtained by coupling multiple second electrical signals with different frequencies;

An optical computing network unit (30), configured to receive the at least one second optical signal and modulate the at least one second optical signal.
The optical computing system according to claim 14, wherein the optical computing network unit (30) comprises:

At least one demultiplexer (32), the at least one demultiplexer (32) is used to receive the at least one second optical signal and decouple the at least one second optical signal.
The optical computing system according to claim 14 or 15, wherein the optical computing network unit (30) includes at least one computing node (41), and the optical computing system further includes:

At least one filter (40), each of the filters (40) is used to filter the optical signal output by the at least one computing node (41), and retain the multiple second electrical signals with different frequencies The corresponding weight along the way.
The optical computing system according to any one of claims 14 to 16, wherein the multiple first optical signals include a first optical signal, a second optical signal, and a third optical signal, and the first The modes of one optical signal and the second optical signal are different, and the wavelengths of the second optical signal and the third optical signal are different.
The optical computing system according to claim 17, wherein the light source unit (11) comprises:

A multi-wavelength light source (111), configured to provide multiple channels of third optical signals with different wavelengths;

A plurality of mode converters (112), each of the mode converters (112) is used to receive one of the third optical signals among the multiple third optical signals with different wavelengths, and is used to convert the received one converting the third optical signal into a first optical signal with different multipath modes;

Alternatively, the light source unit (11) includes:

A multi-mode light source (113), configured to provide multiple third optical signals with different modes;

A plurality of mode converters (112), each of the mode converters (112) is used to receive one of the third optical signals of the multiple third optical signals with different modes, and is used to convert the received one of the third optical signals The third optical signal is converted into multiple channels of first optical signals with different wavelengths.
The optical computing system according to any one of claims 14 to 18, wherein the optical computing network unit (30) includes at least one optical processing unit (300), and the optical computing system further includes:

A light diverting element (21), configured to divert the at least one second optical signal, so that the at least one second optical signal is sent to the at least one optical processing unit (300).
The optical computing system according to claim 19, characterized in that the light redirecting element (21) comprises a liquid crystal on silicon cell (210).
The optical computing system according to any one of claims 14 to 20, wherein the optical computing system further comprises:

A conversion unit (50), configured to add and convert the optical signals output by the optical computing network unit (30) into electrical signals.
The optical calculation system according to any one of claims 14 to 21, wherein the optical calculation system is used for calculation in image processing, or the optical calculation system is used for calculation in digital signal processing, Or the optical computing system is used to perform calculations in a processor.
An optical calculation method, characterized in that the method comprises:

providing multiple first optical signals;

Steering the multiple channels of first optical signals;

Converging the diverted multiple first optical signals into at least one second optical signal, and transmitting the at least one second optical signal to the optical computing network unit;

The at least one second optical signal is modulated by the optical computing network unit.
The optical calculation method according to claim 23, further comprising:

Before modulating the at least one second optical signal, decoupling the at least one second optical signal.
The optical computing method according to claim 23, wherein the optical computing network unit includes at least one optical processing unit, and transmitting the at least one second optical signal to the optical computing network unit includes:

Transmitting the at least one second optical signal to the at least one optical processing unit according to the processing capability of the at least one optical processing unit.
The optical calculation method according to any one of claims 23 to 25, wherein the multiple first optical signals include a first optical signal, a second optical signal, and a third optical signal, and the first The modes of one optical signal and the second optical signal are different, and the wavelengths of the second optical signal and the third optical signal are different.
An optical calculation method, characterized in that the method comprises:

providing multiple first optical signals;

coupling the multiple first optical signals into at least one second optical signal;

Using at least one first electrical signal to modulate the at least one second optical signal to obtain the modulated second optical signal, and each first electrical signal in the at least one first electrical signal is obtained by pairing obtained by coupling multiple second electrical signals with different frequencies;

The at least one second optical signal is modulated by the optical computing network unit.
The optical calculation method according to claim 27, further comprising:

Before modulating the at least one second optical signal, decoupling the at least one second optical signal.
The optical computing method according to claim 27 or 28, wherein the optical computing network unit includes at least one computing node, and the method further includes:

Filtering the optical signal output by the at least one computing node, and retaining a component corresponding to one of the multiple second electrical signals with different frequencies.
The optical calculation method according to any one of claims 27 to 29, wherein the multiple first optical signals include a first optical signal, a second optical signal, and a third optical signal, and the first The modes of one optical signal and the second optical signal are different, and the wavelengths of the second optical signal and the third optical signal are different.
A control device, characterized by comprising a processor and a memory; program codes are stored in the memory, and when the program codes are executed by the processor, any one of claims 23 to 26 or claims The method described in any one of 27 to 30.