Optical coherent computing device and error correction method thereof

Abstract

The present invention relates to the technical field of data calculation, and provides an optical coherent calculation device and an error correction method thereof, wherein the optical coherent calculation device includes a laser device, a first beam splitter, a second beam splitter, a third beam splitter, a first modulator, a second modulator, a third modulator, a first converter, a second converter, a first coupler, a second coupler, an injector, a zero slip frequency detector, and an arithmetic control module. The use of the optical coherent calculation device and the error correction method thereof can effectively improve the calculation speed and calculation accuracy in the process of solving calculation problems to be processed by the optical coherent calculation device.

Description

[Technical Field]

[0001] This application relates to the field of data processing technology, and more particularly to an optical coherent computing device and an error correction method thereof. [Background technology]

[0002] In the field of data computation, it is often necessary to perform several complex computational problems using optical coherent computing devices. A coherent isin machine is a common type of optical coherent computing device.

[0003] A coherent isin machine is a physical solver based on degenerate optical parametric oscillation. It can effectively solve combinatorial optimization problems by mapping combinatorial optimization problems to an isin model and searching for the ground state energy to find the optimal solution.

[0004] However, in conventional technology, coherent isin machines have amplitude heterogeneity problems during calculations, which can cause mapping errors. [Overview of the Initiative] [Problems that the invention aims to solve]

[0005] In view of this, the present invention provides an optical coherent computing device and an error correction method thereof that can effectively improve the computing speed and accuracy in the process of solving the computing problem to be processed by the optical coherent computing device. [Means for solving the problem]

[0006] The technical proposal of this invention is specifically implemented as follows.

[0007] An optical coherent computing device comprising: a laser device, a first beam splitter, a second beam splitter, a third beam splitter, a first modulator, a second modulator, a third modulator, a first converter, a second converter, a first coupler, a second coupler, an injector, and a zerodifference It includes a frequency detector and an arithmetic control module, The output end of the laser device is connected to the input end of the first beam splitter, The output ends of the first beam splitter are respectively connected to the input ends of the first modulator and the second modulator, The output end of the first modulator is connected to the input end of the first converter, The output end of the first converter is connected to the input end of the first combiner, The output end of the first combiner is connected to the input end of the second converter, The output end of the second converter is connected to the input end of the third beam splitter, The output ends of the third beam splitter are respectively connected to the input ends of the zero difference frequency detector and the second combiner, The output end of the second modulator is connected to the input end of the second beam splitter, The output ends of the second beam splitter are respectively connected to the input ends of the zero difference frequency detector and the third modulator, The zero difference The output end of the frequency detector is connected to the input end of the arithmetic control module, The output ends of the arithmetic control module are respectively connected to the input ends of the third modulator and the injector, The output end of the third modulator is connected to the input end of the second combiner, The output end of the second combiner is connected to the input end of the injector, The output end of the injector is connected to the input end of the first combiner, The first combiner, the second converter, the third beam splitter, the second combiner and the injector are sequentially connected to the ring resonator via optical fibers.

[0008] Optionally, the laser device is used to output a pulsed laser having a first wavelength, The first beam splitter is used to split the pulsed laser output by the laser device into two pulsed lasers and output them to the first modulator and the second modulator, respectively. The first modulator is used to periodically output to the first converter a group of first timing pulses, each containing n pulse lasers (where n is an integer greater than 1), based on the received pulse lasers. The first converter is used to convert a received pulsed laser having a first wavelength into a pulsed laser having a second wavelength, and to input it to the second converter in the ring resonator via the first coupler. The second converter is used to convert the received pulsed laser having a second wavelength into a pulsed laser having a first wavelength and transmit it to the third beam splitter. The third beam splitter splits each received pulsed laser into two pulsed lasers, outputs one pulsed laser to the second coupler, and outputs the other pulsed laser to the zero difference Used to output to a frequency detector, The second modulator is used to periodically output a second group of timing pulses, each containing n pulse lasers, to the second beam splitter based on the received pulse laser, wherein a predetermined first time interval is set between the time the second modulator outputs the second group of timing pulses and the time the first modulator outputs the first group of timing pulses. The second beam splitter splits each received pulsed laser into two pulsed lasers, outputs one pulsed laser to the third modulator, and outputs the other pulsed laser to the zero difference Used to output to a frequency detector, The aforementioned zero difference The frequency detector balances the two received pulsed lasers to zero. difference It is used to perform frequency measurement and transmit an electrical signal to the calculation control module based on the measurement results. The aforementioned arithmetic control module is used to calculate the current pulse spin symbol and current isin energy based on the received electrical signal and a preset constraint matrix, to transmit a modulation command corresponding to the third modulator based on the set dissipation strength and coupling strength, and to transmit an injection command corresponding to the injector. The third modulator modulates the intensity and / or phase of the pulsed laser that has passed through based on the modulation command, outputs the modulated pulsed laser to the second coupler, and is used to input the second coupler to the ring resonator. The second coupler is used to output the received pulsed laser to the injector. The injector is used to perform a mutual injection operation on the received first timing pulse group and second timing pulse group based on an injection command, and to output the first timing pulse group and second timing pulse group after the mutual injection operation is completed to the second converter via the first coupler, with a predetermined first time interval between the time the injector outputs the second timing pulse group and the time the injector outputs the first timing pulse group.

[0009] Selectively, the optical coherent computing device is The optical coherent arithmetic unit further includes a timer for sending a stop command to the arithmetic control module if the operating time of the optical coherent arithmetic unit is greater than or equal to a preset time threshold, The aforementioned calculation control module is used to output the current optimal calculation result as the final calculation result based on a stop command.

[0010] Selectively, the injector is A delay module for performing a corresponding delay on each received pulse laser based on a pre-set delay policy, and for simultaneously transmitting the first pulse in the first timing pulse group and the first pulse in the second timing pulse group to the two input terminals of the injection module, respectively. The injection module for performing mutual injection operations on corresponding pulsed lasers in a first timing pulse group and a second timing pulse group based on an injection command, The combiner module includes a preset delay policy which applies a corresponding delay to each received pulsed laser, sequentially outputs each received pulsed laser to a first combiner according to a preset first time interval and transmission order, and forms a first timing pulse group and a second timing pulse group after injection is complete, with a preset first time interval still in place.

[0011] Selectively, the third modulator is An intensity modulator for modulating the intensity of a corresponding pulsed laser based on a modulation command, The system includes a phase modulator for modulating the phase of a corresponding pulsed laser based on a modulation command.

[0012] Selectively, the first converter is a second harmonic generator.

[0013] Selectively, both the first and second transducers are lithium niobate crystals with periodic polarization reversal.

[0014] Selectively, the arithmetic control module is a field-programmable gate array.

[0015] This application relates to an error correction method for an optical coherent computing device, Step A of setting up the optical coherent arithmetic device described in any one of the above embodiments, Step B involves pre-setting constraint matrices in the arithmetic control module of the optical coherent arithmetic device based on the arithmetic problem to be processed, Step C involves starting the optical coherent computing device and periodically outputting a pulsed laser having a first wavelength to the first beam splitter from the laser device in the optical coherent computing device, Step D involves splitting each pulsed laser received by the first beam splitter into two pulsed lasers and outputting them to the first modulator and the second modulator, respectively. Step E involves periodically outputting a first timing pulse group and a second timing pulse group, each having n pulse lasers (where n is an integer greater than 1), with a predetermined first time interval between them, based on the pulse lasers received by the first modulator and the second modulator. Step F involves converting the wavelength of the first timing pulse group output by the first modulator using the first converter, inputting it to the ring resonator via the first coupler, converting the wavelength again using the second converter, and then transmitting it to the third beam splitter. The third beam splitter splits each received pulsed laser into two pulsed lasers, and one of the pulsed lasers is set to zero. difference Step G involves outputting to a frequency detector, continuously outputting the other pulsed laser to a second coupler via a ring resonator, and continuously outputting to an injector via a ring resonator, The second group of timing pulses output by the second modulator is transmitted to the second beam splitter, which splits each received pulse laser into two pulse lasers, each of which is then connected to the third modulator and zero. difference Step H transmitted to the frequency detector, zero difference The frequency detector balances the lasers based on the two pulsed lasers it receives. difference Step I involves measuring the frequency and transmitting an electrical signal to the calculation control module based on the measurement results. Step J involves the arithmetic control module calculating the current pulse spin symbol and current isin energy based on the received electrical signal and a preset constraint matrix, setting the current isin energy as the energy optimal value, setting the current pulse spin symbol and current isin energy as the current optimal calculation result, and transmitting a modulation command corresponding to the third modulator and an injection command corresponding to the injector based on the preset initial values ​​of the dissipation strength and coupling strength. Step K involves the third modulator modulating the intensity and / or phase of the pulsed laser that has passed through it based on the modulation command, outputting the modulated pulsed laser to the second coupler, inputting it to the ring resonator via the second coupler, and outputting it to the injector via the ring resonator. Step L involves the injector performing a mutual injection operation on the received first timing pulse group and second timing pulse group based on the injection command and outputting the result, and then sequentially transmitting the first timing pulse group and second timing pulse group after the mutual injection operation is completed to the third beam splitter via the first coupler and second converter in the ring resonator. The third beam splitter splits each received timing pulse into two pulsed lasers; one pulsed laser is continuously transmitted in the ring resonator, and the other pulsed laser is zeroed out. difference Step M to be output to the frequency detector, zero difference The frequency detector balances the lasers based on the two pulsed lasers received from the two input terminals. difference Step N involves performing frequency measurement and transmitting an electrical signal to a calculation control module based on the measurement results. Step O: The arithmetic control module calculates the current pulse spin symbol and current isin energy based on the electrical signal it receives and a pre-set constraint matrix. Step P determines whether the current isin energy Ei is less than the current optimal energy value. If so, step Q is performed; otherwise, step R is performed. The calculation control module sets the current isin energy as the energy optimal value, sets the current pulse spin symbol and current isin energy as the current optimal calculation result, updates the dissipation intensity and coupling intensity values ​​based on preset adjustment values, sends a modulation command corresponding to the third modulator and an injection command corresponding to the injector based on the updated dissipation intensity and coupling intensity values, returns to step K and executes step Q, The present invention further provides an error correction method for an optical coherent arithmetic unit, which includes a arithmetic control module resetting the values ​​of dissipation intensity and coupling intensity, sending a modulation command corresponding to the third modulator and an injection command corresponding to the injector based on the reset values ​​of dissipation intensity and coupling intensity, and then returning to step K and executing step R.

[0016] Selectively, before returning to step K and executing, the method To determine whether the operating time of the optical coherent computing device is less than a preset time threshold, If so, the process returns to step K and executes; otherwise, the arithmetic control module outputs the current optimal arithmetic result as the final arithmetic result.

[0017] Selectively, updating the values ​​of dissipation intensity and binding intensity based on the aforementioned preset adjustment values ​​is possible. This includes increasing the current dissipation intensity and coupling intensity values ​​by a predetermined adjustment value, and then obtaining the updated dissipation intensity and coupling intensity values.

[0018] Selectively, resetting the values ​​of dissipation strength and binding strength as described above is possible. Randomly selecting one value from a pre-defined first-value range as the current value of the dissipation intensity, This includes setting the initial value of the bond strength to the current value of the bond strength. [Effects of the Invention]

[0019] As can be seen from the above, in the optical coherent computing device and its error correction method of the present invention, a predetermined number of auxiliary error correction pulses that do not involve nonlinear gain are introduced into a laser device pulse network having a predetermined number of interconnections based on a degenerate optical parametric oscillator. These pulses are coupled with nonlinear gain pulses to form pulse neurons, and error correction of the computing process of the optical coherent computing device can be realized by the dynamic characteristics of the pulse neurons. The current isin energy and the magnitude of the current energy optimal value are determined, and subsequent operations are performed based on the determination result to search for the ground state energy, thereby realizing the search for the optimal solution. Furthermore, even if the computing process falls into a fixed point of the local optimal value, the local optimal value can be made unstable by changing the parameter conditions, thereby obtaining a better global optimal solution. Thus, the problem of inaccurate mapping due to amplitude heterogeneity in conventional coherent isin machines is effectively solved, improving the computing performance of the optical coherent computing device and significantly improving the computing speed and accuracy in the process of solving the computing problem to be processed by the optical coherent computing device. [Brief explanation of the drawing]

[0020] [Figure 1] This is a schematic diagram of the structure of an optical coherent computing device in a specific embodiment of the present invention. [Figure 2] This is a schematic diagram of the structure of an injector in a specific embodiment of the present invention. [Figure 3] This is a flowchart of an error correction method for an optical coherent computing device in a specific embodiment of the present invention. [Modes for carrying out the invention]

[0021] To further clarify the technical concept and advantages of the present invention, the present invention will be described in more detail below with reference to the drawings and specific embodiments.

[0022] This invention proposes an optical coherent computing device and an error correction method thereof.

[0023] Figure 1 is a schematic diagram of the structure of an optical coherent computing device in a specific embodiment of the present invention. As shown in Figure 1, in this specific embodiment, the optical coherent computing device specifically comprises a laser device 101, a first beam splitter 102, a second beam splitter 103, a third beam splitter 104, a first modulator 105, a second modulator 106, a third modulator 107, a first converter 108, a second converter 109, a first coupler 110, a second coupler 111, an injector 112, and a zero difference It includes a frequency detector 113 and a calculation control module 114, The output terminal of the laser device 101 is connected to the input terminal of the first beam splitter 102. The output terminals of the first beam splitter 102 are connected to the input terminals of the first modulator 105 and the second modulator 106, respectively. The output terminal of the first modulator 105 is connected to the input terminal of the first converter 108. The output terminal of the first converter 108 is connected to the input terminal of the first coupler 110. The output terminal of the first coupler 110 is connected to the input terminal of the second converter 109. The output terminal of the second converter 109 is connected to the input terminal of the third beam splitter 104. The output terminals of the third beam splitter 104 are, respectively, zero difference It is connected to the input terminals of the frequency detector 113 and the second coupler 111, The output terminal of the second modulator 106 is connected to the input terminal of the second beam splitter 103. The output terminals of the second beam splitter 103 are, respectively, zero difference It is connected to the input terminals of the frequency detector 113 and the third modulator 107, The aforementioned zero difference The output terminal of the frequency detector 113 is connected to the input terminal of the calculation control module 114. The output terminals of the calculation control module 114 are connected to the input terminals of the third modulator 107 and the injector 112, respectively. The output terminal of the third modulator 107 is connected to the input terminal of the second coupler 111. The output terminal of the second coupler 111 is connected to the input terminal of the injector 112. The output terminal of the injector 112 is connected to the input terminal of the first coupler 110. The first coupler 110, the second transducer 109, the third beam splitter 104, the second coupler 111, and the injector 112 are sequentially connected to the ring resonator 115 via optical fibers.

[0024] Furthermore, the optical coherent computing device described in this application can be realized in several specific implementation forms. The technical proposal of this application will be described in more detail below, using several specific implementation forms as examples.

[0025] For example, in one specific example of the present application, The laser device 101 is used to output a pulsed laser having a first wavelength. The first beam splitter 102 is used to split the pulsed laser output by the laser device 101 into two pulsed lasers and output them to the first modulator 105 and the second modulator 106, respectively. The first modulator 105 is used to periodically output a first timing pulse group containing n (where n is an integer greater than 1) pulse lasers to the first converter 108 based on the received pulse laser. The first converter 108 is used to convert the received pulsed laser having a first wavelength into a pulsed laser having a second wavelength and input it to the second converter in the ring resonator 115 via the first coupler 110. The second converter 109 is used to convert the received pulsed laser having a second wavelength into a pulsed laser having a first wavelength and transmit it to the third beam splitter 104. The third beam splitter 104 splits each received pulsed laser into two pulsed lasers, outputs one pulsed laser to the second coupler 111, and outputs the other pulsed laser to the zero difference Used to output to the frequency detector 113, The second modulator 106 is used to periodically output a second timing pulse group, each containing n pulse lasers, to the second beam splitter 103 based on the received pulse laser, and a predetermined first time interval is set between the time the second modulator 106 outputs the second timing pulse group and the time the first modulator 105 outputs the first timing pulse group. The second beam splitter 103 splits each received pulsed laser into two pulsed lasers, outputs one pulsed laser to the third modulator 107, and outputs the other pulsed laser to the zero difference Used to output to the frequency detector 113, The aforementioned zero difference The frequency detector 113 balances the two received pulsed lasers to zero. difference It is used to perform frequency measurement and transmit an electrical signal to the calculation control module 114 based on the measurement result. The calculation control module 114 calculates the current pulse spin symbol B based on the received electrical signal and the preset constraint matrix. i and current Ishin Energy E i Based on the set dissipation intensity β and coupling intensity ε, a modulation command corresponding to the third modulator 107 is transmitted, and an injection command corresponding to the injector 112 is transmitted. The third modulator 107 modulates the intensity and / or phase of the pulsed laser that has passed through it based on the modulation command, outputs the modulated pulsed laser to the second coupler 111, and is used to input it to the ring resonator 115 via the second coupler 111. The second coupler 111 is used to output the received pulsed laser to the injector 112. The injector 112 performs a mutual injection operation on the received first timing pulse group and second timing pulse group based on the injection command, and is used to output the first timing pulse group and second timing pulse group after the mutual injection operation is completed to the second converter 109 via the first coupler 110, with a predetermined first time interval between the time the injector 112 outputs the second timing pulse group and the time the injector 112 outputs the first timing pulse group.

[0026] For example, in one specific embodiment of this application, the laser device 101 may periodically output timing pulses of equal intensity, that is, at regular intervals (e.g., a predetermined time length), it may output pulsed lasers (which may be called timing pulses) of equal intensity (e.g., having a predetermined frequency and predetermined power) having a predetermined first wavelength (e.g., 1560 nanometers). For example, the laser device 101 may output pulsed lasers with equivalent intensity and a wavelength of 1560 nanometers at output periods (e.g., 1 millisecond).

[0027] The pulsed laser output by the laser device 101 may be transmitted to the first beam splitter 102. The first beam splitter 102 may split each received pulsed laser into two pulsed lasers according to a preset ratio and output them to the first modulator 105 and the second modulator 106, respectively.

[0028] Furthermore, as an example, in one specific embodiment of this application, both the first modulator and the second modulator may be intensity modulators (IMs) capable of modulating the intensity of each received pulsed laser. For example, the intensity of some unwanted (or non-permitting) pulsed lasers may be modulated to zero without changing the intensity of some necessary (or permitting) pulsed lasers.

[0029] Therefore, both the first and second modulators output n timing pulses at regular intervals (for example, a preset output period) based on the received pulsed laser. These n timing pulses are called a timing pulse group, and thus the timing pulse group can be output periodically. Furthermore, by controlling the time at which the two modulators output the timing pulse group, a preset first time interval can be created between the two timing pulse groups output by the two modulators.

[0030] For example, assuming that the output periods of both the first and second modulators are 100 milliseconds, both the first and second modulators may output one group of timing pulses every 100 milliseconds, and each group of timing pulses may contain five pulsed lasers (which may also be called timing pulses), with a time interval of 1 millisecond between each pulsed laser.

[0031] Assuming that the time origin of any one output period of the first modulator is t0, the first modulator outputs one timing pulse every millisecond starting from t0 (i.e., the first modulator does not change the intensity of the received pulsed laser at this time), outputs a total of five timing pulses, forming one first timing pulse group (x1, x2, ... x5), and does not output any timing pulses for (t0+5) to (t0+100) milliseconds (i.e., during this time, the first modulator modulates the intensity of all received pulsed lasers to 0).

[0032] The second modulator starts at (t0+10) milliseconds, outputs one timing pulse every millisecond, for a total of five timing pulses, forming one second timing pulse group (k1, k2, ..., k5), and does not output any timing pulses between (t0+15) and (t0+100) milliseconds.

[0033] Therefore, the first modulator and the second modulator can be used to stagger the output times of the two timing pulse groups so that the first timing pulse group and the second timing pulse group can be transmitted in the ring resonator with a predetermined first time interval (for example, 5 milliseconds) between them in subsequent operations.

[0034] The first modulator 105 outputs the first timing pulse group to the first converter 108, which converts all pulse lasers having a first wavelength (e.g., 1560 nanometers) in the first timing pulse group into pulse lasers having a second wavelength (e.g., 780 nanometers), and inputs them to the ring resonator 115 via the first coupler 110, where they are transmitted to the second converter 109.

[0035] The second converter 109 converts the received pulsed laser having a second wavelength (e.g., 780 nanometers) into a pulsed laser having a first wavelength (e.g., 1560 nanometers) and transmits it to the third beam splitter 104 in the ring resonator 115.

[0036] The third beam splitter 104 splits each received pulsed laser into two pulsed lasers according to a preset ratio, and sets one of the pulsed lasers to zero. difference The other pulsed laser is output to the frequency detector 113, and then continuously output to the second coupler 111 via the ring resonator 115, resulting in the first timing pulse group (x1, x2, ... x) shown in Figure 1. n ) is formed and continuously output to the injector 112 via the ring resonator 115.

[0037] Furthermore, the second modulator 106 outputs the generated second timing pulse group to the second beam splitter 103. The second beam splitter 103 splits each received pulsed laser into two pulsed lasers according to a preset ratio, outputting one pulsed laser to the third modulator 107 and the other pulsed laser to zero. difference The output is sent to the frequency detector 113.

[0038] zero difference The frequency detector 113 receives two pulsed lasers from the two input terminals and then balances the two received pulsed lasers to zero. difference Frequency measurement can be performed, and an electrical signal can be transmitted to the arithmetic control module 114 based on the measurement results.

[0039] After receiving an electrical signal, the arithmetic control module 114 calculates the current pulse spin symbol B based on the received electrical signal and a preset constraint matrix. i (i.e., the i-th pulse spin symbol) and the current isin energy E i (i.e., the i-th isin energy) is obtained, and based on the set dissipation intensity β and coupling intensity ε, a modulation command corresponding to the third modulator 107 can be transmitted, and an injection command corresponding to the injector 112 can be transmitted.

[0040] The third modulator 107 modulates the intensity and / or phase of the pulsed laser that has passed through it based on the modulation command, outputs the modulated pulsed laser to the second coupler 111, and inputs it to the ring resonator 115 via the second coupler 111, generating the second timing pulse group (k1, k2, ...k) shown in Figure 1. n ) is formed and continuously output to the injector 112 via the ring resonator 115.

[0041] Because a predetermined first time interval (for example, 5 milliseconds) is set between the time when the second modulator 106 outputs the second timing pulse group and the time when the first modulator 105 outputs the first timing pulse group, the first timing pulse group (x1, x2, ... x) transmitted in the ring resonator 115 n ) is the second timing pulse group (k1, k2, ...k n The pulses reach the injector 112 before the pulses reach the injector 112, and the above-described preset first time interval is maintained between the times of the two timing pulse groups.

[0042] After receiving the first timing pulse group and the second timing pulse group, the injector 112 performs an inter-injection operation on the received first timing pulse group and second timing pulse group based on an injection command, and outputs the first timing pulse group and the second timing pulse group after the inter-injection operation is completed to the second converter 109 via the first coupler 110. A preset first time interval may still be provided between the time when the injector 112 outputs the second timing pulse group and the time when it outputs the first timing pulse group.

[0043] After the injection is completed, the two timing pulse groups are continuously transmitted in the ring resonator 115 and re-transmitted to the third beam splitter 104 via the first coupler 110 and the second converter 109.

[0044] The third beam splitter 104 divides all the received timing pulses into two pulse lasers according to a preset ratio, continuously transmits one pulse laser in the ring resonator 115, and sets the other pulse laser to zero difference and outputs it to the zero frequency detector 113.

[0045] zero difference Based on the two pulse lasers received from the two input ends, the zero frequency detector 113 performs a balance zero difference frequency measurement, and transmits an electrical signal to the arithmetic control module 114 based on the measurement result.

[0046] Also, as an example, in one specific embodiment of the present application, after receiving the electrical signal, the arithmetic control module 114 calculates the current pulse spin symbol B i and the current ising energy E i based on the received electrical signal and a preset constraint matrix, and may determine the magnitudes of the current ising energy E i and the current energy optimum value E0.

[0047] When the current E i is smaller than the current energy optimum value E0, the arithmetic control module uses the current Ei Let E0 be the optimal energy value, and the current pulse spin symbol B i and current Ishin Energy E i This is the current optimal calculation result, and the values ​​of dissipation intensity β and coupling intensity ε are updated based on a preset adjustment value (for example, if the preset adjustment value is 0.1, β i+1 =β i +0.1, ε i+1 =ε i +0.1, β i and ε i These are the current i-th dissipation intensity and binding intensity values, respectively, and β i+1 and ε i+1 (wherein these are the (i+1)th dissipation intensity β and coupling intensity ε values ​​after the update, respectively), and based on the updated dissipation intensity β and coupling intensity ε values, a modulation command corresponding to the third modulator and an injection command corresponding to the injector may be transmitted.

[0048] Current E i If the current energy is greater than or equal to the optimal energy value E0, the arithmetic control module may reset the values ​​of the dissipation intensity β and the coupling intensity ε (for example, by randomly selecting one value from a preset first range as the current value of the dissipation intensity β, and setting the initial value of the coupling intensity ε0 as the current value of the coupling intensity ε), and based on the reset values ​​of the dissipation intensity β and the coupling intensity ε, send a modulation command corresponding to the third modulator and an injection command corresponding to the injector.

[0049] The third modulator 107 modulates the intensity and / or phase of the pulsed laser that has passed through it based on the modulation command, outputs the modulated pulsed laser to the second coupler 111, inputs it to the ring resonator 115 via the second coupler 111, and continues to outputs it to the injector 112 via the ring resonator 115.

[0050] Based on the injection command, the injector 112 performs a mutual injection operation on the received first timing pulse group and second timing pulse group, and outputs the first timing pulse group and second timing pulse group after the mutual injection operation is completed to the second converter 109 via the first coupler 110. A predetermined first time interval is still maintained between the time when the injector 112 outputs the second timing pulse group and the time when it outputs the first timing pulse group.

[0051] The two timing pulse groups after injection are continuously transmitted in the ring resonator 115 to continue the next round cycle, but this will not be explained further here.

[0052] The above-described optical coherent computing device performs the above operations, maps the computational problems to be processed in the computation process (e.g., combinatorial optimization problems, I-SIN problems, etc.) to the corresponding I-SIN model, searches for the ground state energy to realize the search for the optimal solution, effectively solves the computational problems to be processed, and solves the mapping error problem due to amplitude heterogeneity in conventional coherent I-SIN machines.

[0053] In order to use the above structure in the optical coherent computing device of this application, a predetermined number of auxiliary error correction pulses that do not involve nonlinear gain are added to a laser device pulse network having a predetermined number of mutual couplings based on a degenerate optical parametric oscillator, i.e., pulse lasers (k1, k2, ...k) in the second timing pulse group. n In order to introduce a nonlinear gain pulse, i.e., a pulsed laser in the first timing pulse group (x1, x2, ... x n ) combines with to form pulse neurons, and the dynamic characteristics of the pulse neurons enable error correction of the computation process of the optical coherent computing device. Current isin energy E iBy determining the magnitude of the current optimal energy value E0 and performing subsequent operations based on the determination result, the optimal solution (i.e., the optimal energy value E0) can be found by searching for the ground state energy. Furthermore, even if the calculation process falls into a fixed point of the local optimal value, the local optimal value can be made unstable by changing the parameter conditions (for example, updating or resetting the values ​​of the dissipation intensity β and coupling intensity ε), thereby obtaining a better global optimal solution (i.e., the optimal calculation result). This effectively improves the calculation performance of the optical coherent computing device and enhances the calculation speed and accuracy in the process of solving the calculation problem that the optical coherent computing device must process.

[0054] Furthermore, as an example, in one specific embodiment of this application, the optical coherent computing device is If the operating time of the optical coherent arithmetic unit exceeds a preset time threshold, the system further includes a timer (not shown in Figure 1) for sending a stop command to the arithmetic control module. Based on the stop command, the calculation control module determines the current optimal calculation result (i.e., the corresponding pulse spin symbol B of the current optimal energy value E0) i and Ishin Energy E i This is used to output the final calculation result.

[0055] Furthermore, as an example, in one specific embodiment of this application, the timer may be provided independently and connected to the arithmetic control module, or it may be directly integrated into or provided in the arithmetic control module.

[0056] Therefore, with the timer described above, the optical coherent arithmetic unit can automatically output the final calculation result without manual intervention once it has operated for a preset time period.

[0057] Furthermore, once the final calculation result is output, the entire process can be terminated and the optical coherent calculation device can be turned off.

[0058] Furthermore, as an example, as shown in Figure 2, in one specific embodiment of this application, the injector 112 may include a delay module 201, an injection module 202, and a combiner module 203.

[0059] The delay module 201 performs a corresponding delay on each received pulse laser based on a preset delay policy, and the first timing pulse group (x1, x2, ... x n ) The first pulse and the second timing pulse group (k1, k2, ...k n This is used to simultaneously transmit the first pulse in ) to the two input terminals of the injection module 202, respectively.

[0060] The injection module 202 is used to perform mutual injection operations on the corresponding pulsed lasers in the first timing pulse group and the second timing pulse group based on an injection command. For example, a portion of x1 (a preset ratio, e.g., 10% of pulses) may be injected into k1, a portion of k1 (a preset ratio, e.g., 10% of pulses) may be injected into x1, and so on. The ratio of mutual injection between the two pulses can be controlled by the injection command. Through the above mutual injection, each pulse neuron x i k i It is possible to form this.

[0061] The combiner module 203 applies a corresponding delay to each received pulsed laser based on a preset delay policy, and sequentially outputs each received pulsed laser to the first coupler according to a preset first time interval and transmission order, thereby forming a first timing pulse group and a second timing pulse group after injection completion, with a preset first time interval still in place. For example, the second timing pulse group after injection completion is still transmitted after the first timing pulse group.

[0062] Furthermore, as an example, in one specific embodiment of this application, the third modulator 107 may include an intensity modulator (IM) and a phase modulator (PM).

[0063] The intensity modulator is used to modulate the intensity of the corresponding pulsed laser based on a modulation command. The phase modulator is used to modulate the phase of a corresponding pulsed laser based on a modulation command.

[0064] Furthermore, as an example, in one specific embodiment of this application, the first converter may be a second harmonic generator (SHG).

[0065] Furthermore, as an example, in one specific embodiment of this application, the first and second transducers may both be periodically polled lithium niobate crystals (PPLN), or other suitable transducers.

[0066] Furthermore, as an example, in one specific embodiment of this application, the calculation control module may use a field-programmable gate array (FPGA) so that corresponding calculations and controls can be performed by a program.

[0067] Furthermore, as an example, in one specific embodiment of this application, the first wavelength may be 1560 nanometers, and the second wavelength may be 780 nanometers.

[0068] Furthermore, in the proposed technology of this application, based on the above-mentioned optical coherent computing device, this application further proposes an error correction method for the optical coherent computing device.

[0069] Figure 3 is a flowchart of the error correction method for an optical coherent computing device in a specific embodiment of the present invention. As shown in Figure 3, in this specific embodiment, the error correction method for the optical coherent computing device includes the following steps.

[0070] Step 301: Set up the optical coherent computing device.

[0071] In the proposed technical version of the present invention, the optical coherent computing device shown in Figure 1 may be pre-configured based on the computational problem to be processed (for example, a combinatorial optimization problem, an ISIS problem, etc.).

[0072] Step 302: Based on the computational problem to be processed, constraint matrices are pre-set in the computation control module of the optical coherent computation device.

[0073] In the proposed technical version of the present invention, constraint matrices corresponding to the arithmetic control module of the optical coherent arithmetic device may be set in advance based on the arithmetic problem to be processed (for example, a combinatorial optimization problem, an ISIS problem, etc.).

[0074] For example, in one specific embodiment of this application, a corresponding pulse neuron may be provided for each node in the computation problem to be processed, and a corresponding value may be set for each matrix element in the constraint matrix based on the interaction strength and / or connection relationship between each pulse neuron.

[0075] For example, in one specific embodiment of this application, if there is an interaction between the i-th pulse neuron and the j-th pulse neuron, the matrix element J in the constraint matrix ij The corresponding value may be taken as the interaction strength between the i-th pulse neuron and the j-th pulse neuron, and this can be sequentially inferred.

[0076] Of course, in the proposed technical version of this invention, other values ​​may be set for each matrix element in the constraint matrix using other methods as needed for the actual application scenario, and specific setting methods will not be explained here one by one.

[0077] Step 303: The optical coherent computing device is started, and the laser device in the optical coherent computing device periodically outputs a pulsed laser having the first wavelength to the first beam splitter.

[0078] Step 304, the first beam splitter splits each received pulsed laser into two pulsed lasers and outputs them to the first modulator and the second modulator, respectively.

[0079] In this step, upon receiving a pulsed laser output by the laser device, the first beam splitter may split each received pulsed laser into two pulsed lasers according to a preset ratio and output them to the first modulator and the second modulator, respectively.

[0080] Furthermore, in the proposed technology of this application, the value of the above ratio can be set in advance according to the requirements of the actual application scenario. For example, in one specific embodiment of this application, the set ratio may be 1:1, that is, it may be divided into two pulsed lasers on average. Of course, the set ratio may be any other appropriate value, which are not listed here.

[0081] Step 305, the first modulator and the second modulator periodically output a first timing pulse group and a second timing pulse group, each having n pulse lasers and spaced apart by a predetermined first time interval, based on the received pulse laser.

[0082] Furthermore, in the proposed technology of this application, the value of n can be set in advance according to the requirements of the actual application scenario. For example, in one specific embodiment of this application, the value of n may be 3, 4, 5, 10, or 15, or any other appropriate value, which is not listed here.

[0083] Step 306: The wavelength of the first timing pulse group output by the first modulator is converted by the first converter, input to the ring resonator via the first coupler, and then the wavelength is converted by the second converter before being transmitted to the third beam splitter.

[0084] Step 307, the third beam splitter splits each received pulsed laser into two pulsed lasers, and zeros out one of the pulsed lasers. difference The pulsed laser is output to a frequency detector, and the other pulsed laser is continuously output to a second coupler via a ring resonator, and then continuously output to an injector via a ring resonator.

[0085] In the proposed technology of this application, the third beam splitter splits each received pulsed laser into two pulsed lasers according to a preset ratio, and zero difference The output may be sent to the frequency detector and the injector separately.

[0086] Furthermore, in the proposed technology of this application, the value of the above ratio can be set in advance according to the requirements of the actual application scene. For example, in one specific embodiment of this application, the set ratio may be 1:9, that is, 10% pulsed laser for measurement is zeroed difference The frequency detector outputs a pulsed laser signal, and a 90% pulsed laser signal is continuously output to the injector via a ring resonator. Of course, this preset ratio may be any other appropriate value, which are not listed here.

[0087] Step 308: The second timing pulse group output by the second modulator is transmitted to the second beam splitter, which splits each received pulsed laser into two pulsed lasers, each connected to the third modulator and zero. difference The signal is transmitted to the frequency detector.

[0088] In the proposed technology of this application, after receiving a pulsed laser, the second beam splitter splits each received pulsed laser into two pulsed lasers according to a preset ratio, outputs one pulsed laser to the third modulator, and zeros the other pulsed laser. difference Output to the frequency detector.

[0089] Furthermore, in the proposed technology of this application, the value of the above ratio can be set in advance according to the requirements of the actual application scenario. For example, in one specific embodiment of this application, the set ratio may be 1:1, that is, it may be divided into two pulsed lasers on average. Of course, the set ratio may be any other appropriate value, which are not listed here.

[0090] Step 309, Zero difference The frequency detector balances the two received pulsed lasers to zero. difference The system performs frequency measurements and transmits electrical signals to the arithmetic control module based on the measurement results.

[0091] In step 310, the arithmetic control module calculates the current pulse spin symbol and current isin energy based on the received electrical signal and the preset constraint matrix, sets the current isin energy as the energy optimal value, sets the current pulse spin symbol and current isin energy as the current optimal calculation result, and transmits a modulation command corresponding to the third modulator and an injection command corresponding to the injector based on the preset initial values ​​of the dissipation strength and coupling strength.

[0092] In the proposed technology of this application, the computation control module, upon first receiving an electrical signal, calculates the current pulse spin symbol B based on the received electrical signal and a preset constraint matrix. i (For example, B1 represents the first pulse spin symbol) and the current isin energy E i (For example, E1 represents the first isin energy) and obtain the current isin energy E iLet E0 be the optimal energy value, and the current pulse spin symbol B i and current Ishin Energy E i Using this as the current optimal calculation result, and based on the pre-set initial values ​​of the dissipation intensity β and coupling intensity ε (for example, pre-setting the initial value of the dissipation intensity β to β0 and the initial value of the coupling intensity ε to ε0), a modulation command corresponding to the third modulator and an injection command corresponding to the injector can be transmitted.

[0093] Step 311, the third modulator modulates the intensity and / or phase of the pulsed laser that has passed through it based on the modulation command, outputs the modulated pulsed laser to the second coupler, inputs it to the ring resonator via the second coupler, and outputs it to the injector via the ring resonator.

[0094] In the proposed technology of this application, the arithmetic control module transmits a modulation command based on the corresponding dissipation intensity β and coupling intensity ε, and the modulation command may carry information related to the dissipation intensity β and coupling intensity ε. Since the dissipation intensity β may usually relate to the intensity of the pulsed laser and the coupling intensity ε may usually relate to the phase of the pulsed laser, the third modulator can modulate the intensity and / or phase of the pulsed laser that has passed through based on the received modulation command, and output the modulated pulsed laser to the second coupler, input it to the ring resonator via the second coupler, and output it to the injector via the ring resonator.

[0095] In step 312, the injector performs a mutual injection operation on the received first timing pulse group and second timing pulse group based on the injection command and outputs the result. After the mutual injection operation is completed, the first timing pulse group and second timing pulse group are sequentially transmitted to the third beam splitter via the first coupler and second transducer in the ring resonator.

[0096] In the proposed technology of this application, the arithmetic control module transmits an injection command based on the corresponding dissipation intensity β and coupling intensity ε, and the injection command may carry information related to the dissipation intensity β and coupling intensity ε. Since the dissipation intensity β may relate to the intensity of normally injected pulsed lasers, and the coupling intensity ε may relate to the ratio of normally injected pulsed lasers, the injector can perform mutual injection operations on the received first timing pulse group and second timing pulse group and output based on the received injection command.

[0097] Step 313, the third beam splitter splits each received timing pulse into two pulsed lasers, one pulsed laser is transmitted continuously in the ring resonator, and the other pulsed laser is zeroed out. difference Output to the frequency detector.

[0098] In the proposed technology of this application, the third beam splitter splits each received pulsed laser into two pulsed lasers according to a preset ratio, and zero difference The signal can be output to both the frequency detector and the injector.

[0099] Step 314, Zero difference The frequency detector balances zero based on the two pulsed lasers received from the two input terminals. difference The system performs frequency measurements and transmits electrical signals to the arithmetic control module based on the measurement results.

[0100] Step 315, the arithmetic control module calculates the current pulse spin symbol and current isin energy based on the received electrical signal and the preset constraint matrix.

[0101] Step 316: Determine whether the current isin energy Ei is less than the current energy optimal value. If so, perform step 317; otherwise, perform step 318.

[0102] In step 317, the arithmetic control module sets the current isin energy as the energy optimal value, the current pulse spin symbol and current isin energy as the current optimal calculation result, updates the dissipation intensity and coupling intensity values ​​based on preset adjustment values, sends a modulation command corresponding to the third modulator and an injection command corresponding to the injector based on the updated dissipation intensity and coupling intensity values, and returns to step 311 to execute.

[0103] In this step, the current isin energy E i Since is smaller than the current optimal energy value E0, the current isin energy E i Since this indicates that is a better solution, the arithmetic control module is now E i Set the optimal energy value E0, update the optimal energy value E0, update the optimal calculation result with the current pulse spin symbol and current isin energy as the current optimal calculation result, update the values ​​of dissipation intensity β and coupling intensity ε, send a modulation command corresponding to the third modulator and an injection command corresponding to the injector based on the updated values ​​of dissipation intensity β and coupling intensity ε, return to step 311 and start the cycle operation of the next round.

[0104] Furthermore, in the proposed technology of this application, the values ​​of dissipation intensity β and coupling intensity ε can be realized by multiple specific implementations based on pre-set adjustment values.

[0105] For example, in one specific embodiment of this application, updating the values ​​of dissipation strength and bond strength based on the aforementioned preset adjustment values ​​is: The current dissipation intensity and coupling intensity values ​​may be increased by pre-set adjustment values, and then updated to these values.

[0106] For example, if the pre-set adjustment value is 0.1, then β i+1 =β i +0.1, ε i+1 =ε i It is +0.1. β i and ε iThese are the current i-th dissipation intensity and binding intensity values, respectively, and β i+1 and ε i+1 These are the values ​​of the (i+1)th dissipation intensity β and the binding intensity ε after the update, respectively.

[0107] In step 318, the arithmetic control module resets the dissipation intensity and coupling intensity values, and based on the reset dissipation intensity and coupling intensity values, sends a modulation command corresponding to the third modulator and an injection command corresponding to the injector, and returns to step 311 to execute.

[0108] In this step, the current isin energy E i Since the current energy is greater than or equal to the optimal energy value E0, the current energy E i Since this indicates that it is not a better solution, the values ​​of the dissipation intensity β and coupling intensity ε are directly reset without changing the current energy optimal value E0 and the current optimal calculation result, and based on the reset values ​​of the dissipation intensity β and coupling intensity ε, a modulation command corresponding to the third modulator is sent, an injection command corresponding to the injector is sent, and the process returns to step 311 to start the cycle operation for the next round.

[0109] Furthermore, in the proposed technology of this application, the values ​​of dissipation strength β and bonding strength ε can be reset in various ways.

[0110] For example, in one specific embodiment of this application, resetting the values ​​of dissipation strength and bonding strength is, One value is randomly selected from a pre-defined first-value range as the current value of the dissipation intensity β, This may include setting the initial value ε0 of the bond strength ε to the current value of the bond strength ε.

[0111] In the proposed technology of this application, the range of the above-mentioned first value range can be pre-set according to the requirements of the actual application scenario, so no further explanation is provided here.

[0112] Therefore, the error correction method of the optical coherent computing device described above allows the current isin energy E to be corrected. i By determining the magnitude of the current optimal energy value E0, subsequent operations can be performed based on the determination result, and the search for the ground state energy can be realized to find the optimal solution (i.e., the optimal energy value E0). Furthermore, even if the calculation process falls into a fixed point of the local optimal value, the local optimal value can be made unstable by changing the parameter conditions (for example, updating or resetting the values ​​of the dissipation intensity β and coupling intensity ε), thereby obtaining a better global optimal solution (i.e., the optimal calculation result), and thus effectively improving the calculation performance of the optical coherent computing device and improving the calculation speed and accuracy in the process of solving the calculation problem that the optical coherent computing device should process.

[0113] Furthermore, as an example, in one specific embodiment of this application, before returning to step 311, the error correction method of the optical coherent computing device is: A step of determining whether the operating time of the optical coherent computing device is less than a preset time threshold, If so, return to step 311 and execute; otherwise, the arithmetic control module executes the current optimal calculation result (i.e., the corresponding pulse spin symbol B of the current optimal energy value E0). i and Ishin Energy E i The process may further include the step of outputting ) as the final calculation result.

[0114] Therefore, the optical coherent computing device can automatically output the final calculation result after operating for a predetermined time period.

[0115] Furthermore, once the final calculation result is output, the entire process can be terminated and the optical coherent calculation device can be turned off.

[0116] As described above, in the technical proposal of the present invention, a laser device pulse network having a predetermined number of mutual couplings based on a degenerate optical parametric oscillator is provided with a predetermined number of auxiliary error correction pulses that do not involve nonlinear gain, i.e., pulse lasers (k1, k2, ...k) in the second timing pulse group. n In order to introduce a nonlinear gain pulse, i.e., a pulsed laser in the first timing pulse group (x1, x2, ... x n ) combines with to form pulse neurons, and the dynamic characteristics of the pulse neurons enable error correction of the computation process of the optical coherent computing device. Current isin energy E i By determining the magnitude of the current optimal energy value E0 and performing subsequent operations based on the determination result, the optimal solution (i.e., the optimal energy value E0) can be found by searching for the ground state energy. Furthermore, even if the calculation process falls into a fixed point of the local optimal value, the local optimal value can be made unstable by changing the parameter conditions (for example, updating or resetting the values ​​of the dissipation intensity β and coupling intensity ε), thereby obtaining a better global optimal solution (i.e., the optimal calculation result). Thus, this effectively solves the problem of inaccurate mapping due to amplitude heterogeneity in conventional coherent isin machines, improves the calculation performance of the optical coherent calculation device, and significantly improves the calculation speed and accuracy in the solution process of the calculation problem to be processed by the optical coherent calculation device (especially when the problem is complex).

[0117] The above are merely preferred embodiments of the present invention and are not intended to limit the invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the invention should be included within the scope of protection of the present invention.

Claims

1. An optical coherent computing device comprising a laser device, a first beam splitter, a second beam splitter, a third beam splitter, a first modulator, a second modulator, a third modulator, a first converter, a second converter, a first coupler, a second coupler, an injector, a zero-difference frequency detector, and a computing control module, The output terminal of the laser device is connected to the input terminal of the first beam splitter. The output terminal of the first beam splitter is connected to the input terminals of the first modulator and the second modulator, respectively. The output terminal of the first modulator is connected to the input terminal of the first converter. The output terminal of the first converter is connected to the input terminal of the first coupler. The output terminal of the first coupler is connected to the input terminal of the second converter. The output terminal of the second converter is connected to the input terminal of the third beam splitter. The output terminals of the third beam splitter are connected to the zero-difference frequency detector and the input terminals of the second coupler, respectively. The output terminal of the second modulator is connected to the input terminal of the second beam splitter. The output terminals of the second beam splitter are connected to the zero-difference frequency detector and the input terminals of the third modulator, respectively. The output terminal of the zero-difference frequency detector is connected to the input terminal of the arithmetic control module. The output terminals of the aforementioned arithmetic control module are connected to the input terminals of the third modulator and the injector, respectively. The output terminal of the third modulator is connected to the input terminal of the second coupler. The output terminal of the second coupler is connected to the input terminal of the injector. The output terminal of the injector is connected to the input terminal of the first coupler. An optical coherent computing device characterized in that the first coupler, the second converter, the third beam splitter, the second coupler, and the injector are sequentially connected to a ring resonator via optical fibers.

2. The laser device is used to output a pulsed laser having a first wavelength, The first beam splitter is used to split the pulsed laser output by the laser device into two pulsed lasers and output them to the first modulator and the second modulator, respectively. The first modulator is used to periodically output to the first converter a first timing pulse group, each containing n pulse lasers (where n is an integer greater than 1) based on the received pulse lasers. The first converter is used to convert a received pulsed laser having a first wavelength into a pulsed laser having a second wavelength, and to input it to the second converter in the ring resonator via the first coupler. The second converter is used to convert the received pulsed laser having a second wavelength into a pulsed laser having a first wavelength and transmit it to the third beam splitter. The third beam splitter is used to split each received pulsed laser into two pulsed lasers, outputting one pulsed laser to the second coupler and the other pulsed laser to the zero-difference frequency detector. The second modulator is used to periodically output a second timing pulse group, each containing n pulse lasers, to the second beam splitter based on the received pulse laser, wherein a predetermined first time interval is set between the time the second modulator outputs the second timing pulse group and the time the first modulator outputs the first timing pulse group. The second beam splitter is used to split each received pulsed laser into two pulsed lasers, outputting one pulsed laser to the third modulator and the other pulsed laser to the zero-difference frequency detector. The zero-difference frequency detector is used to measure the balanced zero-difference frequency based on the two received pulsed lasers and to transmit an electrical signal to the calculation control module based on the measurement result. The aforementioned arithmetic control module is used to calculate the current pulse spin symbol and current isin energy based on the received electrical signal and a preset constraint matrix, to transmit a modulation command corresponding to the third modulator based on the set dissipation strength and coupling strength, and to transmit an injection command corresponding to the injector. The third modulator modulates the intensity and / or phase of the pulsed laser that has passed through it based on the modulation command, outputs the modulated pulsed laser to the second coupler, and is used to input the second coupler to the ring resonator. The second coupler is used to output the received pulsed laser to the injector. The optical coherent computing device according to claim 1, characterized in that the injector performs a mutual injection operation on the received first timing pulse group and second timing pulse group based on an injection command, and outputs the first timing pulse group and second timing pulse group after the completion of the mutual injection operation to the second converter via the first coupler, and a predetermined first time interval is set between the time the injector outputs the second timing pulse group and the time the injector outputs the first timing pulse group.

3. The optical coherent computing device is The optical coherent arithmetic unit further includes a timer for sending a stop command to the arithmetic control module if the operating time of the optical coherent arithmetic unit is greater than or equal to a preset time threshold, The optical coherent arithmetic apparatus according to claim 2, characterized in that the arithmetic control module is used to output the current optimal arithmetic result as the final arithmetic result based on a stop command.

4. The aforementioned injector is A delay module for performing a corresponding delay on each received pulse laser based on a pre-set delay policy, and for simultaneously transmitting the first pulse in the first timing pulse group and the first pulse in the second timing pulse group to the two input terminals of the injection module, respectively. The injection module for performing mutual injection operations on corresponding pulsed lasers in a first timing pulse group and a second timing pulse group based on an injection command, The optical coherent computing device according to claim 2, comprising a combiner module for forming a first timing pulse group and a second timing pulse group after injection is complete, wherein a preset first time interval is still maintained between each received pulse laser, the combined module applies a corresponding delay to each received pulse laser based on a preset delay policy, and sequentially outputs each received pulse laser to a first combiner according to a preset first time interval and transmission order.

5. The third modulator is An intensity modulator for modulating the intensity of a corresponding pulsed laser based on a modulation command, The optical coherent computing device according to claim 1, further comprising a phase modulator for modulating the phase of a corresponding pulsed laser based on a modulation command.

6. The optical coherent computing device according to claim 1, characterized in that the first converter is a second harmonic generator.

7. The optical coherent computing device according to claim 1, characterized in that both the first and second transducers are periodically polarized reversal lithium niobate crystals.

8. The optical coherent arithmetic apparatus according to claim 1, characterized in that the arithmetic control module is a field-programmable gate array.

9. An error correction method for an optical coherent computing device, Step A of setting up the optical coherent computing device according to any one of claims 1 to 8, Step B involves pre-setting constraint matrices in the arithmetic control module of the optical coherent arithmetic device based on the arithmetic problem to be processed, Step C involves starting the optical coherent computing device and periodically outputting a pulsed laser having a first wavelength to the first beam splitter from the laser device in the optical coherent computing device, Step D involves splitting each pulsed laser received by the first beam splitter into two pulsed lasers and outputting them to the first modulator and the second modulator, respectively. Step E involves periodically outputting a first timing pulse group and a second timing pulse group, each having n pulse lasers (where n is an integer greater than 1), with a predetermined first time interval between them, based on the pulse lasers received by the first modulator and the second modulator. Step F involves converting the wavelength of the first timing pulse group output by the first modulator using the first converter, inputting it to the ring resonator via the first coupler, converting the wavelength again using the second converter, and then transmitting it to the third beam splitter. Step G involves splitting each pulsed laser received by the third beam splitter into two pulsed lasers, outputting one pulsed laser to a zero-difference frequency detector, and continuously outputting the other pulsed laser to the second coupler via a ring resonator, and then continuously outputting it to the injector via a ring resonator. Step H involves transmitting the second timing pulse group output by the second modulator to the second beam splitter, splitting each pulsed laser received by the second beam splitter into two pulsed lasers, and transmitting them respectively to the third modulator and the zero-difference frequency detector. Step I involves a zero-difference frequency detector performing a balance zero-difference frequency measurement based on two pulsed lasers received, and transmitting an electrical signal to a calculation control module based on the measurement result. Step J involves the arithmetic control module calculating the current pulse spin symbol and current isin energy based on the received electrical signal and a preset constraint matrix, setting the current isin energy as the optimal energy value, setting the current pulse spin symbol and current isin energy as the current optimal calculation result, and transmitting a modulation command corresponding to the third modulator and an injection command corresponding to the injector based on the preset initial values ​​of the dissipation strength and coupling strength. Step K involves the third modulator modulating the intensity and / or phase of the pulsed laser that has passed through it based on the modulation command, outputting the modulated pulsed laser to the second coupler, inputting it to the ring resonator via the second coupler, and outputting it to the injector via the ring resonator. Step L: The injector performs a mutual injection operation on the received first timing pulse group and second timing pulse group based on the injection command and outputs the result, and after the mutual injection operation is completed, the first timing pulse group and second timing pulse group are sequentially transmitted to the third beam splitter via the first coupler and second converter in the ring resonator. Step M involves splitting each timing pulse received by the third beam splitter into two pulsed lasers, continuously transmitting one pulsed laser in a ring resonator, and outputting the other pulsed laser to a zero-difference frequency detector. Step N involves a zero-difference frequency detector performing a balanced zero-difference frequency measurement based on two pulsed lasers received from two input terminals, and transmitting an electrical signal to a computation control module based on the measurement result. Step O: The arithmetic control module calculates the current pulse spin symbol and current isin energy based on the electrical signal it receives and a pre-set constraint matrix. Step P determines whether the current isin energy Ei is less than the current optimal energy value. If so, step Q is performed; otherwise, step R is performed. The calculation control module sets the current isin energy as the energy optimal value, sets the current pulse spin symbol and current isin energy as the current optimal calculation result, updates the dissipation intensity and coupling intensity values ​​based on preset adjustment values, sends a modulation command corresponding to the third modulator and an injection command corresponding to the injector based on the updated dissipation intensity and coupling intensity values, and returns to step K to execute step Q. An error correction method for an optical coherent arithmetic device, characterized in that the arithmetic control module resets the values ​​of dissipation intensity and coupling intensity, transmits a modulation command corresponding to the third modulator based on the values ​​of dissipation intensity and coupling intensity after the reset, transmits an injection command corresponding to the injector, and returns to step K to execute step R.

10. Before returning to step K and executing, the method is: To determine whether the operating time of the optical coherent computing device is less than a preset time threshold, The method according to 9, further comprising, if so, returning to step K and executing, otherwise, the arithmetic control module outputting the current optimal arithmetic result as the final arithmetic result.

11. Updating the values ​​of dissipation intensity and binding intensity based on the aforementioned preset adjustment values ​​is: The method according to 9, characterized in that the current dissipation intensity and binding intensity values ​​are increased by a preset adjustment value, and then the updated dissipation intensity and binding intensity values ​​are obtained.

12. Resetting the aforementioned values ​​of dissipation strength and binding strength is Randomly selecting one value from a pre-defined first value range as the current value of the dissipation intensity, The method according to claim 9, characterized in that the initial value of the bond strength is set to the current value of the bond strength.

Images