WO2021083348A1 - 一种光计算设备、光运算方法以及计算系统 - Google Patents

一种光计算设备、光运算方法以及计算系统 Download PDF

Info

Publication number
WO2021083348A1
WO2021083348A1 PCT/CN2020/125385 CN2020125385W WO2021083348A1 WO 2021083348 A1 WO2021083348 A1 WO 2021083348A1 CN 2020125385 W CN2020125385 W CN 2020125385W WO 2021083348 A1 WO2021083348 A1 WO 2021083348A1
Authority
WO
WIPO (PCT)
Prior art keywords
optical
optical signal
phase
waveguide
amplitude
Prior art date
Application number
PCT/CN2020/125385
Other languages
English (en)
French (fr)
Inventor
褚雅妍
董晓文
Original Assignee
华为技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from CN201911329810.0A external-priority patent/CN112783260A/zh
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Publication of WO2021083348A1 publication Critical patent/WO2021083348A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G02—OPTICS
    • G02B—OPTICAL ELEMENTS, SYSTEMS OR APPARATUS
    • G02B6/00—Light guides; Structural details of arrangements comprising light guides and other optical elements, e.g. couplings
    • G02B6/24—Coupling light guides
    • G02B6/26—Optical coupling means
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06E—OPTICAL COMPUTING DEVICES; COMPUTING DEVICES USING OTHER RADIATIONS WITH SIMILAR PROPERTIES
    • G06E3/00—Devices not provided for in group G06E1/00, e.g. for processing analogue or hybrid data
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L27/00—Modulated-carrier systems
    • H04L27/18—Phase-modulated carrier systems, i.e. using phase-shift keying
    • H04L27/22—Demodulator circuits; Receiver circuits
    • H04L27/227—Demodulator circuits; Receiver circuits using coherent demodulation

Definitions

  • This application relates to the field of information technology, and in particular to an optical computing device, an optical computing method, and a computing system.
  • Multiplication and addition is a basic operation. In the multiplication and addition operation, multiple multiplication results are added to obtain the final result.
  • the multiplier-adder is the calculation unit that realizes the multiplication and addition operation.
  • optical multiplier and adder is mainly based on optical multiplier and adder of optical fiber system.
  • This optical multiplier and adder includes an acousto-optic modulator array composed of a plurality of acousto-optic modulators connected by optical fibers and a plurality of detectors.
  • the signal can be modulated on the multiple optical signals through the acousto-optic modulator array to realize the multiplication operation.
  • multiple detectors are used to receive the optical signals modulated by the acousto-optic modulator array, and the received optical signals are converted into electrical signals, and multiple electrical signals obtained by multiple detectors are superimposed to realize the addition operation. In order to obtain the calculation result.
  • the optical multiplier and adder in the prior art uses the optical fiber system as the transmission carrier of the optical signal, the volume and power consumption of the optical multiplier and adder are relatively large. Moreover, since the addition part of this kind of optical multiplier and adder is completed by electricity, the computational efficiency is not high.
  • the present application provides an optical computing device, an optical computing method, and a computing system to provide a multiplier and adder with a small volume and high scalability.
  • the present application provides an optical computing device, which includes an optical splitter, N-1 first waveguides, and a beam combiner.
  • the optical splitter is respectively connected to the input ends of the N-1 first waveguides, and the beam combiner is respectively connected to the output ends of the N-1 first waveguides.
  • the optical splitter can divide the received continuous light into N optical signals, and transmit the N optical signals to N-1 first waveguides and one second waveguide, respectively, wherein each first waveguide receives one optical signal.
  • Signal, the second waveguide receives an optical signal.
  • the N-1 first waveguides have an amplitude modulation function, and can adjust the amplitude of the received N-1 optical signal according to the set amplitude parameter, and output the N-1 intermediate optical signal.
  • each first waveguide is provided with at least two amplitude parameters.
  • the at least two amplitude parameters are used to indicate the same amount of data of the multiplication operation to be performed, and the intermediate optical signal output by the first waveguide may indicate the product of the data indicated by the at least two amplitude parameters.
  • the beam combiner receives N-1 intermediate optical signals, and combines the N-1 intermediate optical signals into a first optical signal.
  • the amplitude of the optical signal is adjusted through the first waveguide, so that the optical computing device can implement multiplication.
  • the N-1 intermediate optical signals output by the N-1 first waveguides pass through the combiner, which can combine the N-1 intermediate optical signals, so that the optical computing device can realize the addition operation and ensure The optical computing device can realize multiplication and addition operations. Because the optical computing device uses waveguides to implement multiplication and addition operations, it can effectively reduce the volume and power consumption of the optical computing device, improve the scalability of the optical computing device, and further enable the optical computing device to calculate more data. Big.
  • the optical computing device may further include a second waveguide and a beam combiner.
  • the optical splitter can be connected to the input end of the second waveguide, and the beam combiner can be connected to the output end of the second waveguide.
  • the optical splitter divides the received continuous light into N optical signals, and transmits the N optical signals to N-1 first waveguides and a second waveguide respectively, and N-1 first waveguides receive N-1 optical signals.
  • the second waveguide receives one optical signal except the N-1 optical signal.
  • the amplitude of the N-1 optical signal can be adjusted according to the set amplitude parameter, and the phase of the optical signal output by the first waveguide can be adjusted according to the set phase parameter ,
  • the phase parameter may indicate the positive or negative value of the sum of the products of the data indicated by the intermediate optical signal output by the first waveguide.
  • the second waveguide After receiving the optical signal, the second waveguide can adjust the phase of the optical signal to output a second optical signal.
  • the phase of the second optical signal reaching the combiner is the first phase.
  • the beam combiner can combine N-1 channels output by the first waveguide and the second optical signal output by the second waveguide into a third optical signal.
  • the amplitude of the optical signal is adjusted through the first waveguide, so that the optical computing device can implement multiplication.
  • the first waveguide also has the function of adjusting the phase of the optical signal, it can change the phase of the optical signal and add a sign to the result of each multiplication operation.
  • the optical signal output by the first waveguide and the second waveguide can be combined by the beam combiner, so that the optical computing device can implement a signed addition operation.
  • the second waveguide can adjust the phase of the optical signal and change the light intensity of the optical signal output by the final combiner, the second waveguide can determine the amplitude of the optical signal after the intermediate optical signal output by the first waveguide is combined (that is, The result of the multiplication and addition operation) to ensure that the optical computing device can realize the signed multiplication and addition operation.
  • the optical computing device uses waveguides to implement multiplication and addition operations, it can effectively reduce the volume and power consumption of the optical computing device, ensure the scalability of the optical computing device, and make the calculation that the optical computing device can implement The amount increases.
  • the second waveguide can also adjust the phase of the received Nth optical signal again, and adjust the received optical signal to the fourth optical signal, and the fourth optical signal reaches the phase of the beam combiner. Is the second phase;
  • the beam combiner then combines the N-1 first waveguides to output N-1 intermediate optical signals and the fourth optical signal into a fifth optical signal.
  • the second waveguide since the second waveguide can adjust the phase of the optical signal multiple times, it can output multiple different optical signals (such as the second optical signal and the fourth optical signal), so that the optical signal output by the beam combiner The light intensity of is changed, so that the optical computing device can implement signed multiplication and addition operations.
  • the optical computing device may also include a detector and a processing circuit.
  • the detector can detect the optical signals output by the beam combiner (such as the first optical signal, the third optical signal, and the fifth optical signal). ) Of the light intensity.
  • the processing circuit can obtain the light intensity of the optical signal output by the beam combiner from the detector, and determine the sum of the product of the data indicated by the N-1 intermediate light signal according to the light intensity.
  • the processing circuit can directly determine the sum of the products of the data indicated by the N-1 intermediate optical signals according to the light intensity of the first optical signal, that is, the calculation result of the unsigned multiplication and addition operation.
  • the processing circuit can also directly determine the sum of the product of the data indicated by the N-1 intermediate optical signal according to the light intensity of the third optical signal and the light intensity of the fifth optical signal, that is, the calculation result of the signed multiplication and addition operation.
  • the processing circuit can conveniently determine the calculation result of the multiplication and addition operation through the light intensity of the optical signal output by the beam combiner, consumes less resources, and the determination method is more convenient, which can effectively improve the optical computing equipment. Realize the efficiency of multiplication and addition operations.
  • each first waveguide includes at least two amplitude modulation units for implementing a group of multiplication operations in the multiplication and addition operations.
  • Any amplitude modulation unit can adjust the amplitude of the received optical signal according to an amplitude parameter, and the amplitude parameter can indicate a piece of data of a multiplication operation that needs to be implemented in a multiplication and addition operation that needs to be implemented.
  • the first waveguide can implement multiplication by adjusting the amplitude, so that the implementation of the multiplication is simpler and more efficient.
  • each first waveguide includes two amplitude modulation units, namely a first amplitude modulation unit and a second amplitude modulation unit, and the input end of the second amplitude modulation unit is connected to the output end of the first amplitude modulation unit .
  • the first amplitude modulation unit may adjust the amplitude of the optical signal according to the first amplitude parameter.
  • the second amplitude modulation unit may adjust the amplitude of the optical signal adjusted by the first amplitude modulation unit according to the second amplitude parameter.
  • the first waveguide further includes a first phase modulation unit, and one first phase modulation unit and at least two amplitude modulation units are connected in series.
  • the connection sequence of the first phase modulation unit and the at least two amplitude modulation units is not limited here.
  • Each unit in the first waveguide adjusts the optical signal output by the previous unit.
  • the first phase modulation unit can adjust the phase of the received optical signal according to the phase parameter, so that the phase difference between the adjusted optical signal (that is, the intermediate optical signal) and the phase when the reference light reaches the beam combiner is 0 or ⁇ , Where the reference light is one of the N optical signals.
  • the first waveguide can implement multiplication by adjusting the amplitude, and adding a sign to the result of the multiplication by adjusting the phase, so that the realization of the multiplication is simpler and more efficient.
  • the first phase modulation unit may be calibrated in advance to determine the first phase parameter that can adjust the phase difference when the optical signal and the reference light reach the beam combiner to 0 or ⁇ . After that, in actual multiplication and addition operations, the phase difference between the intermediate light signal and the reference light when reaching the beam combiner can be adjusted to 0 or ⁇ directly according to the first phase parameter.
  • the first phase modulation unit can more accurately control the phase difference when the intermediate optical signal and the reference light reach the beam combiner.
  • the second waveguide includes a second phase modulation unit.
  • the second phase modulation unit can adjust the phase of the received optical signal so that the phase of the adjusted optical signal when it reaches the combiner is the first phase or the second phase, and the reference light is one of the N optical signals. .
  • first phase and the second phase are not limited here.
  • the first phase and the second phase must meet one of the following conditions:
  • Condition 1 The difference between the first phase and the phase when the reference light reaches the beam combiner is 0, and the difference between the second phase and the phase when the reference light reaches the beam combiner is ⁇ .
  • Condition 2 The difference between the first phase and the phase when the reference light reaches the beam combiner is ⁇ , and the difference between the second phase and the phase when the reference light reaches the beam combiner is 0.
  • the second waveguide only needs to adjust the phase of the optical signal, and change the phase difference between the output optical signal and the reference light when it reaches the combiner, so as to change the light intensity of the optical signal output by the combiner. . That is to say, the different phase difference between the optical signal and the reference light when they reach the combiner makes the combiner output different optical signals, and the light intensity of the optical signal output by the combiner under different phase differences can be conveniently and directly Determine the positive and negative value of the sum of the indicated data of the optical signals after the N-1 intermediate optical signals are combined.
  • the second phase modulation unit may be calibrated in advance, and it has been determined that the phase difference when the optical signal and the reference light reach the beam combiner can be adjusted to a second phase parameter of 0 or ⁇ . After that, in actual multiplication and addition operations, the phase difference when the optical signal and the reference light reach the beam combiner can be adjusted to 0 or ⁇ directly according to the second phase parameter.
  • the first phase modulation unit can more accurately control the phase difference when the optical signal and the reference light reach the beam combiner.
  • the optical computing device further includes a laser, and the laser is connected to the optical splitter to generate continuous light, and the laser outputs the continuous light to the optical splitter.
  • the laser is a single longitudinal mode semiconductor laser.
  • a laser can be used to generate continuous light of a single frequency, which is convenient for subsequent processing.
  • the amplitude modulation unit is EAM, SOA or VOA.
  • the beam combiner is an MMI or cascade Y branch; the splitter is an MMI or cascade Y branch.
  • the present application provides an optical operation method, and the beneficial effects can be referred to the related description of the first aspect, which will not be repeated here.
  • This method is executed by an optical computing device.
  • the optical splitter divides the received continuous light into N optical signals; N-1 first waveguides receive N-1 optical signals among the N optical signals, and adjust the optical signals according to the set amplitude parameters.
  • N-1 optical signal amplitude, output N-1 intermediate optical signal where each first waveguide receives one optical signal, each first waveguide is set with at least two amplitude parameters, at least two amplitude parameters can be Indicating at least two pieces of data to be multiplied, so that the intermediate optical signal output by each first waveguide can indicate the product of at least two pieces of data corresponding to the at least two amplitude parameters set in the first waveguide;
  • the beam combiner receives the N-1 intermediate optical signals, and combines the N-1 intermediate optical signals into a first optical signal.
  • the first optical signal can indicate the sum of the products of the data indicated by the N-1 intermediate optical signals.
  • each of the N-1 first waveguides can also adjust the phase of the optical signal output by the first waveguide according to the set phase parameter; the phase of the optical signal output by the first waveguide can be Indicates the positive or negative value of the product indicated by the intermediate light signal.
  • the second waveguide receives the Nth optical signal among the N optical signals, modulates the phase of the Nth optical signal, and can adjust the received Nth optical signal to the second optical signal, and the second optical signal reaches the beam combiner.
  • the phase is the first phase
  • the beam combiner can combine the N-1 intermediate optical signals and the second optical signal into a third optical signal.
  • the second waveguide can also adjust the phase of the Nth optical signal again, adjusting the received Nth optical signal to the fourth optical signal, and the fourth optical signal reaches the phase of the combiner. Is the second phase;
  • the beam combiner combines the N-1 intermediate optical signal and the fourth optical signal into a fifth optical signal.
  • the detector can detect the light intensity of the first optical signal; the processing circuit can obtain the product of the data indicated by the N-1 intermediate optical signal according to the light intensity of the first optical signal output by the detector. with.
  • the detector can also detect the light intensity of the third light signal and the light intensity of the fifth light signal; the processing circuit obtains the middle of the N-1 path according to the light intensity of the third light signal output by the detector and the light intensity of the fifth light signal The sum of the products of the data indicated by the optical signal.
  • the difference between the first phase and the phase when the reference light reaches the combiner is 0, the difference between the second phase and the phase when the reference light reaches the combiner is ⁇ , and the reference light is N
  • the difference between the first phase and the phase when the reference light reaches the combiner is ⁇
  • the difference between the second phase and the phase when the reference light reaches the combiner is 0, and the reference light is one of the N optical signals. signal.
  • the present application provides a computing system.
  • the computing system may include a processor and an optical computing device as described in the first aspect or any one of the possible implementation manners of the first aspect.
  • the processor can send data to be multiplied and added to the optical computing device.
  • the data to be multiplied and added includes N-1 sets of data, and each group of data includes at least two data to be multiplied in the multiply and add operation.
  • the amplitude parameter of the first waveguide can be set according to a set of data, and optionally, the phase parameter can also be set.
  • the optical calculation device performs optical calculation, outputs the calculation result of the multiplication and addition operation, and feeds back the calculation result to the processing.
  • Figure 1 is a schematic structural diagram of an optical computing device provided by this application.
  • Figure 2 is a schematic structural diagram of another optical computing device provided by this application.
  • FIG. 3 is a schematic structural diagram of a first waveguide provided by this application.
  • FIG. 4 is a schematic structural diagram of a second waveguide provided by this application.
  • FIG. 5 is a schematic diagram of an optical operation method provided by this application.
  • FIG. 6 is a schematic diagram of an optical operation method provided by this application.
  • Fig. 7 is a schematic diagram of a computing system provided by this application.
  • This application provides an optical computing device and an optical computing method to provide an optical multiplier and adder with a small volume and high scalability.
  • FIG. 1 shows an optical computing device provided by an embodiment of the application.
  • the optical computing device includes an optical splitter 110, N-1 first waveguides 120, and a beam combiner 130.
  • the optical splitter 110 is respectively connected to the N-1 first waveguides 120.
  • the optical splitter 110 can divide the received continuous light into N optical signals, and then transmit the N-1 optical signals of the N optical signals to N-1 first waveguides 120; among them, one first waveguide 120 receives one optical signal among N-1 optical signals.
  • the N-1 first waveguides 120 can adjust the amplitude of the received N-1 optical signals according to the amplitude parameter, and output N-1 intermediate optical signals.
  • the first waveguide 120 receives one of the N-1 optical signals, and adjusts the optical signal according to the set at least two amplitude parameters
  • the amplitude of the output is an intermediate optical signal.
  • the at least two amplitude parameters can be used to indicate at least two pieces of data for multiplication, and one amplitude parameter has at least one piece of data. In this way, the output of an intermediate optical signal can indicate the product of the two pieces of data.
  • the beam combiner is respectively connected to N-1 first waveguides, receives N-1 intermediate optical signals output by the N-1 first waveguides 120, and combines the N-1 intermediate optical signals into the first optical signal.
  • the optical signal is used to indicate the sum of the products of the data indicated by the N-1 intermediate optical signals.
  • the first waveguide 120 has an amplitude modulation function, which can adjust the amplitude of the received optical signal.
  • the first waveguide 120 may also have a phase modulation function, and the N-1 first waveguides 120 may adjust the phase of the N-1 optical signals so that the first waveguide 120 outputs the same phase of the intermediate optical signals.
  • the multiplication and addition operation to be realized by the optical computing device is: (Among them, the signs of the N-1 X i W i are the same, and they can all be positive or negative.)
  • the optical computing device uses N-1 first waveguides for 120 pairs of N-1 optical The amplitude of the signal is adjusted to realize the N-1 multiplication operations in the multiplication-addition operation.
  • Each first waveguide 120 adjusts the amplitude of one optical signal in the N-1 optical signal according to the amplitude parameter.
  • the parameter indicates X i W i , so that the amplitude of the intermediate light signal can indicate one of the N-1 multiplication operations X i W i .
  • the first optical signal is output, and the amplitude of the first optical signal is Indicates the sum of N-1 products.
  • the optical computing device can implement unsigned multiplication, and then the beam combiner 140 can output the intermediate optical signal to the first waveguide 120
  • Combining beams enables the optical computing device to implement addition operations, and ensures that the optical computing device can implement unsigned multiplication and addition operations (the sign of the result of each multiplication operation is the same). Since the first waveguide 120 is used in the optical computing device, the volume occupied by the optical computing device can be effectively reduced, the scalability of the optical computing device is improved, and the amount of data that the optical computing device can calculate is larger.
  • the optical computing device as shown in Figure 1 can realize unsigned multiplication and addition operations (each multiplication result can be positive or negative). In order to realize signed multiplication and addition operations, as shown in Figure 1 The optical computing device is improved.
  • another optical computing device provided by this embodiment of the application includes an optical splitter 110, N-1 first waveguides 120, and a second waveguide 140. And the beam combiner 130.
  • the optical splitter 110 is respectively connected to N-1 first waveguides 120 and one second waveguide 140.
  • the optical splitter 110 can divide the received continuous light into N optical signals, and then transmit the N optical signals to N- One first waveguide 120 and one second waveguide 140; among them, one first waveguide 120 receives one of the N optical signals, N-1 first waveguides 120 receive N-1 optical signals, and the second waveguide 120 140 receives one optical signal except for the N-1 optical signal among the N optical signals, that is, the Nth optical signal.
  • any one of the N-1 first waveguides 120 receives one of the N optical signals, in addition to adjusting the amplitude of the optical signal according to the amplitude parameter (the first waveguide 120 adjusts the optical signal
  • the first waveguide 120 adjusts the optical signal
  • the first waveguide 120 outputs an optical signal whose phase and amplitude are adjusted.
  • the optical signal output by the first waveguide 120 may also be referred to as an intermediate optical signal.
  • each first waveguide 120 is provided with a phase parameter indicating the positive and negative value of the product of the at least two data indicated by the intermediate optical signal.
  • the multiplication and addition operation to be realized by the optical computing device is:
  • a phase parameter set by the first waveguide 120 indicates the positive and negative values of X i W i.
  • the second waveguide 140 After the second waveguide 140 receives the Nth optical signal, it can adjust the phase of the received optical signal, and output the phase-adjusted optical signal.
  • the phase of the optical signal reaching the combiner can be the first phase or the first phase. Two-phase. In other words, the second waveguide 140 is different from the first waveguide 120, and only the phase of the received optical signal can be adjusted.
  • the beam combiner 130 connects N-1 first waveguides 120 and a second waveguide 140, respectively, and receives N-1 intermediate optical signals output by the N-1 first waveguides 120 and optical signals output by the second waveguide 140.
  • the N-1 intermediate optical signals and the optical signal output by the second waveguide 140 are combined into one optical signal.
  • the optical computing device may further include a detector 150, which can detect the light intensity of the optical signal output by the beam combiner.
  • the detector 150 may be a photovoltaic type detector or a photoconductive detector.
  • the multiplication and addition operation to be realized by the optical computing device is: For example (the sign of X i W i is not limited here, it can be positive or negative).
  • the optical computing device uses N-1 first waveguides 120 to determine the amplitude and phase of the N-1 optical signal. Adjust to realize the N-1 multiplication operations in the multiplication-addition operation, where each first waveguide 120 adjusts the amplitude of one optical signal in the N-1 optical signals to realize one of the N-1 multiplication operations
  • the multiplication operation X i W i that is, the amplitude of the optical signal after the amplitude adjustment is the result of the multiplication operation.
  • Each first waveguide adjusts the phase of one optical signal in the N-1 optical signal, and adds a sign (such as a positive sign or a negative sign) before the multiplication operation X i W i.
  • a sign such as a positive sign or a negative sign
  • the optical signal A is output, and the amplitude of the optical signal A can indicate Since the light intensity of the optical signal detected by the detector 150 is the square of the amplitude of the optical signal, the sign of the amplitude of the optical signal cannot be distinguished. For this reason, a second waveguide 140 is added to the optical computing device.
  • the optical computing device can adjust the phase of one optical signal through the second waveguide 140, and then change the N-1 intermediate optical signal and the second waveguide 140 after being combined in the beam combiner 130.
  • the amplitude of the optical signal of one path changes the optical intensity of the optical signal, and the sign of the optical signal A is determined by the change of the optical intensity (for a specific method of determining the sign, please refer to the description below).
  • the optical computing device can realize multiplication; and because the first waveguide 120 also has the function of adjusting the phase of the optical signal, it can change the phase of the optical signal. , Add a sign to the result of each multiplication operation; after that, the optical signal output by the first waveguide 120 and the second waveguide 140 can be combined through the beam combiner 130, so that the optical computing device can implement a signed addition operation. Since the second waveguide 140 can adjust the phase of the optical signal and can change the light intensity of the optical signal output by the final combiner, the second waveguide 140 can determine the amplitude of the optical signal after the optical signal output by the first waveguide 120 is combined.
  • the optical computing device can implement signed multiplication and addition operations.
  • the volume occupied by the optical computing device can be effectively reduced, the scalability of the optical computing device is improved, and the calculation capability of the optical computing device is improved. The amount of data.
  • the second waveguide 140 can perform two phase adjustment operations on the received N-th optical signal respectively.
  • the beam combiner can combine two different optical signals.
  • the second waveguide 140 receives the Nth optical signal among the N optical signals, and adjusts the received Nth optical signal into a second optical signal.
  • the phase of the second optical signal reaching the beam combiner is The first phase.
  • the combiner 130 combines the N-1 intermediate optical signals output by the N-1 first waveguides 120 and the second optical signals output by the second waveguide into a third optical signal.
  • the second waveguide 140 receives the Nth optical signal among the N optical signals, and adjusts the received Nth optical signal into a fourth optical signal.
  • the phase of the third optical signal reaching the beam combiner is The second phase.
  • the combiner 130 combines the N-1 intermediate optical signals output by the N-1 first waveguides 120 and the fourth optical signal output by the second waveguide into a fifth optical signal.
  • the optical computing device may also include a detector 150. After the detector 150 detects the light intensities of the third light signal and the fifth light signal, the detector 150 determines the light intensities of the third light signal and the fifth light signal through the detector, and determines the light intensity of the third light signal and the fifth light signal according to the light intensity of the third light signal and the fifth light signal.
  • the light intensity obtains the sum of the products of the data indicated by the N-1 intermediate optical signals (for a specific determination method, please refer to the description below).
  • optical computing device shown in Figure 2 can also implement the functions that the optical computing device shown in Figure 1 can achieve, and control the second The waveguide does not receive the N-th optical signal (for example, the amplitude of the N-th optical signal is adjusted to 0).
  • the optical splitter 110 has a light splitting function.
  • the embodiment of the present application does not limit the specific type of the optical splitter 110.
  • the optical splitter 110 may be a multimode interference coupler (MMI) or a multimode interference coupler (MMI).
  • Link Y branch may be a multimode interference coupler (MMI) or a multimode interference coupler (MMI).
  • the light received by the optical splitter 110 is continuous light, and the so-called continuous light means that there is no discontinuity in the optical signal.
  • the embodiment of the present application does not limit the source of the continuous light, and any device capable of generating continuous light is applicable to the embodiment of the present application.
  • the optical computing device may further include a laser 170, and the laser 170 may generate the continuous light.
  • the laser 170 may be a single longitudinal mode semiconductor laser, which generates continuous light with a single frequency.
  • Single longitudinal mode semiconductor lasers include but are not limited to: distributed feedback (DFB) lasers, distributed Bragg reflector (DBR) lasers, or micro-ring lasers.
  • the N optical signals generated by the optical splitter 110 after splitting they can be the same N optical signals, for example, the phase and amplitude of the N optical signals are the same; it can also be different N optical signals, such as the amplitude of the N optical signals Same but different phase.
  • the optical splitter 110 when the optical splitter 110 performs light splitting, due to the principle of optical splitting by the optical splitter 110 or environmental factors, the amplitude of the N optical signals will be different. In order to ensure that the amplitudes of the optical signals remain consistent, the first waveguide 120 can be used. The amplitude of the N-1 optical signal is compensated. Similarly, due to the principle of light splitting by the optical splitter 110 or environmental factors, the phases of the N optical signals may also be different. In order to ensure that the phases of the N optical signals remain consistent, the first waveguide 120 and the second waveguide 140 can be used to pair the N optical signals. The phase of the optical signal is compensated.
  • the first waveguide 120 has both an amplitude modulation function and a phase modulation function.
  • the first waveguide 120 includes a first phase modulation unit 121 and at least two amplitude modulation units.
  • One first phase modulation unit 121 and at least two amplitude modulation units are connected in series.
  • the embodiment of the present application does not limit the sequence of a first phase modulation unit and at least two amplitude modulation units.
  • the first waveguide 120 includes two amplitude modulation units and One first phase modulation unit 121 and two amplitude modulation units are located before the first phase modulation unit as an example for description. To facilitate the distinction between the two amplitude modulation units, they are represented by an amplitude modulation unit 122A and an amplitude modulation unit 122B, respectively.
  • One of the N optical signals generated by the optical splitter 110 enters the first waveguide 120, first passes through the amplitude modulation unit 122A, then passes through the amplitude modulation unit 122B, and finally passes through the first phase modulation unit 121.
  • the amplitude modulation unit 122A and the amplitude modulation unit 122B can respectively adjust the amplitude of the received optical signal according to the amplitude parameter.
  • the optical signal successively passes through the amplitude modulation unit 122A and the amplitude modulation unit 122B, which can realize a set of multiplication operations in the multiplication and addition operations.
  • each first waveguide 120 is used to implement a set of multiplication operations X i W i in the multiplication and addition operations.
  • Amplitude parameter may indicate the X i and W i
  • the amplitude modulation section 120 a first optical waveguide 122A may adjust the amplitude of the signal received from the amplitude parameter indicative of X i, then, according to a further indication of the amplitude modulation section 122B W i
  • the amplitude parameter adjustment amplitude modulation unit 122A adjusts the amplitude of the optical signal after the amplitude.
  • One amplitude modulation unit 122A in the first waveguide 120 can also adjust the amplitude of the received optical signal according to the amplitude parameter indicating W i .
  • the other amplitude modulation unit 122B adjusts the amplitude modulation unit 122A according to the amplitude parameter indicating X i.
  • multiplication first waveguide 120 can have a positive or negative by adjusting the phase of the light signal.
  • the first waveguide 120 includes two amplitude modulation units as an example.
  • the first waveguide 120 may also include three or even more amplitude modulation units; the number of amplitude modulation units is calculated by the light.
  • the number of data in a group of multiplication operations to be implemented by the device is related to the number of data in a group of multiplication operations.
  • the first waveguide 120 may include three amplitude modulation units. That is, the number of amplitude modulation units can be equal to the number of operation parameters.
  • the amplitude modulation unit is not limited in the embodiments of the present application, and any device that can adjust the amplitude of the optical signal can be used as the amplitude modulation unit.
  • the amplitude modulation unit may be an electro-absorption modulator (EAM), a semiconductor optical amplifier (SOA), or a variable optical attenuator (VOA).
  • EAM electro-absorption modulator
  • SOA semiconductor optical amplifier
  • VOA variable optical attenuator
  • the absorption coefficient of the optical signal is adjusted by controlling the reverse bias voltage applied to the EAM or VOA to achieve the attenuation of the optical signal and adjust the amplitude of the optical signal; wherein, the reverse bias The set voltage is determined by the amplitude parameter.
  • the amplification factor of the optical signal is adjusted by controlling the forward bias current applied to the SOA to realize the attenuation or amplification of the optical signal and adjust the amplitude of the optical signal.
  • the forward-biased current is the amplitude parameter; when the forward-biased current is lower than the threshold, the optical signal can be attenuated and the amplitude of the optical signal can be reduced; when the forward-biased current is higher than the threshold At this time, the optical signal can be amplified and the amplitude of the optical signal can be increased.
  • the amplitude of the optical signal is changed, and the amplitude-adjusted optical signal passes through the first phase modulation unit 121, and the first phase modulation unit 121 adjusts the amplitude modulation unit.
  • the phase of the optical signal output by 122B that is, the phase of the optical signal output by the first waveguide 120 can be adjusted, and the sign of the result of the multiplication operation can be adjusted (for example, adjusted to positive or negative).
  • one optical signal among the N optical signals can be used as the reference light, and the first phase modulation unit 121 can adjust the phase of the first waveguide 120 to receive one optical signal, so that the reference light and the optical signal are different from each other.
  • the phase difference between is 0 or ⁇ , which realizes the sign adjustment of the result of the multiplication operation.
  • the phase difference between the reference light and the optical signal is 0, and the signs of the amplitude of the reference light and the optical signal are the same. Assuming that the product of the two data indicated by the reference light is a positive sign, it can be regarded as the first The sign of the result of the multiplication operation implemented by the waveguide 120 is a positive sign.
  • the phase difference between the reference light and the optical signal is ⁇ , and the signs of the amplitude of the reference light and the optical signal are different. Assuming that the product of the two data indicated by the reference light is a positive sign, it can be regarded as the first The sign of the result of the multiplication operation implemented by the waveguide 120 is a negative sign.
  • the optical signal needs to continue to be transmitted after passing through the first phase modulation unit 121 to reach the beam combiner 130, in order to ensure that the beam combiner 130 detects the optical signals from the N-1 first waveguides 120 and the optical signals from the second waveguide 140
  • the addition operation can be accurately realized, that is, the sign of the result of each multiplication operation can be accurate, and the first phase modulation unit 121 can adjust the phase of the optical signal, so that the phase-adjusted optical signal and the reference light
  • the phase difference when reaching the beam combiner 130 is 0 or ⁇ .
  • the first phase modulation unit 121 can be calibrated in advance to determine that the When the received optical signal and the reference light reach the beam combiner 130, the phase difference is adjusted to a first phase parameter of 0 or ⁇ .
  • the optical splitter 110 divides the received continuous light into N optical signals, and then determines the reference light in the N optical signals.
  • N-1 first waveguides 120 receive the N-1 optical signals, and one first waveguide 120 receives one optical signal among the N optical signals.
  • the second waveguide 140 receives the remaining one optical signal except for the N-1 optical signal among the N optical signals.
  • the first waveguide 120B adjusts the amplitude modulation unit of the first waveguide 120 except the first waveguide 120B so that the The amplitude of the optical signal received by the first waveguide 120 other than 120B is attenuated to zero, so that the first waveguide 120 other than the first waveguide 120B will not output an optical signal.
  • the second waveguide 140 may not adjust the phase of the optical signal.
  • the amplitude modulation unit 122 in the first waveguide 120B may not adjust the amplitude of the optical signal, and gradually adjust the working parameters of the first phase modulation unit 121 in the first waveguide 120B, adjust the phase of the optical signal received by the first waveguide 120B, and combine the beams.
  • the receiver 130 receives the optical signals output by the second waveguide 140 and the first waveguide 120B, and combines the optical signals output by the second waveguide 140 and the first waveguide 120B into one optical signal.
  • the optical signal received by the second waveguide 140 is Where A 1 is the amplitude of the optical signal received by the second waveguide 140, Is the phase of the optical signal.
  • a 1 can be measured and determined in advance.
  • the optical signal received by the first waveguide 120B is Where A 2 is the amplitude of the optical signal received by the first waveguide 120B, Is the phase of the optical signal. A 2 can be measured and determined in advance.
  • the optical signal after being combined by the beam combiner 130 is
  • Optical signal detected by detector 150 The light intensity of is (A 1 +A 2 ) 2 .
  • Optical signal detected by detector 150 The light intensity of is (A 1 -A 2 ) 2 .
  • the light intensity of the optical signal output by the beam combiner 130 can be detected by the detector 150 to determine the optical signal output by the beam combiner 130 When the light intensity of is (A 1 +A 2 ) 2 , the first working parameter of the first phase modulation unit 121 of the first waveguide 120B; the optical signal output by the beam combiner 130 is determined When the light intensity of is (A 1 -A 2 ) 2 , the second operating parameter of the first phase modulation unit of the first waveguide 120B.
  • the first working parameter and the second working parameter may be used as the first phase parameter of the first phase modulation unit 121 of the first waveguide 120B.
  • the continuous light received by the beam combiner 130 is the same as the continuous light received by the beam combiner 130 when the subsequent light calculation device performs multiplication and addition operations.
  • the N optical signals generated by the beam combiner 130 using the continuous light are the same as the N optical signals generated by the beam splitter 110 using the continuous light when the subsequent optical computing device performs multiplication and addition operations.
  • the amplitude modulation unit 122 in the first waveguide 120 can be operated to adjust the amplitude of the optical signal; and the calibration method of the first phase modulation unit 121 is only an example, and the embodiment of the present application does not limit the operation of the first phase modulation unit 121
  • the specific calibration method, any calibration method capable of determining the first phase parameter is applicable to the embodiment of the present application.
  • the specific form of the first phase modulation unit 121 is not limited in the embodiment of the present application, and any device that can adjust the phase of the optical signal can be used as the first phase modulation unit 121.
  • the first phase modulation unit 121 may be a passive waveguide capable of transmitting optical signals, and the refractive index of the passive waveguide is changed by means of carrier injection, carrier depletion, quantum confinement Stark effect, etc., and then Change the phase of the optical signal.
  • the use of carrier injection to change the refractive index of the passive waveguide refers to adjusting the magnitude of the forward current applied to the passive waveguide to change the number of carriers in the passive waveguide and change the refraction of the passive waveguide. rate.
  • Using carrier depletion to change the refractive index of the passive waveguide refers to adjusting the magnitude of the reverse voltage applied to the passive waveguide to change the number of carriers in the passive waveguide and change the refractive index of the passive waveguide .
  • Using the quantum confinement Stark effect to change the refractive index of the passive waveguide refers to adjusting the magnitude of the reverse voltage applied to the passive waveguide, so that the energy band of the passive waveguide material is bent, and the refraction of the passive waveguide is changed. rate.
  • the first phase modulation unit 121 in the N-1 first waveguides only needs to adjust the phases of the N-1 optical signals to the same value.
  • the first phase modulation unit 121 in the N-1 first waveguides can adjust the phase difference between the N-1 intermediate optical signals and the reference light reaching the beam combiner to zero.
  • At least two amplitude modulation units 122 in the first waveguide 120 are connected in series with the first phase modulation unit 121, in order to ensure that the at least two amplitude modulation units and the first phase modulation unit can work normally, At least two amplitude modulation units are electrically isolated from two adjacent units in the first phase modulation unit. Electrical isolation refers to electrical isolation between two adjacent units to ensure that the two adjacent units are electrically isolated. The working voltage or current of the unit will not crosstalk, and the embodiment of the present application does not limit the specific method of electrical isolation. For example, methods such as proton implantation, oxygen ion implantation, isolation tank, etc. may be used.
  • the second waveguide 140 may only adjust the phase of the received optical signal.
  • the second waveguide 140 includes a second phase modulation unit 141.
  • the second phase modulation unit 141 has the same function as the first phase modulation unit 121, and can adjust the phase difference when the received optical signal and the reference light reach the beam combiner 130 to 0 or ⁇ .
  • the second phase modulation unit 141 adjusts the phase difference between the second optical signal and the reference light reaching the beam combiner 130 to 0, and adjusts the phase difference between the fourth optical signal and the reference light reaching the beam combiner 130 to ⁇ .
  • the second phase modulation unit 141 adjusts the phase difference between the second optical signal and the reference light reaching the beam combiner 130 to ⁇ , and adjusts the phase difference between the fourth optical signal and the reference light reaching the beam combiner 130 to zero.
  • the second phase modulation unit 141 can also be calibrated in advance to determine that it can The phase difference when the received optical signal and the reference light reach the beam combiner 130 is adjusted to a second phase parameter of 0 or ⁇ .
  • a calibrated optical signal received by the first phase modulation unit 121 can be used as the reference light.
  • the method of calibrating the second phase modulation unit 141 is the same as that of the first phase modulation unit 121.
  • the calibration method is the same. For details, please refer to the foregoing content, which will not be repeated here.
  • the second waveguide 141 includes an amplitude modulation unit 142.
  • the amplitude modulation unit 142 can adjust the light intensity of the received Nth optical signal to zero. That is, the amplitude of the N-th optical signal is adjusted to zero.
  • the detector 150 can only detect the light intensity of the optical signal output by the beam combiner 130, the light intensity is equal to the square of the amplitude of the optical signal, which means that the phase of the optical signal cannot be determined, that is, the multiplication and addition operation of the final output is determined. symbol.
  • the second waveguide 140 is provided in the embodiment of the present application. Through the second waveguide 140, the light intensity of the optical signal output by the combiner can be changed, so that the subsequent processing circuit 160 can determine the middle of the N-1 path according to the light intensity of different optical signals The sum of the products of the data indicated by the optical signal.
  • the beam combiner 130 has a beam combining function, and the operation performed by the beam combiner 130 is the inverse operation of the operation performed by the beam splitter 110.
  • the embodiment of the present application does not limit the specific type of the beam combiner 130, for example
  • the beam combiner 130 may be an MMI or a cascade Y branch.
  • processing circuit 160 determines the final output result of the multiplication and addition operation.
  • a N can be measured and determined in advance when calibrating the second phase modulation unit.
  • the optical signal output by the second waveguide 140 at this time is the second optical signal
  • the optical signal output by the beam combiner is the third optical signal.
  • the optical signal output by the second waveguide 140 at this time is the fourth optical signal
  • the optical signal output by the beam combiner is the fifth optical signal.
  • the optical signal Take the optical signal as an example, the fifth optical signal detected by the detector 150
  • the second phase modulation unit 140 performs two phase adjustment operations of the optical signal, respectively, adjusts the phase difference between the optical signal output by the second waveguide 140 and the reference light to the beam combiner 130 to 0 and ⁇ , and detects
  • the device 150 respectively determines the light intensity M 0 of the third optical signal output by the beam combiner 130 and the light intensity M ⁇ of the fifth optical signal output by the beam combiner 130; after that, the middle of the N-1 path is determined according to M 0 and M ⁇ .
  • the sum of the product of the data indicated by the optical signal is the result of the multiplication and addition operation.
  • the embodiment of the present application does not limit the specific type of the processing circuit 160.
  • the processing circuit 160 can be a simple logic circuit, such as an adder.
  • the function of the processing circuit 160 can also be implemented by other devices, for example,
  • the functions of the processing circuit 160 can be composed of a central processing unit (CPU), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), and artificial intelligence (artificial intelligence).
  • AI AI chip, system on chip (system on chip, SoC) or complex programmable logic device (complex programmable logic device, CPLD) and other implementations.
  • Any circuit or device that can determine the sum of the product of the data indicated by the N-1 intermediate optical signal according to the light intensity of the optical signal output by the beam combiner is applicable to the embodiments of the present application.
  • the optical splitter 110 receives the continuous light, it divides the continuous light into N optical signals (optionally, the optical splitter may also divide the continuous light into N-1 optical signals), and N- One first waveguide 120 receives N-1 optical signals among the N optical signals, and one first waveguide 120 receives one optical signal among the N-1 optical signals.
  • Each first waveguide 120 adjusts the amplitude of a received optical signal according to the set amplitude parameter, and each first waveguide outputs an intermediate optical signal.
  • the beam combiner 130 receives the intermediate optical signals output by the N-1 first waveguides 120, and combines the N-1 intermediate optical signals into one optical signal.
  • the detector 150 detects the light intensity of the optical signal, and the processing circuit 160 determines the sum of the products of the data indicated by the N-1 intermediate optical signals according to the light intensity of the optical signal.
  • the optical splitter 110 receives the continuous light, it divides the continuous light into N optical signals, and a second waveguide 140 and N-1 first waveguides 120 receive the N optical signals, where: One first waveguide 120 receives one optical signal among the N optical signals, and one second waveguide 140 receives the Nth optical signal among the N optical signals.
  • Each first waveguide 120 adjusts the amplitude of the received optical signal according to the set amplitude parameter, adjusts the phase of the optical signal according to the set phase parameter, and each first waveguide outputs an intermediate optical signal.
  • the second waveguide 140 adjusts the phase of the received optical signal, so that the phase of the adjusted optical signal reaching the combiner 130 is the first phase or the second phase.
  • the beam combiner 130 receives the optical signals output by the N-1 first waveguide 120 and the second waveguide 140, and combines the N-1 intermediate optical signals and the optical signal output by the second waveguide 140 into one optical signal.
  • the detector 150 detects the light intensity of the optical signal output by the beam combiner 130, and the processing circuit 160 determines the sum of the products of the data indicated by the N-1 intermediate optical signals according to the light intensity of the optical signal output by the beam combiner 130.
  • the combiner 130 receives N-1 intermediate optical signals and the second waveguide 140 outputs The second optical signal is combined with the N-1 intermediate optical signals and the second optical signal output by the second waveguide 140 into a third optical signal; the detector 150 detects the light intensity of the third optical signal.
  • the combiner 130 receives N-1 intermediate optical signals and the fourth optical signal output by the second waveguide 140, and divides N -1 intermediate optical signal and the fourth optical signal output by the second waveguide 140 are combined into a fifth optical signal; the detector 150 detects the light intensity of the fifth optical signal.
  • the processing circuit 160 obtains the sum of the products of the data indicated by the N-1 intermediate optical signals according to the light intensity of the third optical signal output by the detector 150 and the light intensity of the fifth optical signal.
  • the difference between the first phase and the phase when the reference light reaches the combiner is 0, the difference between the second phase and the phase when the reference light reaches the combiner is ⁇ , and the reference light is the N-way One optical signal in the optical signal.
  • the difference between the first phase and the phase when the reference light reaches the combiner is ⁇
  • the difference between the second phase and the phase when the reference light reaches the combiner is 0, and the reference light is among the N optical signals Light signal.
  • the computing system 10 includes an optical computing device 100 and a processor 200.
  • the structure of the optical computing device 100 may refer to the optical computing device shown in FIG. 1 or FIG. 2.
  • the processor 200 is connected to the optical computing device 100, and the processor 200 may send the data to be multiplied and added to the optical computing device 100 to instruct the optical computing device 100 to perform optical calculation on the received data.
  • the data to be multiplied and added includes N-1 groups of data, and each group of data includes at least two data to be multiplied in the multiplication and addition operation.
  • the amplitude parameter of the N-1 first waveguide can be set, and optionally, the phase parameter of the N-1 first waveguide can also be set.
  • the optical computing device 100 can set at least two amplitude parameters of the first waveguide according to a set of data, and set the phase parameter of the first waveguide according to the positive and negative values of the set of data. After that, the optical computing device 100 may perform optical calculations, output the calculation result of the multiplication and addition operation, and feed back the calculation result to the processor 200. In practical applications, the calculation result may be determined by the optical computing device 100 (the processing circuit in) according to the first optical signal, or may be determined according to the light intensity of the third optical signal and the light intensity of the fifth optical signal.

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Optics & Photonics (AREA)
  • Optical Modulation, Optical Deflection, Nonlinear Optics, Optical Demodulation, Optical Logic Elements (AREA)

Abstract

一种光计算设备（100）、光运算方法以及计算系统（10），光计算设备（100）包括分光器（110）、N-1个第一波导（120）、一个第二波导（140）以及一个合束器（130）。分光器（110）可以将接收的连续光分为N路光信号，并将N路光信号分别传输到N-1个第一波导（120）和一个第二波导（140）中，每个第一波导（120）接收一路相干光，第二波导（140）接收一路相干光。N-1个第一波导（120）调整接收到的N-1路相干光的振幅和相位。第二波导（140）调整接收的一路光信号的相位。合束器（130）可以将N-1个第一波导（120）以及第二波导（140）输出的相干光合并为一路光信号。第一波导（120）能够调整相干光的振幅和相位，使得光计算设备（120）能够实现带符号乘法运算，且由于光计算设备（120）中借助了第一波导（120）和第二波导（140），能够有效减少光计算设备（100）所占用的体积。

Description

一种光计算设备、光运算方法以及计算系统
本申请要求在2019年12月20日提交中华人民共和国知识产权局、申请号为201911329810.0、发明名称为“一种光计算设备、光运算方法以及计算系统”的中国专利申请的优先权以及在2019年11月1日提交中华人民共和国知识产权局、申请号为201911061911.4、发明名称为“一种片上集成乘加器”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及信息技术领域，尤其涉及一种光计算设备、光运算方法以及计算系统。
背景技术
乘加运算是一种基础运算，乘加运算中将多个乘法结果相加获取最后的结果，乘加器是实现乘加运算的计算单元，乘法累加器的种类有许多，其中，光学乘加器鉴于其采用了光学方法，具备了高效，低功耗的优点。
目前，常见的光学乘加器的实现方式主要是基于光纤系统的光学乘加器。这种光学乘加器中包括由通过光纤连接的多个声光调制器构成的声光调制器阵列以及多个探测器。这种基于光纤实现的光学乘加器实现乘加运算的过程中，可以先通过声光调制器阵列在多路光信号上调制信号，实现乘法运算。然后利用多个探测器分别接收声光调制器阵列调制后的光信号，并将接收的光信号转换为电信号，并将多个探测器获得的多个电信号进行叠加，以实现加法运算，从而获得计算结果。
由于现有技术中的光学乘加器利用了光纤系统作为光信号的传输载体，导致该种光学乘加器体积和功耗都较大。并且，由于该种光学乘加器中的加法运算部分是通过电学完成的，运算效率并不高。
发明内容
本申请提供一种光计算设备、光运算方法以及计算系统，用以提供一种体积较小，扩展性较高的乘加器。
第一方面，本申请提供了一种光计算设备，该光计算设备包括分光器、N-1个第一波导以及合束器。所述分光器分别连接N-1个第一波导的输入端，所述合束器分别连接N-1个第一波导的输出端。所述分光器可以将接收的连续光分为N路光信号，并将N路光信号分别传输到N-1个第一波导和一个第二波导中，其中，每个第一波导接收一路光信号，所述第二波导接收一路光信号。所述N-1个第一波导具备振幅调制功能，可以根据设置的振幅参数调整接收到的N-1路光信号的振幅，输出N-1路中间光信号。其中，每个第一波导设置有至少两个振幅参数。所述至少两个振幅参数用于指示待进行的乘法运算的相同数量的数据，该第一波导输出的中间光信号可以指示该至少两个振幅参数所指示的数据的乘积。所述合束器接收N-1路中间光信号，将N-1路中间光信号合并为第一光信号。
在本申请提供的设备中，通过第一波导调整光信号的振幅，使得该光计算设备能够实现乘法运算。并且，N-1个第一波导输出的N-1路中间光信号经过合束器，合束器能够对N-1路中间光信号进行合束，使得该光计算设备能够实现加法运算，保证该光计算设备能够实现 乘加运算。由于该光计算设备中采用波导来实现乘加运算，能够有效减少光计算设备所占用的体积和功耗，提高了该光计算设备的扩展性，进而使得该光计算设备能够计算的数据量更大。
在一种可能的设计中，该光计算设备还可以包括第二波导以及一个合束器。分光器可以连接第二波导的输入端，合束器连接第二波导的输出端。
分光器将接收的连续光分为N路光信号，将N路光信号分别传输到N-1个第一波导和一个第二波导中，N-1个第一波导接收N-1路光信号，第二波导接收除N-1路光信号外的一路光信号。
N-1个第一波导调整接收到的N-1路光信号后，可以根据设置的振幅参数调整N-1路光信号的振幅，根据设置的相位参数调整第一波导输出的光信号的相位，该相位参数可以指示该第一波导输出的中间光信号所指示的数据的乘积的和的正负值。
第二波导接收到光信号后，可以调整该路光信号的相位，输出第二光信号，该第二光信号到达合束器的相位为第一相位。
合束器可以N-1个第一波导输出的N-1路以及第二波导输出的第二光信号合并为第三光信号。
在本申请提供的设备中，通过第一波导调整光信号的振幅，使得该光计算设备能够实现乘法运算。又由于第一波导还具备调整光信号相位的功能，能够改变光信号的相位，为每个乘法运算的结果附加符号。之后经过合束器能够对第一波导以及第二波导输出的光信号进行合束，使得该光计算设备能够实现带符号的加法运算。由于第二波导能够调整光信号的相位，能够改变最终合束器输出的光信号的光强，借助第二波导可以确定第一波导输出的中间光信号合束后的光信号的振幅(也就是乘加运算的结果)，保证该光计算设备能够实现带符号的乘加运算。同样的，由于该光计算设备中采用波导来实现乘加运算，能够有效减少光计算设备所占用的体积和功耗，确保了该光计算设备的扩展性，使得光计算设备所能实现的计算量增大。
在一种可能的设计中，第二波导还可以再次调整接收的第N路光信号的相位，将接收的一路光信号调整为第四光信号，第四光信号到达所述合束器的相位为第二相位；
所述合束器之后将N-1个第一波导输出N-1路中间光信号以及第四光信号合并为第五光信号。
在本申请提供的设备中，由于第二波导能够多次调整光信号的相位，能够输出多个不同的光信号(如第二光信号和第四光信号)，使得合束器输出的光信号的光强发生变化，以使得该光计算设备能够实现带符号的乘加运算。
在一种可能的设计中，该光计算设备中还可以包括探测器和处理电路，该探测器可以探测合束器输出的光信号(如第一光信号、第三光信号以及第五光信号)的光强。处理电路能够从探测器获取合束器输出的光信号的光强，根据该光强确定N-1路中间光信号所指示的数据的乘积的和。
例如，处理电路可以直接根据第一光信号的光强确定N-1路中间光信号所指示的数据的乘积的和，也就是不带符号的乘加运算的计算结果。处理电路还可以直接根据第三光信号的光强和第五光信号的光强确定N-1路中间光信号所指示的数据的乘积的和，也就是带符号的乘加运算的计算结果。
在本申请提供的设备中，处理电路可以较为便捷的通过合束器输出的光信号的光强确定乘加运算的计算结果，消耗的资源较少，确定方式更加便捷，可以有效提高光计算设备实现 乘加运算的效率。
在一种可能的设计中，每个第一波导包括至少两个振幅调制单元，用于实现乘加运算中的一组乘法运算。
任一振幅调制单元可以根据一个振幅参数调整接收到的光信号的振幅，该一个振幅参数可以指示所需实现的乘加运算中一个需要实现的乘法运算的一个数据。
本申请提供的设备中，第一波导能够通过调整振幅的方式实现乘法运算，使得乘法运算的实现更加简单、高效。
在一种可能的设计中，每个第一波导包括两个振幅调制单元，分别为第一振幅调制单元和第二振幅调制单元，第二振幅调制单元的输入端连接第一振幅调制单元输出端。
第一振幅调制单元可以根据第一振幅参数调整所述光信号的振幅。之后，第二振幅调制单元可以根据第二振幅参数调整第一振幅调制单元调整后的光信号的振幅。
在一种可能的设计中，第一波导还包括第一相位调制单元，一个第一相位调制单元和至少两个振幅调制单元串联连接。这里不限定第一相位调制单元和至少两个振幅调制单元的连接顺序。第一波导中的各个单元对前一个单元输出的光信号进行调整。
第一相位调制单元能够根据相位参数调整接收到的光信号的相位，使调整相位后的光信号(也即中间光信号)与参考光到达所述合束器时的相位的差为0或π，其中，参考光为N路光信号中的一路光信号。
本申请提供的设备中，第一波导能够通过调整振幅的方式实现乘法运算，通过调整相位的方式为乘法运算的结果附加符号，乘法运算的实现更加简单、高效。
在一种可能的设计中，第一相位调制单元可以预先进行校准，以确定能够将光信号和参考光到达合束器时的相位差调整为0或π的第一相位参数。之后，在实际进行乘加运算时，可以直接根据第一相位参数将中间光信号与参考光到达合束器时的相位差调整为0或π。
在本申请提供的设备中，第一相位参数由于是预先确定的，能够使得第一相位调制单元可以更加精准的控制中间光信号和参考光到达合束器时的相位差。
在一种可能的设计中，第二波导包括第二相位调制单元。第二相位调制单元可以调整接收到的光信号的相位，使调整相位后的光信号到达合束器时的相位为第一相位或第二相位，参考光为N路光信号中的一路光信号。
这里不限定第一相位和第二相位的具体数值，第一相位和第二相位需满足如下条件之一：
条件一、第一相位与参考光到达所述合束器时的相位的差为0，第二相位与参考光到达所述合束器时的相位的差为π。
条件二、第一相位与参考光到达所述合束器时的相位的差为π，第二相位与参考光到达所述合束器时的相位的差为0。
在本申请提供的光计算设备中，第二波导仅需要调整光信号的相位，改变输出的光信号与参考光到达合束器时的相位差，以改变合束器输出的光信号的光强。也就是说，光信号与参考光到达合束器时的不同相位差，使得合束器输出不同的光信号，通过不同相位差下合束器输出的光信号的光强，可以方便、直接的确定N-1个中间光信号合束后的光信号的所指示的数据的乘积之和的正负值。
在一种可能的设计中，第二相位调制单元可以预先进行校准，已确定能够将光信号和参考光到达合束器时的相位差调整为0或π的第二相位参数。之后，在实际进行乘加运算时，可以直接根据第二相位参数将光信号与参考光到达合束器时的相位差调整为0或π。
在本申请提供的设备中，第二相位参数由于是预先确定的，能够使得第为相位调制单元 可以更加精准的控制光信号和参考光到达合束器时的相位差。
在一种可能的设计中，该光计算设备中还包括激光器，激光器与分光器连接，能够产生连续光，激光器将该连续光输出值分光器。其中，激光器为单纵模半导体激光器。
在本申请提供的设备中，利用激光器能够产生单一频率的连续光，便于后续处理。
在一种可能的设计中，振幅调制单元为EAM、SOA或VOA。
在一种可能的实现方式中，合束器为MMI或级联Y分支；分光器为MMI或级联Y分支。
第二方面，本申请提供了一种光运算方法，有益效果可以参见第一方面的相关描述，此处不再赘述。该方法由光计算设备执行，分光器将接收的连续光分为N路光信号；N-1个第一波导接收N路光信号中的N-1路光信号，根据设置的振幅参数调整该N-1路光信号的振幅，输出N-1路中间光信号，其中，每个第一波导接收一路光信号，每个第一波导中设置有至少两个振幅参数，至少两个振幅参数可以指示待进行乘法计算的至少两个数据，这样，每个第一波导输出的中间光信号可以指示第一波导中设置的至少两个振幅参数对应的至少两个数据的乘积；
合束器接收N-1路中间光信号，将N-1路中间光信号合并为第一光信号，第一光信号可以指示N-1路中间光信号所指示的数据的乘积的和。
在一种可能的设计中，N-1个第一波导中的每个第一波导还可以根据设置的相位参数调整第一波导输出的光信号的相位；第一波导输出的光信号的相位能够指示中间光信号所指示的乘积的正负值。
第二波导接收N路光信号中的第N路光信号，调制第N路光信号的相位，可以将接收的第N路光信号调整为第二光信号，第二光信号到达合束器的相位为第一相位；
之后，合束器可以将的N-1路中间光信号以及第二光信号合并为第三光信号。
在一种可能的设计中，第二波导还可以再次对第N路光信号的相位进行调整，将接收的第N路光信号调整为第四光信号，第四光信号到达合束器的相位为第二相位；
之后，合束器将N-1路中间光信号以及第四光信号合并为第五光信号。
在一种可能的设计中，探测器可以探测第一光信号的光强；处理电路可以根据探测器输出的第一光信号的光强获得N-1路中间光信号所指示的数据的乘积的和。
探测器还可以探测第三光信号的光强以及探测第五光信号的光强；处理电路根据探测器输出的第三光信号的光强以及第五光信号的光强获得N-1路中间光信号所指示的数据的乘积的和。
在一种可能的设计中，第一相位与参考光到达合束器时的相位的之差为0，第二相位与参考光到达合束器时的相位的之差为π，参考光为N路光信号中的一路光信号。或第一相位与参考光到达合束器时的相位的之差为π，第二相位与参考光到达合束器时的相位的之差为0，参考光为N路光信号中的一路光信号。
第三方面，本申请提供了一种计算系统，所述计算系统可以包括处理器和如第一方面或第一方面的任意一种可能的实现方式中所述的光计算设备。
处理器可以向光计算设备发送待进行乘加运算的数据。具体的，该待进行乘加运算的数据中包括N-1组数据，每组数据包括该乘加运算中的待进行乘法运算的至少两个数据。所述光计算设备接收到该待进行乘加运算的数据后，可以根据一组数据设置第一波导的振幅参数，可选的，还可以设置相位参数。之后，该光计算设备进行光计算，输出该乘加运算的计算结果，将该计算结果反馈给处理。
附图说明
图1为本申请提供的一种光计算设备的结构示意图；
图2为本申请提供的另一种光计算设备的结构示意图；
图3为本申请提供的一种第一波导的结构示意图；
图4为本申请提供的一种第二波导的结构示意图；
图5为本申请提供的一种光运算方法示意图；
图6为本申请提供的一种光运算方法示意图；
图7为本申请提供的一种计算系统示意图。
具体实施方式
本申请提供了一种光计算设备和光运算方法，用以提供一种体积较小、扩展性高的光学乘加器。
如图1所示为本申请实施例提供的一种光计算设备，该光计算设备包括分光器110、N-1个第一波导120、合束器130。
分光器110分别与N-1个第一波导120连接，分光器110可以将接收到的连续光分为N路光信号，之后将这N路光信号中的N-1路光信号分别传输至N-1个第一波导120；其中，一个第一波导120接收N-1路光信号中的一路光信号。
N-1个第一波导120可以根据振幅参数调整接收到的N-1路光信号的振幅，输出N-1路中间光信号。
对于N-1个第一波导120中的任一第一波导120，该第一波导120接收到N-1路光信号中的一路光信号，根据设置的至少两个振幅参数调整该路光信号的振幅，输出一路中间光信号。该至少两个振幅参数可用于指示带进行乘法运算的至少两个数据，一个振幅参数至少一个数据，这样输出的一路中间光信号可以指示该指示两个数据的乘积。
合束器分别连接N-1个第一波导，接收N-1个第一波导120输出的N-1路中间光信号，将N-1路中间光信号合并为第一光信号，N第一光信号用于指示所述N-1路中间光信号所指示的数据的乘积的和。
在本申请实施例中，第一波导120具备振幅调制功能，能够对接收到的光信号的振幅进行调整。可选的，第一波导120也可以具备相位调制功能，N-1个第一波导120可以调整N-1路光信号的相位，使第一波导120输出中间光信号的相位相同。
以光计算设备所要实现的乘加运算为:
Figure PCTCN2020125385-appb-000001
(其中，N-1个X iW i的符号相同，可以均为正号，也可以均为负号)为例，该光计算设备利用N-1个第一波导120对N-1路光信号的振幅进行调整，实现该乘加运算中的N-1个乘法运算，其中，每个第一波导120根据振幅参数对N-1路光信号中的一路光信号的振幅进行调整，该振幅参数指示X iW i，这样中间光信号的振幅可以指示N-1个乘法运算中一个乘法运算X iW i。之后，N-1个第一波导120输出的中间光信号在合束器140中合束之后，输出第一光信号，该第一光信号的振幅为
Figure PCTCN2020125385-appb-000002
指示了N-1个乘积的和。
在本申请实施例中，由于第一波导120能够调整光信号的振幅，使得该光计算设备能够实现不带符号的乘法运算，之后经过合束器140能够对第一波导120输出的中间光信号进行合束，使得该光计算设备能够实现加法运算，保证该光计算设备能够实现不带符号的乘加运算(每个乘法运算的结果的符号相同)。由于该光计算设备中借助了第一波导120，能够有效 缩小光计算设备所占用的体积，提高了该光计算设备的扩展性，进而使得该光计算设备能够计算的数据量更大。
如图1所示的光计算设备中能够实现不带符号的乘加运算(每个乘法结果可以为正数，也可以为负数)，为了实现带符号的乘加运算，对如图1所示的光计算设备进行改进，如图2所示，为本申请实施例提供的另一种光计算设备，该光计算设备包括分光器110、N-1个第一波导120、一个第二波导140以及合束器130。
分光器110分别与N-1个第一波导120以及一个第二波导140连接，分光器110可以将接收到的连续光分为N路光信号，之后将这N路光信号分别传输至N-1个第一波导120以及一个第二波导140；其中，一个第一波导120接收N路光信号中的一路光信号，N-1个第一波导120接收N-1路光信号，第二波导140接收N路光信号中除N-1路光信号外的一路光信号，也即第N路光信号。
N-1个第一波导120中的任一第一波导120在接收到N路光信号中的一路光信号后，除了根据振幅参数调整该路光信号的振幅(第一波导120调整该路光信号的振幅的方式可以参见前述内容，此处不再赘述)，还可以根据相位参数调整接收到的该路光信号相位。之后，第一波导120输出调整了相位以及振幅后的光信号，在如图2所示的光计算设备中，第一波导120输出的光信号也可以称为中间光信号。其中，每个第一波导120上设置有相位参数，该相位参数指示中间光信号所指示的至少两个数据的乘积的正负值。以光计算设备所要实现的乘加运算为:
Figure PCTCN2020125385-appb-000003
为例，一个第一波导120设置的相位参数指示的是X iW i的正负值。
第二波导140在接收到第N路光信号后，可以调整接收到的该路光信号的相位，输出调整了相位的光信号，该光信号到达合束器的相位可以为第一相位或第二相位。也就是说，第二波导140与第一波导120不同，可以只对接收到的光信号的相位进行调整。
合束器130分别连接N-1个第一波导120和一个第二波导140，接收N-1个第一波导120输出的N-1个中间光信号和第二波导140输出的光信号，可以将N-1个中间光信号和第二波导140输出的光信号合并为一路光信号。
可选的，该光计算设备中还可以包括探测器150，探测器150可以探测合束器输出的光信号的光强。探测器150可以是光伏型探测器，也可以是光电导探测器。
以光计算设备所要实现的乘加运算为:
Figure PCTCN2020125385-appb-000004
为(这里不限制X iW i的符号，可以为正号，也可以为负号)例，该光计算设备利用N-1个第一波导120对N-1路光信号的振幅和相位的调整，实现该乘加运算中的N-1个乘法运算，其中，每个第一波导120对N-1路光信号中的一路光信号的振幅进行调整，实现N-1个乘法运算中一个乘法运算X iW i，也就是说，调整了振幅后的该路光信号的振幅即为该乘法运算的结果。每个第一波导对N-1路光信号中的一路光信号的相位进行调整，为该乘法运算X iW i前添加符号(如正号、或负号)。之后，N-1个中间光信号在合束器130中合束之后，输出光信号A，该光信号A的振幅可以指示
Figure PCTCN2020125385-appb-000005
由于探测器150所探测的光信号的光强即为光信号的振幅的平方，不能区分光信号的振幅的符号，为此，在光计算设备中增加了第二波导140，第二波导140能够调整第N路光信号的相位，该光计算设备可以通过第二波导140调整一路光信号的相位，进而改变N-1个中间光信号与第二波导140在合束器130中合束后的一路光信号的振幅大小，进而改变该路光信号的光强，通过光强的变化确定光信号A的符号(具体确定符号的方式可参见后文中描述)。
在本申请实施例中，由于第一波导120能够调整光信号的振幅，使得该光计算设备能够实现乘法运算；又由于第一波导120还具备调整光信号相位的功能，能够改变光信号的相位， 为每个乘法运算的结果附加符号；之后经过合束器130能够对第一波导120以及第二波导140输出的光信号进行合束，使得该光计算设备能够实现带符号的加法运算。由于第二波导140能够调整光信号的相位，能够改变最终合束器输出的光信号的光强，借助第二波导140可以确定第一波导120输出的光信号合束后的光信号的振幅的正负，保证该光计算设备能够实现带符号的乘加运算。同样的，由于该光计算设备中借助了第一波导120和第二波导140，能够有效缩小光计算设备所占用的体积，改善该光计算设备的扩展性，提升该光计算设备所能计算的数据量。
在如图2所示的光计算设备中，第二波导140可以对接收到第N路光信号分别执行两次相位调整操作，相应的，合束器可以合并出两种不同的光信号。
第一次相位调整操作：
第二波导140接收所述N路光信号中的第N路光信号，并将接收的所述第N路光信号调整为第二光信号，第二光信号到达所述合束器的相位为第一相位。
合束器130将N-1个第一波导120输出的所述N-1路中间光信号以及第二波导输出的第二光信号合并为第三光信号。
第二次相位调整操作：
第二波导140接收所述N路光信号中的第N路光信号，并将接收的所述第N路光信号调整为第四光信号，第三光信号到达所述合束器的相位为第二相位。
合束器130将N-1个第一波导120输出的所述N-1路中间光信号以及第二波导输出的第四光信号合并为第五光信号。
为了确定带符号的乘加运算的运算结果，也就是确定N-1个中间光信号合束后的光信号的振幅。
该光学计算设备中还可以包括探测器150。探测器150探测到第三光信号和第五光信号的光强后，探测器150通过探测器确定第三光信号和第五光信号的光强，根据第三光信号和第五光信号的光强获得N-1路中间光信号所指示的数据的乘积的和(具体确定的方式可参见后文中描述)。
值得注意的是，由图1所示和图2所示的光计算设备可知，图2所示的光计算设备也同样可以实现图1所示的光计算设备能实现的功能，进行控制第二波导不接收第N路光信号(如将第N路光信号的振幅调整为0)即可。
下面以图2所示的光计算设备为例，下面结合附图对该光计算设备中的各个组成部分进行说明：
(1)、分光器110
在本申请实施例中，分光器110具备分光功能，本申请实施例并不限定分光器110的具体类型，例如分光器110可以是多模干涉耦合器(multimode interference,MMI)，也可以是级联Y分支。
分光器110接收的光为连续光，所谓连续光是指光信号不存在间断。本申请实施例并不限定该连续光的来源，凡是能够产生连续光的器件均适用于本申请实施例。
作为一种可能的实施方式，该光计算设备中还可以包括激光器170，激光器170可以产生该连续光。示例性的，该激光器170可以是单纵模半导体激光器，产生具有单一频率的连续光。单纵模半导体激光器包括但不限于：分布反馈式(distributed feedback,DFB)激光器、分布布拉格反射式(distributed bragg reflector,DBR)激光器、或微环激光器。
对于分光器110分光后产生的N路光信号，可以是相同的N路光信号，如N路光信号的 相位和振幅都相同；也可以不同的N路光信号，例如N路光信号的振幅相同，相位不同。
需要说明的是，在分光器110进行分光时，由于分光器110本身分光的原理或环境因素会导致N路光信号的振幅存在差异，为了保证光信号的振幅保持一致，可以利用第一波导120对其中N-1路光信号的振幅进行补偿。同样的，由于分光器110本身分光的原理或环境因素也可能导致N路光信号的相位存在差异，为了保证N路光信号的相位保持一致，可以利用第一波导120和第二波导140对N路光信号的相位进行补偿。
(2)、第一波导120
在本申请实施例中第一波导120兼具振幅调制功能和相位调制功能，如图3所示，第一波导120包括一个第一相位调制单元121和至少两个振幅调制单元。一个第一相位调制单元121和至少两个振幅调制单元是串联连接的。本申请实施例并不限定一个第一相位调制单元和至少两个振幅调制单元的先后顺序，在如图2所示的第一波导120中，以第一波导120中包括两个振幅调制单元和一个第一相位调制单元121，两个振幅调制单元位于第一相位调制单元之前为例进行说明。为方便区分两个振幅调制单元分别用振幅调制单元122A和振幅调制单元122B表示。
分光器110产生的N路光信号中的一路光信号进入第一波导120，先经过振幅调制单元122A，之后再经过振幅调制单元122B，最后通过第一相位调制单元121。
该光信号在经过振幅调制单元122A和振幅调制单元122B时，振幅调制单元122A和振幅调制单元122B可以分别根据振幅参数调整接收到的光信号的振幅。该光信号在相继通过振幅调制单元122A和振幅调制单元122B，可以实现乘加运算中的一组乘法运算。
例如，光计算设备所要实现的乘加运算为:
Figure PCTCN2020125385-appb-000006
其中，每一第一波导120用于实现乘加运算中的一组乘法运算X iW i。振幅参数可以指示X i和W i，第一波导120中的一个振幅调制单元122A可以根据指示X i的振幅参数调整接收到的光信号的振幅，之后，另一个振幅调制单元122B根据指示W i的振幅参数调整振幅调制单元122A调整了振幅后的光信号的振幅。第一波导120中的一个振幅调制单元122A也可以根据指示W i的振幅参数调整接收到的光信号的振幅，之后，另一个振幅调制单元122B根据指示X i的振幅参数调整振幅调制单元122A调整了振幅后的光信号的振幅。这里不考虑X i和W i的正负，乘法运算的正负可有第一波导120通过调整光信号的相位实现。
需要说明的是，这里以第一波导120中包括两个振幅调制单元为例，第一波导120中也可以包括三个，甚至数目更多的振幅调制单元；振幅调制单元的数目与该光计算设备所要实现的乘加运算中一组乘法运算中的数据的数量有关，当数据的数量为三个，则第一波导120中可以包括三个振幅调制单元。也就是说，振幅调制单元的数目可以等于运算参数的数目。
在本申请实施例中并不限定振幅调制单元的具体形式，凡是可以调整光信号的振幅的器件均可作为振幅调制单元。例如，振幅调制单元可以是电吸收调制器(electro-absorption modulator，EAM)、半导体光放大器(semiconductor optical amplifier，SOA)、可调光衰减器(Variable Optical Attenuator,VOA)。
当振幅调制单元采用EAM或VOA时，通过控制施加在EAM或VOA的反向偏置的电压调节对光信号的吸收系数，实现对光信号的衰减，调整光信号的振幅；其中，反向偏置的电压即为振幅参数确定的。
当振幅调制单元采用SOA，通过控制施加在SOA的正向偏置的电流调节对光信号的放大系数，实现对光信号的衰减或放大，调整光信号的振幅。其中，正向偏置的电流即为振幅参数；当正向偏置的电流低于阈值时，可以实现对光信号的衰减，减小光信号的振幅；当正 向偏置的电流高于阈值时，可以实现对光信号的放大，增大光信号的振幅。
该光信号在相继通过振幅调制单元122A和振幅调制单元122B后，光信号的振幅发生了改变，调整了振幅的光信号再通过第一相位调制单元121，第一相位调制单元121调整振幅调制单元122B输出的光信号的相位，也就是可以调整第一波导120输出的光信号的相位，可以实现对乘法运算的结果的符号调整(如调整为正或负)。
在实际运算中，可以将N路光信号中的一路光信号作为参考光，第一相位调制单元121可以通过调整该第一波导120接收一路光信号的相位，使参考光与该路光信号之间的相位差为0或者π，实现对乘法运算的结果的符号调整。
参考光与该光信号之间的相位差为0，参考光和该光信号的振幅的符号相同，假设参考光所指示的两个数据的乘积为正号的情况下，可以看做该第一波导120所实现的乘法运算的结果的符号为正号。
参考光与该光信号之间的相位差为π，参考光和该光信号的振幅的符号不同，假设参考光所指示的两个数据的乘积为正号的情况下，可以看做该第一波导120所实现的乘法运算的结果的符号为负号。
由于光信号在通过第一相位调制单元121后，需要继续传输才能到达合束器130，为了保证合束器130对来自N-1个第一波导120的光信号以及第二波导140的光信号进行合束时，能够准确实现加法运算，也即能够各个乘法运算的结果的符号是准确的，第一相位调制单元121可以调整该光信号的相位，以使得相位调整后的光信号与参考光到达合束器130时的相位差为0或π。
为了保证第一相位调制单元121可以准确的将接收到的光信号与参考光到达合束器130时的相位差调整为0或π，可以预先对第一相位调制单元121进行校准，确定能够将接收到的光信号与参考光到达合束器130时的相位差调整为0或π的第一相位参数。
下面以校准一个第一波导120中的第一相位调制单元121为例，对预先对N-1个第一波导120中的第一相位调制单元121进行校准的方式进行说明：
分光器110将接收到的连续光分为N路光信号后，确定该N路光信号中的参考光。
N-1个第一波导120接收该N-1路光信号，一个第一波导120接收N路光信号中的一路光信号。第二波导140接收N路光信号中除N-1路的光信号外的剩余一路光信号。
以第二波导140接收的光信号为参考光，需要校准的第一波导120为第一波导120B为例，调节除第一波导120B外的第一波导120的振幅调制单元，使除第一波导120B外的第一波导120接收的光信号的振幅衰减为零，这样，除第一波导120B外的第一波导120不会输出光信号。
第二波导140可以不调整光信号的相位。第一波导120B中的振幅调制单元122可以不调整光信号的振幅，逐渐调整第一波导120B中的第一相位调制单元121的工作参数，调整第一波导120B接收的光信号的相位，合束器130接收第二波导140和第一波导120B输出的光信号，并将第二波导140和第一波导120B输出的光信号合并为一路光信号。
例如，第二波导140接收的光信号为
Figure PCTCN2020125385-appb-000007
其中，A 1为第二波导140接收的光信号的振幅，
Figure PCTCN2020125385-appb-000008
为该光信号的相位。A 1可以预先测量确定。
第一波导120B接收的光信号为
Figure PCTCN2020125385-appb-000009
其中，A 2为第一波导120B接收的光信号的振幅，
Figure PCTCN2020125385-appb-000010
为该光信号的相位。A 2可以预先测量确定。
经过合束器130合束后的光信号为
Figure PCTCN2020125385-appb-000011
若第二波导140和第一波导120B输出的光信号到达合束器130的相位差为0，
Figure PCTCN2020125385-appb-000012
探测器150检测的光信号
Figure PCTCN2020125385-appb-000013
的光强为(A 1+A 2) 2。
若第二波导140和第一波导120B输出的光信号到达合束器130的相位差为π，
Figure PCTCN2020125385-appb-000014
探测器150检测的光信号
Figure PCTCN2020125385-appb-000015
的光强为(A 1-A 2) 2。
每调整一次第一波导120B中的第一相位调制单元121的工作参数，可以通过探测器150探测合束器130输出的光信号的光强，确定合束器130输出的光信号
Figure PCTCN2020125385-appb-000016
的光强为(A 1+A 2) 2时，第一波导120B的第一相位调制单元121的第一工作参数；确定合束器130输出的光信号
Figure PCTCN2020125385-appb-000017
的光强为(A 1-A 2) 2时，第一波导120B的第一相位调制单元的第二工作参数。第一工作参数和第二工作参数，可以作为第一波导120B的第一相位调制单元121的第一相位参数。
需要说明的是，在进行校准时，合束器130接收的连续光与后续光计算设备进行乘加运算时，合束器130接收的连续光相同。相应的，在进行校准时，合束器130利用该连续光产生的N路光信号与后续光计算设备进行乘加运算时，分光器110利用该连续光产生的N路光信号相同。这样可以保证第一相位调制单元进行校准时，确定的第一相位参数是准确的，后续第一相位调制单元才可以较为准确的将接收到的光信号与参考光到达合束器130时的相位差调整为0或π。另外，在上述校准第一波导120中的第一相位调制单元的方式中，仅是以校准一个第一波导120中的第一相位调制单元为例进行说明的，作为一种可能的实施方式，也可以同时校准多个第一波导120的第一相位调制单元，分别确定多个第一波导120中每个第一波导120的第一相位调制单元的第一相位参数。
应需注意的是，在上述对第一波导120中的第一相位调制单元121进行校准的过程中，忽略了振幅调制单元122对光信号的相位的影响。事实上，第一波导120中的振幅调制单元122在工作时，也可能会导致光信号的相位发生变化，为此，在对第一波导120中的第一相位调制单元121进行校准时，也可以使该第一波导120中的振幅调制单元122工作，调整光信号的振幅；且上述第一相位调制单元121的校准方式仅是举例，本申请实施例并不限定第一相位调制单元121的校准的具体方式，凡是能够确定第一相位参数的校准方式均适用于本申请实施例。
在本申请实施例中并不限定第一相位调制单元121的具体形式，凡是可以调整光信号的相位的器件均可作为第一相位调制单元121。例如，第一相位调制单元121可以是一段能够传输光信号的无源波导，通过载流子注入、载流子耗尽、量子限制斯塔克效应等方式改变该无源波导的折射率，进而改变光信号的相位。其中，利用载流子注入改变该无源波导的折射率是指调节施加在该无源波导上的正向电流大小，以改变该无源波导中载流子的数目，改变无源波导的折射率。利用载流子耗尽改变该无源波导的折射率是指调节施加在该无源波导上的反向电压大小，以改变该无源波导中载流子的数目，改变无源波导的折射率。利用量子限制斯塔克效应改变该无源波导的折射率是指调节施加在该无源波导上的反向电压大小，使该无源波导的材料的能带发生弯曲，改变无源波导的折射率。需要说明的是，当仅需实现不带符号的乘加运算时，N-1个第一波导中第一相位调制单元121只需将N-1个光信号的相位调整为相同值即可，例如N-1个第一波导中第一相位调制单元121可以将N-1个中间光信号与参考光到达合束器的相位差调整为0。
作为一种可能的实施方式，由于第一波导120中至少两个振幅调制单元122与第一相位调制单元121串联连接，为了保证至少两个振幅调制单元与第一相位调制单元能够正常工作，在至少两个振幅调制单元与第一相位调制单元中相邻的两个单元之间进行电隔离，电隔离是指在相邻的两个单元中实现电气上的隔离，以保证相邻的两个单元的工作电压或电流不会串扰，本申请实施例并不限定电隔离的具体方式，例如可以采用质子注入、氧离子注入、隔离 槽等方式。
(3)、第二波导140
在本申请实施例中第二波导140可以只调整接收到的光信号的相位，如图4所示，第二波导140包括第二相位调制单元141。第二相位调制单元141与第一相位调制单元121的作用相同，可以将接收到的光信号与参考光到达合束器130时的相位差调整为0或π。例如，第二相位调制单元141将第二光信号与参考光到达合束器130的相位差调整为0，将第四光信号与参考光到达合束器130的相位差调整为π。又例如，第二相位调制单元141将第二光信号与参考光到达合束器130的相位差调整为π，将第四光信号与参考光到达合束器130的相位差调整为0。
为了保证第二相位调制单元141可以准确的将接收到的光信号与参考光到达合束器130时的相位差调整为0或π，也可以预先对第二相位调制单元141进行校准，确定能够将接收到的光信号与参考光到达合束器130时的相位差调整为0或π的第二相位参数。在对第二相位调制单元141校准时，可以将一个已校准的第一相位调制单元121接收的光信号作为参考光，对第二相位调制单元141进行校准的方式与对第一相位调制单元121进行校准的方式相同，具体可以参见前述内容，此处不再赘述。
可选的，第二波导141包括振幅调制单元142，当光计算设备用于实现不带符号的乘加运算时，振幅调制单元142可以将接收的第N路光信号的光强调整为零，也就是将第N路光信号的振幅调整为零。
由于探测器150只能探测合束器130输出的光信号的光强，光强等于光信号的振幅的平方，也就是说并不能确定该光信号的相位，也就是确定最终输出的乘加运算的符号。鉴于此，本申请实施例中设置第二波导140，通过第二波导140可以改变合束器输出的光信号的光强，便于后续处理电路160根据不同光信号的光强确定N-1路中间光信号所指示的数据的乘积的和。
(4)、合束器130
在本申请实施例中，合束器130具备合束功能，合束器130执行的操作是分光器110执行的操作的逆操作，本申请实施例并不限定合束器130的具体类型，例如合束器130可以是MMI，也可以是级联Y分支。
(5)、处理电路160。
下面对处理电路160确定最终输出的乘加运算的结果的方式进行说明：
假设，N-1个第一波导120输出的N-1中间光信号经过合束后输出的光信号的振幅Y为：
Figure PCTCN2020125385-appb-000018
假设，第二波导140输出的光信号为
Figure PCTCN2020125385-appb-000019
该光信号的振幅为A N。A N可以在校准第二相位调制单元时预先测量确定。
若第二波导140输出的光信号与参考光到达合束器130的相位差为0，以此时第二波导140输出的光信号为第二光信号，合束器输出的光信号为第三光信号为例，则探测器150检测的第三光信号
Figure PCTCN2020125385-appb-000020
的光强为M 0＝(Y+A N) 2。
若第二波导140输出的光信号与参考光到达合束器130的相位差为π，以此时第二波导140输出的光信号为第四光信号，合束器输出的光信号为第五光信号为例，则探测器150检测的第五光信号
Figure PCTCN2020125385-appb-000021
的光强为M π＝(Y-A N) 2。
Figure PCTCN2020125385-appb-000022
Figure PCTCN2020125385-appb-000023
由此可知：
Figure PCTCN2020125385-appb-000024
从上述过程可知，第二相位调制单元140分别进行两次光信号的相位调整操作，分别将第二波导140输出的光信号与参考光到达合束器130的相位差调整为0和π，探测器150分别确定合束器130输出的第三光信号的光强M 0以及合束器130输出的第五光信号的光强M π；之后，根据M 0和M π确定N-1路中间光信号所指示的数据的乘积的和，也就是该乘加运算的运算结果。
本申请实施例并不限定处理电路160的具体类型，处理电路160可以是简单的逻辑电路，如加法器，作为一种可能的实现方式，处理电路160的功能也可以由其他装置实现，例如，处理电路160的功能可以由中央处埋器(central processing unit，CPU)、专用集成电路(application specific integrated circuit，ASIC)、现场可编程门阵列(field programmable gate array，FPGA)、人工智能(artificial intelligence，AI)芯片、片上系统(system on chip，SoC)或复杂可编程逻辑器件(complex programmable logic device，CPLD)等实现。凡是能够根据合束器输出的光信号的光强确定N-1路中间光信号所指示的数据的乘积的和的电路或装置均适用于本申请实施例。
为了使方案描述更加清楚，下面将结合前面的实施例，以如图1所示的光计算设备和图5所示的光运算方法为例，对本发明实施例中提供的光计算设备的工作流程进行概括介绍。在本申请实施例中，分光器110接收到连续光后，将该连续光分为N路光信号(可选的，分光器也可以将连续光分为N-1路光信号)，N-1个第一波导120接收该N路光信号中的N-1路光信号，其中，一个第一波导120接收N-1路光信号中的一路光信号。每个第一波导120根据设置的振幅参数调整接收的一路光信号的振幅，每个第一波导输出一路中间光信号。合束器130接收N-1个第一波导120输出的中间光信号，将N-1个中间光信号合束为一路光信号。探测器150探测该光信号的光强，处理电路160根据该光信号的光强确定N-1路中间光信号所指示的数据的乘积的和。
为了使方案描述更加清楚，下面将结合前面的实施例，以如图2所示的光计算设备和图6所示的光运算方法为例，对本发明实施例中提供的光计算设备的工作流程进行概括介绍。在本申请实施例中，分光器110接收到连续光后，将该连续光分为N路光信号，一个第二波导140以及N-1个第一波导120接收该N路光信号，其中，一个第一波导120接收N路光信号中的一路光信号，一个第二波导140接收N路光信号中的第N路光信号。每个第一波导120根据设置的振幅参数调整接收的一路光信号的振幅，根据设置的相位参数调整该路光信号的相位，每个第一波导输出一路中间光信号。第二波导140调整接收到的光信号的相位，使调整了相位后的光信号到达合束器130的相位为第一相位或第二相位。合束器130接收N-1个第一波导120和第二波导140输出的光信号，将N-1个中间光信号和第二波导140输出的光信号合束为一路光信号。探测器150探测合束器130输出的光信号的光强，处理电路160根据合束器130输出的该光信号的光强确定N-1路中间光信号所指示的数据的乘积的和。
作为一种可能的实施方式，当第二波导140输出的第二光信号到达合束器130的相位为第一相位时，合束器130接收N-1个中间光信号和第二波导140输出的第二光信号，将N-1 个中间光信号和第二波导140输出的第二光信号合束为第三光信号；探测器150探测第三光信号的光强。
当第二波导140输出的第四光信号到达合束器130的相位为第二相位时，合束器130接收N-1个中间光信号和第二波导140输出的第四光信号，将N-1个中间光信号和第二波导140输出的第四光信号合束为第五光信号；探测器150探测第五光信号的光强。
处理电路160根据探测器150输出的所述第三光信号的光强以及所述第五光信号的光强获得所述N-1路中间光信号所指示的数据的乘积的和。
其中，第一相位与参考光到达合束器时的相位的之差为0，述第二相位与参考光到达合束器时的相位的之差为π，所述参考光为所述N路光信号中的一路光信号。或第一相位与参考光到达合束器时的相位的之差为π，述第二相位与参考光到达合束器时的相位的之差为0，所述参考光为N路光信号中的一路光信号。
如图7所示，为本申请实施例提供的一种计算系统10，所述计算系统10包括光计算设备100和处理器200。光计算设备100的结构可参见如图1或图2所示的光计算设备。
处理器200连接光计算设备100，处理器200可以将待进行乘加运算的数据发送给光计算设备100，以指示光计算设备100对接收的数据进行光计算。该待进行乘加运算的数据中包括N-1组数据，每组数据包括该乘加运算中的待进行乘法运算的至少两个数据。
光计算设备100接收到该待进行乘加运算的数据后，可以设置N-1第一波导的振幅参数，可选的，还可以设置N-1第一波导的相位参数。
针对任一第一波导，光计算设备100可以根据一组数据设置该第一波导的至少两个振幅参数，根据该组数据的正负值设置该第一波导的相位参数。之后，该光计算设备100可以进行光计算，输出该乘加运算的计算结果，将该计算结果反馈给处理器200。实际应用中，该计算结果可以是光计算设备100(中的处理电路)根据第一光信号确定的，也可以是根据第三光信号的光强和第五光信号的光强确定的。
需要说明的是，本申请所提供的实施例仅仅是示意性的。所属领域的技术人员可以清楚的了解到，为了描述的方便和简洁，在上述实施例中，对各个实施例的描述都各有侧重，某个实施例中没有详述的部分，可以参见其他实施例的相关描述。在本发明实施例、权利要求以及附图中揭示的特征可以独立存在也可以组合存在。在本发明实施例中以硬件形式描述的特征可以通过软件来执行，反之亦然。在此不做限定。

Claims (15)

  1. 一种光计算设备，其特征在于，包括：
    分光器，用于将接收的连续光分为N路光信号；
    N-1个第一波导，分别与所述分光器连接，用于接收所述N路光信号中的N-1路光信号，并根据设置的振幅参数调整所述N-1路光信号的振幅，输出N-1路中间光信号，其中，每个所述第一波导接收一路光信号，每个所述第一波导中设置有至少两个振幅参数，所述至少两个振幅参数用于指示待进行乘法计算的至少两个数据，每个所述第一波导输出的中间光信号用于指示所述第一波导中设置的至少两个振幅参数对应的至少两个数据的乘积；
    合束器，连接所述N-1个第一波导，用于将所述N-1个第一波导输出的N-1路中间光信号合并为第一光信号，所述第一光信号用于指示所述N-1路中间光信号所指示的数据的乘积的和。
  2. 如权利1所述的光计算设备，其特征在于：所述N-1个第一波导中的每个第一波导还用于根据设置的相位参数调整所述第一波导输出的光信号的相位，所述第一波导输出的光信号的相位用于指示所述第一波导输出的中间光信号所指示的至少两个数据的乘积的正负值；
    所述光计算设备还包括：
    第二波导，连接所述分光器，并用于接收所述N路光信号中的第N路光信号，并将接收的所述第N路光信号调整为第二光信号，所述第二光信号到达所述合束器的相位为第一相位；
    所述合束器，还用于连接所述第二波导，并用于将所述N-1个第一波导输出的所述N-1路中间光信号以及所述第二波导输出的第二光信号合并为第三光信号。
  3. 根据权利要求2所述的光计算设备，其特征在于：
    所述第二波导，还用于将接收的所述第N路光信号调整为第四光信号，所述第四光信号到达所述合束器的相位为第二相位；
    所述合束器，还用于将所述N-1个第一波导输出的N-1路中间光信号以及所述第四光信号合并为第五光信号。
  4. 根据权利要求3所述的光计算设备，其特征在于，所述光计算设备还包括：
    探测器，用于探测所述第三光信号的光强以及探测所述第五光信号的光强；
    处理电路，用于根据所述探测器输出的所述第三光信号的光强以及所述第五光信号的光强获得所述N-1路中间光信号所指示的数据的乘积的和。
  5. 如权利要求1或2所述的设备，其特征在于，每个所述第一波导包括至少两个振幅调制单元，
    所述至少两个振幅调制单元中的任一振幅调制单元，用于根据一个所述振幅参数调整所述第一波导接收的一路所述光信号的振幅。
  6. 如权利要求5所述的设备，其特征在于，
    所述至少两个振幅调制单元包括第一振幅调制单元和第二振幅调制单元，
    所述第一振幅调制单元，用于根据第一振幅参数调整所述光信号的振幅；
    所述第二振幅调制单元，用于根据第二振幅参数调整所述第一振幅调制单元调整后的光信号的振幅。
  7. 如权利要求5或6所述的设备，其特征在于，每个所述第一波导还包括所述第一相位调制单元，所述第一相位调制单元和所述至少两个振幅调制单元串联连接；
    所述第一相位调制单元，用于根据所述相位参数调整所述光信号的相位，使所述中间光 信号与参考光到达所述合束器时的相位差为0或π，所述参考光为所述N路光信号中的任意一路光信号。
  8. 如权利要求3所述的设备，其特征在于，所述第二波导包括第二相位调制单元；
    所述第二相位调制单元，用于调整所述第N路光信号的相位，使所述第一相位与参考光到达所述合束器时的相位之差为0，并使所述第二相位与所述参考光到达所述合束器时的相位的之差为π，其中，所述参考光为所述N路光信号中的任意一路光信号。
  9. 如权利要求5或6所述的设备，其特征在于，所述振幅调制单元为电吸收调制器EAM或半导体光放大器SOA或可调光衰减器VOA。
  10. 如权利要求1～8任一所述的设备，其特征在于，所述合束器为多模干涉耦合器MMI或级联Y分支；
    所述分光器为MMI或级联Y分支。
  11. 一种光运算方法，其特征在于，所述方法由光计算设备执行，所述光计算设备包括分光器、N-1个第一波导以及合束器，所述方法包括：
    所述分光器将接收的光分为N路光信号；
    所述N-1个第一波导接收所述N路光信号中的N-1路光信号，并根据设置的振幅参数调整所述N-1路光信号的振幅，输出N-1路中间光信号，其中，每个所述第一波导接收一路光信号，每个所述第一波导中设置有至少两个振幅参数，所述至少两个振幅参数用于指示待进行乘法计算的至少两个数据，每个第一波导输出的中间光信号用于指示所述第一波导中设置的至少两个振幅参数对应的至少两个数据的乘积；
    所述合束器将所述N-1个第一波导输出的N-1路中间光信号合并为第一光信号，所述第一光信号用于指示所述N-1路中间光信号所指示的数据的乘积的和。
  12. 如权利要求11所述的方法，其特征在于，所述光计算设备还包括第二波导，所述方法还包括：
    所述N-1个第一波导中的每个第一波导根据设置的相位参数调整所述第一波导输出的光信号的相位；
    所述第二波导接收所述N路光信号中的第N路光信号，并将接收的所述第N路光信号调整为第二光信号，所述第二光信号到达所述合束器的相位为第一相位；
    所述合束器将所述N-1个第一波导输出的所述N-1路中间光信号以及所述第二波导输出的第二光信号合并为第三光信号。
  13. 如权利要求12所述的方法，其特征在于，所述方法还包括：
    所述第二波导将接收的所述第N路光信号调整为第四光信号，所述第四光信号到达所述合束器的相位为第二相位；
    所述合束器将所述N-1个第一波导输出N-1路中间光信号以及所述第四光信号合并为第五光信号。
  14. 如权利要求13所述的方法，其特征在于，所述光计算设备还包括探测器和处理电路，所述方法还包括：
    所述探测器探测所述第三光信号的光强以及探测所述第五光信号的光强；
    所述处理电路根据所述探测器输出的所述第三光信号的光强以及所述第五光信号的光强获得所述N-1路中间光信号所指示的数据的乘积的和。
  15. 一种计算系统，其特征在于，所述系统包括处理器和如权利要求1～10任一所述的光计算设备，其中，所述处理器用于向所述光计算设备发送待进行乘加运算的数据。
PCT/CN2020/125385 2019-11-01 2020-10-30 一种光计算设备、光运算方法以及计算系统 WO2021083348A1 (zh)

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
CN201911061911 2019-11-01
CN201911061911.4 2019-11-01
CN201911329810.0 2019-12-20
CN201911329810.0A CN112783260A (zh) 2019-11-01 2019-12-20 一种光计算设备、光运算方法以及计算系统

Publications (1)

Publication Number Publication Date
WO2021083348A1 true WO2021083348A1 (zh) 2021-05-06

Family

ID=75714469

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/125385 WO2021083348A1 (zh) 2019-11-01 2020-10-30 一种光计算设备、光运算方法以及计算系统

Country Status (1)

Country Link
WO (1) WO2021083348A1 (zh)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0980123A2 (en) * 1998-08-11 2000-02-16 TRW Inc. High average power fiber laser system with phase front control
US20070201036A1 (en) * 2006-02-27 2007-08-30 The Regents Of The University Of California Coherent hybrid electromagnetic field imaging
CN101667983A (zh) * 2009-09-16 2010-03-10 华为技术有限公司 调制信号的产生方法和传输设备
CN102130739A (zh) * 2010-12-31 2011-07-20 华为技术有限公司 信号再生装置和方法
US20120070151A1 (en) * 2010-09-17 2012-03-22 Samsung Electronics Co., Ltd. Optical Modulators for Generating De-Emphasis Formatted Optical Signals and Methods of Operating the Same
CN107425405A (zh) * 2017-06-29 2017-12-01 武汉电信器件有限公司 一种可调谐半导体激光器
CN109477938A (zh) * 2016-06-02 2019-03-15 麻省理工学院 用于光学神经网络的设备和方法

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0980123A2 (en) * 1998-08-11 2000-02-16 TRW Inc. High average power fiber laser system with phase front control
US20070201036A1 (en) * 2006-02-27 2007-08-30 The Regents Of The University Of California Coherent hybrid electromagnetic field imaging
CN101667983A (zh) * 2009-09-16 2010-03-10 华为技术有限公司 调制信号的产生方法和传输设备
US20120070151A1 (en) * 2010-09-17 2012-03-22 Samsung Electronics Co., Ltd. Optical Modulators for Generating De-Emphasis Formatted Optical Signals and Methods of Operating the Same
CN102130739A (zh) * 2010-12-31 2011-07-20 华为技术有限公司 信号再生装置和方法
CN109477938A (zh) * 2016-06-02 2019-03-15 麻省理工学院 用于光学神经网络的设备和方法
CN107425405A (zh) * 2017-06-29 2017-12-01 武汉电信器件有限公司 一种可调谐半导体激光器

Similar Documents

Publication Publication Date Title
Amiri et al. Spatial continuous wave laser and spatiotemporal VCSEL for high-speed long haul optical wireless communication channels
Al-Qadasi et al. Scaling up silicon photonic-based accelerators: Challenges and opportunities
CN112232503B (zh) 计算装置、计算方法以及计算系统
CN113805641B (zh) 一种光子神经网络
JP7391215B2 (ja) 光計算装置および光信号処理方法
US6882758B2 (en) Current tuned Mach-Zehnder optical attenuator
CN110138455A (zh) 一种基于光谱处理的微波光子射频移相器及其移相方法
Maji et al. Numerical simulations of an all-optical parity generator and checker utilizing a reflective semiconductor optical amplifier at 200 Gbps
JP6957354B2 (ja) ディザなしのバイアス制御
CN112101540A (zh) 光学神经网络芯片及其计算方法
US8164819B2 (en) Photonic technique for generating arbitrary waveforms
WO2021083348A1 (zh) 一种光计算设备、光运算方法以及计算系统
CN112783260A (zh) 一种光计算设备、光运算方法以及计算系统
KR20230042333A (ko) 광자 프로세서에서 광자 자유도를 이용하기 위한 시스템들 및 방법들
CN210745163U (zh) 一种量子通信时间相位编码装置及密钥分发系统
Al-Qadasi et al. Scaling up silicon photonic-based accelerators: challenges and opportunities, and roadmapping with silicon photonics 2.0
US20230288638A1 (en) Multi-directional adaptive optical device
CN114966197A (zh) 基于受激布里渊散射效应的瞬时微波频率测量装置和方法
Hutchinson et al. Power photodiodes for high dynamic range photonic links
CN113541806A (zh) 基于并联马赫曾德尔调制器的16倍频毫米波信号产生装置及方法
CN112232487A (zh) 光学神经网络芯片及其计算方法
US20230236621A1 (en) Photonic computing system
CN116739063A (zh) 一种基于多模干涉仪以及相干检测的神经网络加速器
Belkin et al. Modeling Electroabsorption Modulated Laser for Digital Optical Transmission
CN116318393B (zh) 一种马赫曾德尔调制器的功率监测装置及系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20882934

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20882934

Country of ref document: EP

Kind code of ref document: A1