Photon simulation iterative computing device

Abstract

The photon simulation iterative computing device comprises the following steps: outputting a first optical signal; dividing the first optical signal into a second optical signal, a third optical signal, a fourth optical signal and a fifth optical signal; modulating the power of the second optical signal according to the first modulation voltage corresponding to the original value to generate a sixth optical signal; equally dividing the sixth optical signal into a seventh optical signal and an eighth optical signal; performing photon multiplication and addition calculation on the seventh optical signal by using a second modulation voltage corresponding to the solution matrix to generate a ninth optical signal; a tenth optical signal corresponding to a proportional term is generated by using the eighth optical signal, an eleventh optical signal corresponding to an original value term is generated by using the fourth optical signal and the first modulation voltage, and a first output current is generated by performing photoelectric conversion on optical domain accumulated values of the ninth, tenth and eleventh optical signals by using the fifth optical signal; a second output current corresponding to the differential term is generated using the third optical signal and the first modulated voltage, and an output voltage corresponding to the new value is generated using the first output current and the second output current.

Description

Technical Field

[0001] This invention relates to the field of photonic simulation computing, and in particular to a photonic simulation iterative computing device for solving optimization problems. Background Technology

[0002] Solving combinatorial optimization problems has a wide range of applications in logistics and supply chain management, production and manufacturing, financial portfolio optimization, networks and communications, and artificial intelligence learning. Some well-known problems include vehicle routing, warehouse layout optimization, workshop scheduling, packing problems, and portfolio optimization, all of which play a significant role in social production and daily life.

[0003] However, combinatorial optimization problems are generally NP-hard problems, with their solution time and problem size increasing exponentially. For problems with a scale of tens of thousands, traditional computers may need to spend months to find the optimal or suboptimal solution. For even larger-scale problems, traditional computers are incapable. Therefore, solutions utilizing optical computing have been proposed to address combinatorial optimization problems. Optical computing is an emerging analog computing method. By using light as a carrier wave and converting analog electrical signals into light through an electro-optic modulator for numerical computation, optical computing can achieve low-energy consumption, low-latency, and high-computing-power requirements, making it particularly suitable for solving combinatorial optimization problems.

[0004] Analog Iterative Machine (AIM) is a typical optoelectronic hybrid computing architecture that combines analog electrical computing. By effectively combining optical computing and analog electrical computing, AIM demonstrates advantages over traditional von Neumann architectures in terms of low memory access ratio and high bandwidth. Figure 5 This diagram illustrates a traditional AIM implementation, using dashed arrows to represent electrical signals and solid arrows to represent optical signals. Figure 5 As shown, the entire system includes a laser (LD), multiple optoelectronic modulators (EOMs), an opto-matrix computation unit (oMAC), multiple transimpedance amplifiers (TIAs), multiple adders, and multiple iterative modules. The optoelectronic modulators process the electrical signals x1(t) to x2(t) from the iterative modules, corresponding to the calculation results of the iterative computations. N (t) are modulated into the laser signal emitted by the laser and provided to the photonic matrix computing unit. The photonic matrix computing unit, in conjunction with multiple photodiodes (PDs) configured at its internal rear end, performs photonic multiplication and addition calculations on the optical signal from the photoelectric modulator based on the solution matrix W, followed by photoelectric conversion. After passing through a transimpedance amplifier and adder at the rear end, the calculated signal is converted into a photoelectric signal. electrical analog signal These are provided to each iteration module. Each iteration module includes a proportional unit consisting of an amplifier and a proportional term (β(t)), and a differentiating unit consisting of a differentiator. The iteration module utilizes a nonlinear limiting amplifier to process the analog electrical signal. The matrix is ​​nonlinearized, and the results of the nonlinearized linear matrix operations from the scaling unit, the differential unit, and the limiting amplifier are summed using two summing circuits (add) to obtain the final calculation result. Summary of the Invention

[0005] The technical problem to be solved by the present invention

[0006] Although the iteration latency of the existing AIM implementation has been greatly reduced compared to traditional computers, apart from matrix-vector multiplication, operations such as differentiation, scaling, and nonlinearity are still performed by electrical computing. Therefore, the photoelectric conversion from optical computing to analog electrical computing has greatly reduced the energy consumption advantage of AIM compared to traditional computers.

[0007] Furthermore, existing AIM implementations employ a large number of analog electrical components, which have bandwidth-limited characteristics, further restricting the latency advantages of analog computing systems.

[0008] This invention was made to solve the above-mentioned technical problems. Its purpose is to provide a photonic simulation iterative computing device that can effectively improve the system bandwidth, thereby reducing the latency of simulation iterative computing, improving the computing accuracy, and significantly reducing the system power consumption.

[0009] Technical solutions to solve technical problems

[0010] The photonic simulation iterative computing apparatus of the first aspect of the present invention performs iterative calculations on a variable vector based on a solution matrix and according to an iterative formula, comprising: a laser generating module that outputs a first optical signal; a first power allocation module that divides the first optical signal into a second, third, fourth, and fifth optical signal with equal power; a photoelectric modulation module that modulates the power of the second optical signal according to a first modulation voltage corresponding to the original value of the variable vector to generate a sixth optical signal; a second power allocation module that divides the sixth optical signal into a seventh and eighth optical signal with equal power; and a photonic multiply-accumulate module that utilizes a solution matrix... The corresponding second modulation voltage performs photon multiplication and addition calculations on the seventh optical signal to generate a ninth optical signal; and an iteration module, which uses the eighth optical signal to generate a tenth optical signal corresponding to the proportional term in the iteration formula, uses the fourth optical signal and the first modulation voltage to generate an eleventh optical signal corresponding to the original value term in the iteration formula, uses the fifth optical signal to perform photoelectric conversion on the optical domain accumulation value of the ninth, tenth, and eleventh optical signals to generate a first output current, uses the third optical signal and the first modulation voltage to generate a second output current corresponding to the differential term in the iteration formula, and uses the first and second output currents to generate an output voltage corresponding to the new value of the variable vector.

[0011] Optionally, the iterative module includes a differentiating unit that uses the first modulation voltage and the third optical signal to generate a twelfth optical signal, and uses the twelfth optical signal, the adjustment voltage, and a fourth modulation voltage corresponding to the momentum factor in the differential term to generate the second output current.

[0012] Optionally, the differentiating unit includes: a first phase modulator that modulates the phase of the third optical signal according to the first modulation voltage to generate the twelfth optical signal; a first power divider that divides the twelfth optical signal into a thirteenth optical signal and a fourteenth optical signal with the same power; a second power divider that divides the thirteenth optical signal into a fifteenth optical signal and a sixteenth optical signal with the same power; a second phase modulator that adjusts the time delay and phase of the fifteenth optical signal according to the adjustment voltage to generate a seventeenth optical signal; a combiner that combines the sixteenth optical signal and the seventeenth optical signal into an eighteenth optical signal; a first photoelectric modulation unit that modulates the power of the eighteenth optical signal according to the fourth modulation voltage to generate a nineteenth optical signal; and a first photoelectric conversion unit that couples the fourteenth optical signal and the nineteenth optical signal through multimode interference and performs photoelectric conversion to generate the second output current.

[0013] Optionally, the iterative module includes: a multiplexing unit that performs optical domain accumulation of the ninth, tenth, and eleventh optical signals to generate a twentieth optical signal; and a second photoelectric conversion unit that couples the fifth and twentieth optical signals through multimode interference and performs photoelectric conversion to generate the first output current.

[0014] Optionally, the iteration module includes a scaling unit that modulates the power of the eighth optical signal according to a third modulation voltage corresponding to the annealing factor in the scaling term to generate the tenth optical signal.

[0015] Optionally, the iterative module includes a second photoelectric modulation unit, which modulates the power of the fourth optical signal according to the first modulation voltage to generate the eleventh optical signal.

[0016] Optionally, the iterative module includes a transimpedance amplifier unit that converts the sum of the first output current and the second output current into a voltage and amplifies it to obtain the output voltage.

[0017] Optionally, the photon simulation iterative calculation device adjusts the voltage amplification factor of the transimpedance amplifier unit so that the new value of the variable vector is linearly related to the original value of the variable vector, thereby calibrating the iterative formula.

[0018] Optionally, the iterative module further includes an optical domain nonlinear unit, which uses a nonlinear function to map the variables of the ninth optical signal to binary or continuous time values ​​to obtain a twenty-first optical signal. The iterative module uses the twenty-first optical signal to replace the ninth optical signal to generate the first output current.

[0019] Optionally, the electrical domain portion of the photonic simulation iterative computing device is electrically connected via an electrically adjustable time delay line, which ensures that the time delay of the electrical domain portion remains consistent.

[0020] Optionally, the photon simulation iterative computing device includes N iteration modules, where N is a positive integer corresponding to the dimension of the variable vector. The first power allocation module generates N second optical signals, N third optical signals, N fourth optical signals, and N fifth optical signals, and provides the N third optical signals, N fourth optical signals, and N fifth optical signals to the corresponding iteration modules.

[0021] Optionally, the photoelectric modulation module includes N third photoelectric modulation units, each of which modulates the power of a corresponding second optical signal among the N second optical signals according to the first modulation voltage from the corresponding iteration module among the N iteration modules, thereby generating N sixth optical signals.

[0022] Optionally, the second power distribution module includes N third power distributors, each of which divides each sixth optical signal into a seventh optical signal and an eighth optical signal, and provides each eighth optical signal to the corresponding iteration module.

[0023] Optionally, the solution matrix is ​​an N×N matrix, and the photon multiply-add module is composed of an array of N×N photoelectric conversion devices. The light transmittance of the photoelectric conversion devices corresponds to the scaling factor in the iterative formula. The photoelectric conversion devices modulate the power of the corresponding seventh optical signal according to the second modulation voltage, and accumulate the modulated optical signals in the same column in the optical domain to generate N ninth optical signals, which are then provided to the corresponding iterative modules.

[0024] Invention Effects

[0025] The photonic simulation iterative computing device according to the present invention can effectively improve the system bandwidth, thereby reducing the latency of simulation iterative computing, improving the computing accuracy, and significantly reducing the system power consumption. Attached Figure Description

[0026] Figure 1This is a schematic diagram showing the structure of the photon simulation iterative computing device according to Embodiment 1 of the present invention.

[0027] Figure 2 It is shown Figure 1 A schematic diagram of the structure of the differential unit in the diagram.

[0028] Figure 3 This is a schematic diagram showing the structure of the photon simulation iterative computing device according to Embodiment 2 of the present invention.

[0029] Figure 4 This is a schematic diagram showing the structure of the photon simulation iterative computing device according to Embodiment 3 of the present invention.

[0030] Figure 5 This is a schematic diagram illustrating a traditional structure for implementing AIM. Detailed Implementation

[0031] The following description provides specific application scenarios and requirements for this specification, intended to enable those skilled in the art to make and use the contents of this specification. Various partial modifications to the disclosed embodiments will be apparent to those skilled in the art, and the general principles defined herein can be applied to other embodiments and applications without departing from the spirit and scope of this specification. Therefore, this specification is not limited to the embodiments shown, but rather to the widest scope consistent with the claims.

[0032] Furthermore, considering the following description, these and other features of this specification, as well as the operation and function of the related components of the structure, and the economy of assembly and manufacture of the parts, can be significantly improved. All of these form part of this specification with reference to the accompanying drawings. However, it should be clearly understood that the drawings are for illustrative and descriptive purposes only and are not intended to limit the scope of this specification. It should also be understood that the drawings are not drawn to scale.

[0033] First, the basic principles of simulated iterative calculation will be explained.

[0034] The method of simulating iterative calculation can be implemented based on the iterative formula of the following equation (1).

[0035] [Mathematical Expression 1]

[0036]

[0037] in, Let N be a variable vector with dimensions [N, 1], representing continuous real-valued state variables at time iteration t, where N is the variable vector. Represents the original value of the variable vector. Represents the new value of the variable vector. This represents the value of the variable vector in the previous iteration period; W is the solution matrix with dimensions [N, N]; f nonlinear (·) represents a nonlinear function of the elements within the brackets, obtained by mapping the variables within the brackets to binary or continuous time values; α, β, and γ are constants, representing the scaling factor, annealing factor, and momentum factor in the iterative formula, respectively, and are typically less than 1. Additionally, in equation (1)... For the original value, For matrix operations, For the proportion term, This is the differential term.

[0038] The above iterative formula shows the new values ​​of the variable vector. Compared with the original value The relationship between these relationships applies to binary optimization problems, such as Quadratic Unconstrained Binary Optimization (QUBO) and the Traveling Salesman Problem (TSP). These problems can be effectively solved using the Analog Iterative Machine (AIM) scheme shown in the iterative formulas above. Furthermore, the AIM using the aforementioned iterative formulas can also solve more continuous variable optimization problems, such as financial portfolio optimization problems.

[0039] Hereinafter, with reference to the accompanying drawings, a method for implementing iterative calculations corresponding to the above-described iterative formulas using the photon simulation iterative computing apparatus according to a preferred embodiment of the present invention will be described in detail.

[0040] Implementation Method 1

[0041] Figure 1 This is a schematic diagram showing the structure of the photon simulation iterative computing device according to Embodiment 1. The solid lines in the diagram represent optical signals, and the dashed lines represent electrical signals. The photon simulation iterative computing device of this embodiment, based on the solution matrix W, applies the iterative formula described later to the variable vector... Iterative calculations are performed. Furthermore, in this embodiment, a variable vector is defined. The dimension of the variable is [1, 1], which is a one-dimensional vector relative to time. Therefore, the variable vector can also be represented as x(t).

[0042] like Figure 1 As shown, the photonic simulation iterative computing device includes a laser generation module 1, a first power distribution module 2, an optoelectronic modulation module 3, a second power distribution module 4, a photonic multiply-accumulate module 5, and an iteration module 6.

[0043] The laser generating module 1 is composed of a device capable of emitting laser light, such as a laser diode (LD), and is used to generate and output a first optical signal, which is, for example, a laser signal with a certain power and phase. In this embodiment, it is assumed that the first optical signal output by the laser generating module 1 can be represented by the following formula (2).

[0044] [Mathematical Expression 2]

[0045]

[0046] Where A0 is the electric field intensity of the optical power, ω0 is the angular frequency of the light, and φ0 is the initial phase of the light.

[0047] The first power distribution module 2, for example, is composed of a power separator (PS), used to distribute the first optical signal... The optical signals are divided into four signals with the same power: a second optical signal, a third optical signal, a fourth optical signal, and a fifth optical signal. In this embodiment, each of the second to fifth optical signals after being divided by the first power distribution module 2 can be represented by the following formula (3).

[0048] [Mathematical Expression 3]

[0049]

[0050] The optoelectronic modulation module 3 is, for example, composed of an electro-optic modulator (EOM). This modulator can employ electro-optic conversion devices such as a Mach-Zehnder modulator (MZM) or a voltage-controlled optical attenuator (VOA) combined with linearization modulation technology to achieve linear modulation of the amplitude of the analog electrical signal into an optical signal. The optoelectronic modulation module 3 obtains the second optical signal from the first power distribution module 2. Based on the first modulation voltage x(t) corresponding to the original value of the variable vector, the acquired second optical signal is... The power is modulated to generate the sixth optical signal. In this embodiment, the sixth optical signal output by the photoelectric modulation module 3 at time t can be represented by the following formula (4).

[0051] [Mathematical Expression 4]

[0052]

[0053] The second power distribution module 4, for example, is composed of a power divider, used to distribute the sixth optical signal... The optical signals are divided into a seventh optical signal and an eighth optical signal, each with the same power. In this embodiment, the seventh and eighth optical signals after being divided by the second power distribution module 4 can be represented by the following formula (5).

[0054] [Mathematical Expression 5]

[0055]

[0056] The photon multiply-accumulate module 5, for example, is composed of a photon matrix calculator (oMAC), which obtains the second modulation voltage corresponding to the solved matrix W from an external source and obtains the seventh optical signal from the second power distribution module 4. Using the second modulation voltage W, the seventh optical signal Perform photon multiplication and addition calculations to generate the ninth optical signal.

[0057] Specifically, in this embodiment, the solution matrix W is, for example, an N×1 matrix, where N is a positive integer. In this case, the photonic multiply-add module 5 can be composed of an array of photoelectric conversion devices such as N×1 micro-ring modulators, Mach-Zehnder modulators, and electro-absorption modulators, with the light transmittance of each photoelectric conversion device corresponding to the scaling factor α in the iterative formula. Each photoelectric conversion device modulates the power of the seventh optical signal according to the second modulation voltage of each row. The photonic multiply-add module 5 performs optical domain accumulation on the modulated optical signal to generate the ninth optical signal.

[0058] In this embodiment, the ninth optical signal after photon multiplication and addition calculation by photon multiplication and addition module 5 can be represented by the following formula (6).

[0059] [Mathematical Expression 6]

[0060]

[0061] Iteration module 6 obtains the third optical signal from the first power allocation module 2. Fourth optical signal and the fifth optical signal The eighth optical signal is obtained from the second power distribution module 4. The ninth optical signal is obtained from photon multiplying module 5. Using the eighth optical signal To generate the proportional term in the iterative formula The corresponding tenth optical signal Using the fourth optical signal The first modulation voltage x(t) is used to generate the original value term in the iterative formula. The corresponding eleventh optical signal Using the fifth optical signal For the ninth optical signal Tenth optical signal and the eleventh optical signal The optical domain accumulation value is used for photoelectric conversion to generate the first output current I.out1 Using third optical signals The first modulation voltage x(t) is used to generate the differential term in the iterative formula. The corresponding second output current I out,2 Using the first output current I out1 and the second output current I out,2 To generate new values ​​for the variable vector The corresponding output voltage.

[0062] Specifically, such as Figure 1 As shown, in this embodiment, the iteration module 6 includes a differentiating unit 11, a proportional unit 12, a second photoelectric modulation unit 13, a multiplexing unit 14, a second photoelectric conversion unit 15, and a transimpedance amplifier unit 16.

[0063] Differentiating unit 11 obtains the first modulation voltage x(t) from transimpedance amplifier unit 16 and the third optical signal from first power distribution module 2. Using the first modulation voltage x(t) and the third optical signal To generate the twelfth optical signal, and to use the twelfth optical signal to adjust the voltage V diff Sum and Differential Terms The momentum factor γ in the fourth modulation voltage is used to generate the second output current I. out,2 The specific structure of differential unit 11 will be explained in detail later.

[0064] The scaling unit 12 is, for example, composed of an electro-optic modulator (EOM), according to the scaling term. The annealing factor β corresponds to the third modulation voltage β(t) for the eighth optical signal. The power is modulated to generate the tenth optical signal. In this embodiment, the scaling unit 12 can calculate the tenth optical signal using the following formula (7).

[0065] [Mathematical Expression 7]

[0066]

[0067] The second photoelectric modulation unit 13 is, for example, composed of a photoelectric modulator (EOM), which modulates the fourth optical signal according to the first modulation voltage x(t). The power is modulated to generate the eleventh optical signal. In this embodiment, the second photoelectric modulation unit 13 can calculate the eleventh optical signal using the following formula (8).

[0068] [Mathematical Expression 8]

[0069]

[0070] The multiplexing unit 14, for example, is composed of a multiplexer (MUX) to multiplex the ninth optical signal. Tenth optical signal and the eleventh optical signal Perform optical domain accumulation to generate the twentieth optical signal. In this embodiment, the multiplexing unit 14 can calculate the twentieth optical signal using the following formula (9).

[0071] [Mathematical Expression 9]

[0072]

[0073] The second photoelectric conversion unit 15, for example, is composed of a balanced photodiode (BPD), and may include a multimode interferometer and two photodiodes. The second photoelectric conversion unit 15 acquires the fifth optical signal from the first power distribution module 2. The twentieth optical signal is obtained from multiplexing unit 14. Using a multimode interferometer to transmit the fifth optical signal 20th optical signal Coupled together, and using two photodiodes for photoelectric conversion, a first output current I is generated. out,1 In this embodiment, the second photoelectric conversion unit 15 can convert the fifth optical signal using the following formula (10). 20th optical signal Multimode interference is performed to achieve coupling.

[0074] [Mathematical Expression 10]

[0075]

[0076] Furthermore, the second photoelectric conversion unit 15 can use two photodiodes to convert the two optical signals in the above formula (10) into two optical signals. and The signal is converted into an electrical signal, and the first output current I is calculated using the following equation (11). out,1 .

[0077] [Mathematical Expression 11]

[0078]

[0079] The transimpedance amplifier unit 16, for example, is composed of a transimpedance amplifier (TIA), which converts the first output current I... out,1With the second output current I from the differentiating unit 11 out,2 The sum is converted into voltage and amplified to obtain a new value with respect to the variable vector. The corresponding output voltage. A specific embodiment for calculating the output voltage of the transimpedance amplifier unit 16 will be described in detail later.

[0080] The following describes a specific structural example of the differential unit 11 in iterative module 6.

[0081] Figure 2 It is shown Figure 1 The diagram shows the structure of the differential unit 11, where the solid lines represent optical signals and the dashed lines represent electrical signals.

[0082] like Figure 2 As shown, the differential unit 11 includes a first phase modulator 101, a first power divider 102, a second power divider 103, a second phase modulator 104, a beam combiner 105, a first photoelectric modulation unit 106, and a first photoelectric conversion unit 107.

[0083] The first phase modulator 101 can be constructed using an existing phase modulator (PM). It obtains the third optical signal from the first power distribution module 2 and the first modulation voltage from the transimpedance amplifier unit 16. Based on the first modulation voltage, it modulates the phase of the third optical signal, that is, modulates the first modulation voltage into the phase of the third optical signal to generate the twelfth optical signal.

[0084] The first power divider 102 can be constructed using an existing power divider (PS), which divides the twelfth optical signal from the first phase modulator 101 into two optical signals with the same power, namely the thirteenth and fourteenth optical signals. The fourteenth optical signal directly enters the first photoelectric conversion unit 107 (described later), while the thirteenth optical signal enters the second power divider 103. In this embodiment, the thirteenth and fourteenth optical signals can be represented by the following equation (12).

[0085] [Mathematical Expression 12]

[0086]

[0087] The second power divider 103 can also be constructed using an existing power divider PS, which further divides the thirteenth optical signal from the first power divider 102 into two optical signals with the same power, namely the fifteenth optical signal and the sixteenth optical signal. The sixteenth optical signal directly enters the optical combiner 105 (described later), and the fifteenth optical signal enters the second phase modulator 104.

[0088] The second phase modulator 104 can also be constructed using an existing phase modulator PM, obtaining a variable adjustment voltage V from an external source for adjusting the time delay of the optical signal. diff According to the adjusted voltage V diff The time delay and phase of the fifteenth optical signal from the second power divider 103 are adjusted to generate the seventeenth optical signal.

[0089] Optical combiner 105 combines the sixteenth optical signal from the second power divider 103 with the seventeenth optical signal from the second phase modulator 104 to form the eighteenth optical signal. Since the second phase modulator 104 uses an adjustment voltage to adjust the time delay and phase of the seventeenth optical signal, which has the same power and phase as the sixteenth optical signal, the combined eighteenth optical signal can achieve signal subtraction in the optical domain. The time delay difference is τ. Furthermore, since this time delay difference τ is relatively small, the eighteenth optical signal output by the optical combiner 105 can be written in integral form.

[0090] The first photoelectric modulation unit 106, for example, is composed of a photoelectric modulator (EOM). Based on a fourth modulation voltage corresponding to the momentum factor γ obtained from the outside, it modulates the power of the eighteenth optical signal from the combiner 105 to generate the nineteenth optical signal. In this embodiment, the nineteenth optical signal output by the first photoelectric modulation unit 106 can be... It is expressed as equation (13).

[0091] [Mathematical Expression 13]

[0092]

[0093] The first photoelectric conversion unit 107 is similar to the second photoelectric conversion unit 15, for example, composed of a balanced photodiode (BPD), and may include a multimode interferometer and two photodiodes. The first photoelectric conversion unit 107 acquires the fourteenth optical signal from the first power divider 102. The nineteenth optical signal is obtained from the first photoelectric modulation unit 106. The fourteenth optical signal was transmitted using a multimode interferometer. and the nineteenth optical signal Coupled together, and using two photodiodes for photoelectric conversion, a second output current I is generated. out,2 In this embodiment, the first photoelectric conversion unit 107 can convert the fourteenth optical signal using the following formula (14). and the nineteenth optical signal Multimode interference is performed to achieve coupling.

[0094] [Mathematical Expression 14]

[0095]

[0096] Furthermore, the first photoelectric conversion unit 107 can use two photodiodes to convert the two optical signals in the above equation (14) into two optical signals. and The signal is converted into an electrical signal, and the second output current I is calculated using the following equation (15). out,2 .

[0097] [Mathematical Expression 15]

[0098]

[0099] Back Figure 1 In this embodiment, the transimpedance amplifier 16 can calculate the new value of the variable vector using the following equation (16). The corresponding output voltage.

[0100] [Mathematical Expression 16]

[0101]

[0102] Where, μ TIA The voltage amplification factor of the transimpedance amplifier unit 16 is determined by μ. TIA Adjustments can make That is, to make the new value of the variable vector Compared with the original value The relationship is linear, which allows the iteratively obtained value to be scaled back to the original range, effectively calibrating the iterative formula. α″, β″(t), and γ″ are the scaling factor, annealing factor, and momentum factor in the simplified iterative formula. Through parameter matching, adjustment, and calibration, the matching between α″, β″(t), and γ″ and α, β(t), and γ can be achieved. Let be the noise vector of the system, representing the computational error of the system. Through system optimization, it can be... Adjust to zero.

[0103] As described above, in the photonic simulation iterative computing apparatus according to this embodiment, the matrix-vector multiplication in the photonic multiply-add module 5 and the scaling and differentiation operations in the iterative module 6 are all performed using optical signals. They only need to be converted to electrical signals at the balanced detector (PD) terminals, such as the first photoelectric conversion unit 107 and the second photoelectric conversion unit 15. Therefore, by replacing analog electrical computation with photonic computation, the system bandwidth can be effectively improved, the computational accuracy increased, and the overall system processing latency significantly lower than that of traditional AIM.

[0104] Furthermore, the photon simulation iterative computing device according to this embodiment eliminates numerous electrical operational amplifiers, amplifiers, and electrical processing units, thus the power consumption of the entire system is significantly lower than that of traditional AIM.

[0105] The photonic simulation iterative computing device according to Embodiment 1 has been described above. In this invention, the electric domain portion of the photonic simulation iterative computing device, i.e. Figure 3 The circuit section shown by the dashed line can also be electrically connected using an adjustable delay line. By adjusting the length of this adjustable delay line, the delay of each electrical domain can be kept consistent, thereby ensuring the consistency of the timing of the calculation results for each electrical domain and further ensuring the accuracy of the iterative calculation.

[0106] Implementation Method 2

[0107] Figure 3 This is a schematic diagram showing the structure of the photonic simulation iterative computing device according to Embodiment 2. The solid lines in the diagram represent optical signals, and the dashed lines represent electrical signals. The difference between the photonic simulation iterative computing device according to Embodiment 2 and Embodiment 1 is that the iteration module 6 further includes an optical domain nonlinear unit 17. Figure 2 In this document, components identical to those in the photon simulation iterative computing apparatus of Embodiment 1 are labeled with the same reference numerals. The photon simulation iterative computing apparatus according to this embodiment will now be described, focusing on the differences from Embodiment 1.

[0108] like Figure 3 As shown, the iteration module 6 also includes an optical domain nonlinear unit 17. This optical domain nonlinear unit 17 is composed, for example, of optical nonlinear devices such as a saturable absorption detector or a microring, and utilizes a nonlinear function to process the ninth optical signal from the photon multiply-accumulate module 5. The variables are mapped to binary or continuous time values ​​to obtain the twenty-first optical signal.

[0109] Furthermore, iteration module 6 uses the twenty-first optical signal Replacing the ninth optical signal To generate the first output current I out,1 Regarding the generation of the first output current I out,1 For the specific method, please refer to the above explanation of equations (9) to (11). Just replace equation (9) with the following equation (17).

[0110] [Mathematical Expression 17]

[0111]

[0112] As described above, in the photonic simulation iterative computing device according to this embodiment, since the nonlinearization in the iteration module 6 is also accomplished through optical signals, it will not have any adverse effects on the time delay advantage and bandwidth advantage of the photonic simulation iterative computing device.

[0113] Furthermore, since nonlinearity is introduced through the optical domain nonlinear unit 17, the robustness of the entire system can be improved, resulting in higher stability and stronger adaptability.

[0114] Implementation Method 3

[0115] Figure 4 This is a schematic diagram showing the structure of the photon simulation iterative computing device according to Embodiment 3. The solid lines in the diagram represent optical signals, and the dashed lines represent electrical signals. The difference between the photon simulation iterative computing device according to Embodiment 3 and Embodiments 1 and 2 is that the photon simulation iterative computing device in this embodiment uses a solution matrix W of dimension [N, N] to process a variable vector of dimension [N, 1]. Iterative calculations are performed, where N is a positive integer greater than 1. Therefore, based on Embodiment 2 described above, this embodiment modifies the structure of each part of the photon simulation iterative calculation device except for the laser generation module 1. The photon simulation iterative calculation device involved in this embodiment will now be described, focusing on the differences between this embodiment and Embodiment 2.

[0116] like Figure 4 As shown, the photonic simulation iterative computing device includes a laser generation module 1, a first power allocation module 2, a photoelectric modulation module 3, a second power allocation module 4, a photonic multiply-accumulate module 5, and N iteration modules 6 (X1(t) to X... in the figure). N The iterative module of (t), where N is the variable vector. (The dimension corresponds to a positive integer greater than 1).

[0117] The structure of laser generating module 1 is the same as that of embodiments 1 and 2 described above, and the description is omitted here.

[0118] The first power distribution module 2 will convert the first optical signal The optical signals are divided into N groups of N×4, consisting of N second, N third, N fourth, and N fifth optical signals with the same power, and each group is provided to a corresponding iteration module. Specifically, the first group of four optical signals (second, third, fourth, and fifth optical signals) is provided to iteration module x1(t), the second group of four optical signals is provided to iteration module x2(t), and so on. In this embodiment, the second to fifth optical signals in each group of optical signals after being divided by the first power distribution module 2 can be represented by the following equation (18).

[0119] [Mathematical Expression 18]

[0120]

[0121] The photoelectric modulation module 3 includes N third photoelectric modulation units 18, each of which modulates according to the first modulation voltage (x1(t), x2(t), ... x) from the corresponding iteration module 6 among the N iteration modules 6. N (t)), for N second optical signals The power of the corresponding second optical signal is modulated to generate N sixth optical signals. In this embodiment, the sixth optical signal output by the i-th third optoelectronic modulation unit 18 in the optoelectronic modulation module 3 at time t can be represented by the following formula (19).

[0122] [Mathematical Expression 19]

[0123]

[0124] The second power distribution module 4 includes N third power distributors 19, each of which distributes the sixth optical signal... The optical signal is divided into a seventh optical signal and an eighth optical signal, and each eighth optical signal is provided to the corresponding iteration module 6. In this embodiment, the seventh optical signal and the eighth optical signal after being divided by the third power divider 19 in the second power distribution module 4 can be represented by the following formula (20).

[0125] [Mathematical Expression 20]

[0126]

[0127] The photon multiply-add module 5 consists of an N×N array of photoelectric conversion devices. The light transmittance of each photoelectric conversion device corresponds to the scaling factor α in the iterative formula. Each photoelectric conversion device modulates the corresponding seventh optical signal according to the second modulation voltage corresponding to each element in the solution matrix W. The power is modulated, and the modulated optical signals in the same column are accumulated in the optical domain to generate N ninth optical signals, which are then provided to the corresponding iteration module 6. In this embodiment, the ninth optical signal of the i-th path among the N ninth optical signals after photon multiplication and addition calculation by the photon multiplication and addition module 5 can be represented by the following formula (21).

[0128] [Mathematical Expression 21]

[0129]

[0130] Where W(i, n) represents the element in the i-th row and n-th column of matrix W, and W(i, ∶) represents all elements in the i-th row of matrix W.

[0131] In this embodiment, the structures of each of the N iteration modules 6 are identical. The following explanation will use the x1(t) iteration module 6 as an example.

[0132] like Figure 4 As shown, the iterative module 6 includes a differentiating unit 11, a scaling unit 12, a second photoelectric modulation unit 13, a multiplexing unit 14, a second photoelectric conversion unit 15, a transimpedance amplifier unit 16, and an optical domain nonlinear unit 17.

[0133] The optical domain nonlinear unit 17 uses a nonlinear function to transform the ninth optical signal from the i-th path of the photon multiply-accumulate module 5. The variables are mapped to binary or continuous time values ​​to obtain the twenty-first optical signal.

[0134] Proportional unit 12 is based on the proportional term The annealing factor β corresponds to the third modulation voltage β(t) for the eighth optical signal. The power is modulated to generate the tenth optical signal. In this embodiment, the scaling unit 12 can calculate the tenth optical signal using the following formula (22).

[0135] [Mathematical Expression 22]

[0136]

[0137] The second photoelectric modulation unit 13 modulates the fourth optical signal according to the first modulation voltage x1(t). The power is modulated to generate the eleventh optical signal. In this embodiment, the second photoelectric modulation unit 13 can calculate the eleventh optical signal using the following formula (23).

[0138] [Mathematical Expression 23]

[0139]

[0140] Multiplexing unit 14 will use the twenty-first optical signal Tenth optical signal and the eleventh optical signal Perform optical domain accumulation to generate the twentieth optical signal. In this embodiment, the multiplexing unit 14 can calculate the twentieth optical signal using the following formula (24).

[0141] [Mathematical Expression 24]

[0142]

[0143] The second photoelectric conversion unit 15 obtains the fifth optical signal from the first power distribution module 2. The twentieth optical signal is obtained from multiplexing unit 14. Using a multimode interferometer to transmit the fifth optical signal 20th optical signal Coupled together, and using two photodiodes for photoelectric conversion, a first output current I is generated. out,1 In this embodiment, the second photoelectric conversion unit 15 can convert the fifth optical signal using the following formula (25). 20th optical signal Multimode interference is performed to achieve coupling.

[0144] [Mathematical Expression 25]

[0145]

[0146] Furthermore, the second photoelectric conversion unit 15 can use two photodiodes to convert the two optical signals in the above formula (25) into two optical signals. and The signal is converted into an electrical signal, and the first output current I is calculated using the following equation (26). out,1 .

[0147] [Mathematical Expression 26]

[0148]

[0149] The structure of the differential unit 11 can be the same as in Embodiment 1 described above. Therefore, referring to... Figure 2 Let me explain.

[0150] The differentiating unit 11 of the i-th path includes a first phase modulator 101, a first power divider 102, a second power divider 103, a second phase modulator 104, a beam combiner 105, a first photoelectric modulation unit 106, and a first photoelectric conversion unit 107.

[0151] The first phase modulator 101 obtains the third optical signal from the first power distribution module 2 and the first modulation voltage from the transimpedance amplifier unit 16. Based on the first modulation voltage, it modulates the phase of the third optical signal, that is, modulates the first modulation voltage into the phase of the third optical signal, thereby generating the twelfth optical signal.

[0152] The first power divider 102 divides the twelfth optical signal from the first phase modulator 101 into two optical signals with the same power, namely the thirteenth optical signal and the fourteenth optical signal. The fourteenth optical signal directly enters the first photoelectric conversion unit 107, and the thirteenth optical signal enters the second power divider 103. In this embodiment, the thirteenth optical signal and the fourteenth optical signal can be represented by the following equation (27).

[0153] [Mathematical Expression 27]

[0154]

[0155] The second power divider 103 further divides the thirteenth optical signal from the first power divider 102 into two optical signals with the same power, namely the fifteenth optical signal and the sixteenth optical signal. The sixteenth optical signal directly enters the optical combiner 105, and the fifteenth optical signal enters the second phase modulator 104.

[0156] The second phase modulator 104 obtains a variable adjustment voltage V from an external source. diff According to the adjusted voltage V diff The time delay and phase of the fifteenth optical signal from the second power divider 103 are adjusted to generate the seventeenth optical signal.

[0157] Optical combiner 105 combines the sixteenth optical signal from second power divider 103 with the seventeenth optical signal from second phase modulator 104 to form the eighteenth optical signal, thus achieving signal subtraction in the optical domain after beam combining. The time delay difference is τ. Furthermore, since this time delay difference τ is relatively small, the eighteenth optical signal output by the optical combiner 105 can be written in integral form.

[0158] The first photoelectric modulation unit 106 modulates the power of the eighteenth optical signal from the combiner 105 according to the fourth modulation voltage corresponding to the momentum factor γ obtained from the outside, thereby generating the nineteenth optical signal. In this embodiment, the nineteenth optical signal output by the first photoelectric modulation unit 106 can be... It is expressed as equation (28).

[0159] [Mathematical Expression 28]

[0160]

[0161] The first photoelectric conversion unit 107 obtains the fourteenth optical signal from the first power divider 102. The nineteenth optical signal is obtained from the first photoelectric modulation unit 106. The fourteenth optical signal was transmitted using a multimode interferometer. and the nineteenth optical signal Coupled together, and using two photodiodes for photoelectric conversion, a second output current I is generated. out,2 In this embodiment, the first photoelectric conversion unit 107 can convert the fourteenth optical signal using the following formula (29). and the nineteenth optical signal Multimode interference is performed to achieve coupling.

[0162] [Mathematical Expression 29]

[0163]

[0164] Furthermore, the first photoelectric conversion unit 107 can use two photodiodes to convert the two optical signals in the above equation (29) into two optical signals. and The signal is converted into an electrical signal, and the second output current I is calculated using the following equation (30). out,2 .

[0165] [Mathematical Expression 30]

[0166]

[0167] Back Figure 4 In this embodiment, the transimpedance amplifier 16 can calculate the new value of the variable vector using the following equation (31). The corresponding output voltage.

[0168] [Mathematical Expression 31]

[0169]

[0170] Where, μ TIA The voltage amplification factor of the transimpedance amplifier unit 16 is determined by μ. TIA Adjustments can make That is, to make the new value of the variable vector Compared with the original value The relationship is linear, which allows the iteratively obtained value to be scaled back to the original range, effectively calibrating the iterative formula. α″, β″(t), and γ″ are the scaling factor, annealing factor, and momentum factor in the simplified iterative formula. Through parameter matching, adjustment, and calibration, the matching between α″, β″(t), and γ″ and α, β(t), and γ can be achieved. Let be the noise vector of the system, representing the computational error of the system. Through system optimization, it can be... Adjust to zero.

[0171] As described above, the photon simulation iterative computing device according to this embodiment can perform high-precision calculations on [N,1]-dimensional variable vectors based on the [N,N]-dimensional solution matrix W with high bandwidth, low power consumption, and low latency. Perform iterative calculations.

[0172] The photonic simulation iterative computing device involved in this invention has been described above. It should be considered that all aspects of the embodiments disclosed herein are merely illustrative and not restrictive. The scope of this disclosure is defined by the claims, and not by the above embodiments. The scope of this disclosure also includes all modifications and variations within the meaning and scope equivalent to the claims.

[0173] Industrial practicality

[0174] As described above, the photonic simulation iterative computing device according to the present invention is useful for solving combinatorial optimization problems and any other optimization problems in scenarios such as logistics and supply chain management, production and manufacturing, financial portfolio optimization, network and communication, and artificial intelligence learning.

[0175] Label Explanation

[0176] 1. Laser Generating Module

[0177] 2 First Power Distribution Module

[0178] 3. Optoelectronic modulation module

[0179] 4 Second power distribution module

[0180] 5 Photon Multiply-Accumulate Module

[0181] 6. Iteration Module

[0182] 11 Differential Units

[0183] 12 Scale Units

[0184] 13 Second photoelectric modulation unit

[0185] 14 Multiplexing Units

[0186] 15 Second photoelectric conversion unit

[0187] 16 Transimpedance Amplifier Units

[0188] 17 Optical Domain Nonlinear Units

[0189] 18 Third photoelectric modulation unit

[0190] 19 Third Power Divider

[0191] 101 First Phase Modulator

[0192] 102 First Power Divider

[0193] 103 Second Power Divider

[0194] 104 Second Phase Modulator

[0195] 105 light combiner

[0196] 106 First photoelectric modulation unit

[0197] 107 First photoelectric conversion unit

Claims

1. A photonic simulation iterative computing device, which performs iterative calculations on a variable vector based on a solution matrix and according to an iterative formula, characterized in that, include: A laser generating module that outputs a first optical signal; A first power distribution module divides the first optical signal into a second optical signal, a third optical signal, a fourth optical signal, and a fifth optical signal, each with the same power. A photoelectric modulation module modulates the power of the second optical signal according to a first modulation voltage corresponding to the original value of the variable vector, thereby generating a sixth optical signal; The second power distribution module divides the sixth optical signal into a seventh optical signal and an eighth optical signal with the same power. A photonic multiply-add module, which uses a second modulation voltage corresponding to the solution matrix to perform photonic multiply-add calculations on the seventh optical signal to generate a ninth optical signal; as well as An iterative module uses the eighth optical signal to generate a tenth optical signal corresponding to the proportional term in the iterative formula, uses the fourth optical signal and the first modulation voltage to generate an eleventh optical signal corresponding to the original value term in the iterative formula, uses the fifth optical signal to perform photoelectric conversion on the optical domain accumulation values ​​of the ninth, tenth, and eleventh optical signals to generate a first output current, uses the third optical signal and the first modulation voltage to generate a second output current corresponding to the differential term in the iterative formula, and uses the first and second output currents to generate an output voltage corresponding to the new value of the variable vector.

2. The photon simulation iterative computing device as described in claim 1, characterized in that, The iterative module includes a differentiating unit that uses the first modulation voltage and the third optical signal to generate a twelfth optical signal, and uses the twelfth optical signal, the adjustment voltage, and a fourth modulation voltage corresponding to the momentum factor in the differential term to generate the second output current.

3. The photon simulation iterative computing device as described in claim 2, characterized in that, The differential unit includes: A first phase modulator modulates the phase of the third optical signal according to the first modulation voltage to generate the twelfth optical signal; A first power divider divides the twelfth optical signal into a thirteenth optical signal and a fourteenth optical signal with the same power. A second power divider divides the thirteenth optical signal into a fifteenth optical signal and a sixteenth optical signal with the same power. A second phase modulator adjusts the time delay and phase of the fifteenth optical signal according to the adjustment voltage to generate a seventeenth optical signal; A beam combiner that combines the sixteenth optical signal and the seventeenth optical signal into an eighteenth optical signal; A first photoelectric modulation unit, wherein the second photoelectric modulation unit modulates the power of the eighteenth optical signal according to the fourth modulation voltage to generate a nineteenth optical signal; and The first photoelectric conversion unit couples the fourteenth and nineteenth optical signals through multimode interference and performs photoelectric conversion to generate the second output current.

4. The photon simulation iterative computing apparatus according to any one of claims 1 to 3, characterized in that, The iteration module includes: A multiplexing unit that performs optical domain summation on the ninth, tenth, and eleventh optical signals to generate a twentieth optical signal; and The second photoelectric conversion unit couples the fifth optical signal and the twentieth optical signal through multimode interference and performs photoelectric conversion to generate the first output current.

5. The photonic simulation iterative computing apparatus according to any one of claims 1 to 3, characterized in that, The iterative module includes a scaling unit that modulates the power of the eighth optical signal according to a third modulation voltage corresponding to the annealing factor in the scaling term, thereby generating the tenth optical signal.

6. The photonic simulation iterative computing apparatus according to any one of claims 1 to 3, characterized in that, The iterative module includes a second photoelectric modulation unit, which modulates the power of the fourth optical signal according to the first modulation voltage to generate the eleventh optical signal.

7. The photon simulation iterative computing apparatus according to any one of claims 1 to 3, characterized in that, The iterative module includes a transimpedance amplifier unit, which converts the sum of the first output current and the second output current into a voltage and amplifies it to obtain the output voltage.

8. The photon simulation iterative computing device as described in claim 7, characterized in that, The voltage amplification factor of the transimpedance amplifier unit is adjusted so that the new value of the variable vector is linearly related to the original value of the variable vector, thereby calibrating the iterative formula.

9. The photonic simulation iterative computing apparatus according to any one of claims 1 to 3, characterized in that, The iterative module also includes an optical domain nonlinear unit, which uses a nonlinear function to map the variables of the ninth optical signal to binary or continuous time values, thereby obtaining the twenty-first optical signal. The iterative module uses the twenty-first optical signal instead of the ninth optical signal to generate the first output current.

10. The photonic simulation iterative computing apparatus according to any one of claims 1 to 3, characterized in that, The electrical domain portion of the photonic simulation iterative computing device is electrically connected via an electrically adjustable time delay line, which ensures that the time delay of the electrical domain portion remains consistent.

11. The photonic simulation iterative computing apparatus according to any one of claims 1 to 3, characterized in that, It includes N iteration modules, where N is a positive integer greater than 1 corresponding to the dimension of the variable vector. The first power allocation module generates N second optical signals, N third optical signals, N fourth optical signals, and N fifth optical signals, and provides the N third optical signals, N fourth optical signals, and N fifth optical signals to the corresponding iteration modules.

12. The photon simulation iterative computing device as described in claim 11, characterized in that, The photoelectric modulation module includes N third photoelectric modulation units. Each third photoelectric modulation unit modulates the power of the corresponding second optical signal among the N second optical signals according to the first modulation voltage from the corresponding iteration module among the N iteration modules, thereby generating N sixth optical signals.

13. The photon simulation iterative computing device as described in claim 12, characterized in that, The second power distribution module includes N third power distributors, each of which divides each sixth optical signal into a seventh optical signal and an eighth optical signal, and provides each eighth optical signal to the corresponding iteration module.

14. The photon simulation iterative computing device as described in claim 13, characterized in that, The solution matrix is ​​an N×N matrix. The photon multiply-add module is composed of an array of N×N photoelectric conversion devices. The light transmittance of the photoelectric conversion devices corresponds to the scaling factor in the iterative formula. The photoelectric conversion devices modulate the power of the corresponding seventh optical signal according to the second modulation voltage, and accumulate the modulated optical signals in the same column in the optical domain to generate N ninth optical signals, which are then provided to the corresponding iterative modules.

Images