A packaging method, control device, system and medium of a light quantum chip

Abstract

The application discloses a kind of encapsulation method, control device, system and medium of optical quantum chip, it is related to quantum computing hardware technical field, to solve the phase drift problem caused by thermal crosstalk in large-scale optical quantum chip.The method comprises: the thermal crosstalk matrix between thermo-optic regulation unit is established and compensation value is calculated;Chip is run and the state quantity characterizing phase drift is monitored;When state quantity exceeds threshold value, identify and extract the thermo-optic regulation unit corresponding to the important mode that contributes most to drift as the subset to be updated, only recalibrate and update the thermal crosstalk matrix of the subset, then calculate new compensation value based on the updated matrix.The application realizes low overhead, high efficiency continuous phase compensation by updating only the closed-loop calibration of key subset, and enhances the feasibility of large-scale chip application.

Description

Technical Field

[0001] This invention relates to the field of quantum computer hardware technology, and in particular to a packaging method, control device, system and medium for an optical quantum chip. Background Technology

[0002] Optical quantum chips perform quantum information processing by precisely controlling the phase of the optical field; therefore, the phase control accuracy of their internal phase shifter array directly determines the fidelity of quantum operations. Large-scale optical quantum chips integrate numerous thermo-optical control units (such as microheaters), which generate dynamic and non-uniform thermal fields during operation. Heat diffuses laterally through the chip substrate, creating significant thermal crosstalk. This means that the operation of one control unit can affect the phase of neighboring or even more distant waveguides, causing the chip's phase accuracy to deviate from the requirements for fault-tolerant computation after only a few minutes of operation.

[0003] Existing phase drift compensation schemes mainly have the following problems: One approach is to perform a one-time calibration, which involves calibrating the phase and fixing the compensation parameters after packaging. This method implicitly assumes that the chip's thermal environment is static. This may be suitable for classic optical communication chips with low precision requirements, but for quantum optical chips that require the phase error to be maintained at an extremely low level (e.g., on the order of π / 100), the effective time for a one-time calibration is extremely short due to the continuous dynamic changes in the thermal field during operation, which cannot meet the requirements for continuous and stable operation.

[0004] Another approach is to periodically pause quantum operations and scan and recalibrate all the thermo-optical control units on the chip. This "full array recalibration" method is extremely expensive, requiring a number of measurements proportional to the square of the number of channels. As the chip size (number of channels N) increases, the calibration time increases dramatically, severely impacting effective quantum computing time. When the number of channels reaches a certain scale, this method becomes impractical in engineering.

[0005] Therefore, existing technologies have failed to provide a packaging and calibration method that can effectively control calibration overhead while ensuring high-precision phase control and adapt to the continuous operation requirements of large-scale optical quantum chips. This has become one of the technical bottlenecks restricting the development of practical optical quantum computers. Summary of the Invention

[0006] This invention provides a packaging method, control device, system, and medium for optical quantum chips, which can solve the problem of effectively controlling calibration overhead in related technologies and adapt to the packaging and calibration methods required for continuous operation of large-scale optical quantum chips. The technical solution is as follows: On the one hand, a packaging method for a photonic quantum chip is provided, the photonic quantum chip being configured to satisfy the following conditions: inter-channel crosstalk is lower than a first threshold, phase control error is lower than a second threshold, and the number of packaged channels is greater than a third threshold, the method comprising: Step A: Establish a thermal crosstalk matrix between multiple thermo-optical control units on the photonic quantum chip. The dimension of the thermal crosstalk matrix is ​​determined by the third threshold mentioned above, and the coupling strength distribution of the thermal crosstalk matrix is ​​determined by the first threshold mentioned above. The first threshold represents the crosstalk tolerance between channels, and the third threshold represents the channel size of the chip. Step B: Based on the above thermal crosstalk matrix and the above second threshold, obtain the compensation value of the driving parameter of each thermo-optical control unit, wherein the above second threshold represents the preset phase control accuracy requirement; Step C: Run the photonic quantum chip under the condition of applying the compensation value of the driving parameters, and monitor the state quantity representing the phase drift. When it is determined that the state quantity exceeds the drift threshold corresponding to the second threshold, proceed to steps D to G; when it is determined that the state quantity does not exceed the drift threshold, continue to monitor the state quantity representing the phase drift. Step D: Determine the important modes in the thermal crosstalk matrix that need to maintain update validity based on the first threshold and the third threshold; Step E: Extract the thermo-optical modulation units corresponding to the important modes as the subset to be updated; Step F: Perform recalibration measurement on the subset to be updated, and update the elements in the thermal crosstalk matrix corresponding to the subset to be updated according to the measurement results to obtain the updated thermal crosstalk matrix; Step G: Based on the updated thermal crosstalk matrix, re-execute step B to update the compensation values ​​of the driving parameters of each thermo-optical control unit; then continue to execute step C.

[0007] On the other hand, a photonic quantum chip packaging control device is provided, comprising: A matrix establishment module is used to establish a thermal crosstalk matrix between multiple thermo-optical control units on a photonic quantum chip. The dimension of the thermal crosstalk matrix is ​​determined by a third threshold, and the coupling strength distribution of the thermal crosstalk matrix is ​​determined by a first threshold. The first threshold characterizes the crosstalk tolerance between channels, and the third threshold characterizes the channel size of the chip. The compensation calculation module is used to obtain the compensation value of the driving parameters of each thermo-optical control unit based on the thermal crosstalk matrix and the second threshold, wherein the second threshold represents the preset phase control accuracy requirement. A monitoring module is used to monitor state quantities characterizing phase drift when the photonic quantum chip is operated under conditions where a compensation value of the driving parameters is applied. The judgment module is used to determine whether the state quantity exceeds the drift threshold corresponding to the second threshold; A mode determination module is used to determine, based on the first threshold and the third threshold, important modes in the thermal crosstalk matrix that need to maintain update validity when the judgment module determines that the state quantity exceeds the drift threshold. The subset extraction module is used to extract the thermo-optical modulation units corresponding to the important modes as the subsets to be updated; The matrix update module is used to perform recalibration measurements on the subset to be updated, and update the elements in the thermal crosstalk matrix corresponding to the subset to be updated according to the measurement results, so as to obtain the updated thermal crosstalk matrix. The compensation calculation module is also used to recalculate the compensation value of the driving parameters of each thermo-optical control unit based on the updated thermal crosstalk matrix. The monitoring module is also used to continue operating the quantum chip under the condition of applying a recalculated compensation value and to monitor the state quantities characterizing phase drift.

[0008] On the other hand, a photonic quantum chip packaging system is provided, comprising: A photonic quantum chip integrates multiple thermo-optical control units. The photonic quantum chip is configured to meet the following conditions: inter-channel crosstalk is lower than a first threshold, phase control error is lower than a second threshold, and the number of packaged channels is greater than a third threshold. A packaging substrate supports the photonic quantum chip; a control device, such as the aforementioned photonic quantum chip packaging control device, is electrically connected to the packaging substrate and is used to perform packaging calibration on the photonic quantum chip.

[0009] On the other hand, a computer-readable storage medium is provided, wherein a computer program is stored therein, and when the computer program is executed by a processor, the steps of the above-described packaging method for a photonic quantum chip are implemented.

[0010] On the other hand, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the above-described packaging method for a photonic quantum chip.

[0011] The technical solution provided by this invention can bring at least the following beneficial effects: phase drift is suppressed by establishing a thermal crosstalk matrix and calculating compensation values; at the same time, by continuously monitoring the state variables characterizing phase drift, recalibration is triggered only when the drift exceeds the threshold (instead of fixed-period calibration), and during recalibration, only a subset of the thermal crosstalk matrix is ​​updated instead of the entire matrix, so that the number of measurements for a single calibration is reduced from the order of N² to the order of M² (M is much smaller than N), thereby achieving a dynamic balance between calibration accuracy and calibration overhead. This method embeds calibration into the chip operation process to form a continuous closed loop, which can continuously maintain the phase drift within the quantum fault tolerance threshold, while significantly reducing the calibration time overhead, enabling the packaged calibration method to have scalability to support large-scale optical quantum chips. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a flowchart of a packaging method for a photonic quantum chip according to an embodiment of the present invention; Figure 2 This is a packaging control device for an optical quantum chip provided in one embodiment of the present invention; Figure 3 This is a structural diagram of a packaging system for a photonic quantum chip provided in an embodiment of the present invention; Figure 4 This is a hardware architecture diagram of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0015] As mentioned above, the method in this embodiment achieves a dynamic balance between calibration accuracy and calibration overhead. This method forms a continuous closed loop in the calibration embedded chip operation process, which can continuously maintain the phase drift within the quantum fault tolerance threshold, while significantly reducing the calibration time overhead, thus enabling the packaged calibration method to support the scalability of large-scale optical quantum chips.

[0016] The specific implementation of the above concept is described below.

[0017] Please refer to Figure 1 This invention provides a packaging method for a photonic quantum chip, wherein the photonic quantum chip is configured to satisfy the following conditions: inter-channel crosstalk is lower than a first threshold, phase control error is lower than a second threshold, and the number of packaged channels is greater than a third threshold. The method includes: Step A: Establish a thermal crosstalk matrix among multiple thermo-optical control units on the optical quantum chip. The dimension of the thermal crosstalk matrix is ​​determined by the third threshold, and the coupling strength distribution of the thermal crosstalk matrix is ​​determined by the first threshold. The first threshold characterizes the crosstalk tolerance between channels, and the third threshold characterizes the channel size of the chip. In this embodiment, the thermal crosstalk matrix refers to a numerical matrix used to quantitatively describe the degree to which the heat generated by a thermo-optical control unit (such as a microheater) on a quantum chip affects the optical phase of other waveguide channels. Specifically, the rows of this matrix can correspond to the various waveguide channels on the chip, and the columns can correspond to the various thermo-optical control units. Each element in the matrix (e.g., Then, ) represents the amount of phase change caused by the j-th thermo-optical control unit to the i-th waveguide channel under a unit driving amount (such as unit power or voltage).

[0018] Step B: Based on the thermal crosstalk matrix and the second threshold, obtain the compensation value of the driving parameter of each thermo-optical control unit, wherein the second threshold represents the preset phase control accuracy requirement; Step C: Run the photonic quantum chip under the condition of applying the compensation value of the driving parameters, and monitor the state quantity representing the phase drift. When it is determined that the state quantity exceeds the drift threshold corresponding to the second threshold, proceed to steps D to G; when it is determined that the state quantity does not exceed the drift threshold, continue to monitor the state quantity representing the phase drift. Step D: Determine the important modes in the thermal crosstalk matrix that need to maintain update validity based on the first threshold and the third threshold; Step E: Extract the thermo-optical modulation units corresponding to the important modes as the subset to be updated; Step F: Perform recalibration measurement on the subset to be updated, and update the elements in the thermal crosstalk matrix corresponding to the subset to be updated according to the measurement results to obtain the updated thermal crosstalk matrix; Step G: Based on the updated thermal crosstalk matrix, re-execute step B to update the compensation values ​​of the driving parameters of each thermo-optical control unit; then continue to execute step C.

[0019] In this embodiment, the state quantity refers to one or more physical quantities that can be monitored in real time during chip operation, and whose changes can indirectly or directly reflect the on-chip phase drift. The state quantity can be the output spectral drift of one or more on-chip integrated reference interferometers (such as a Mach-Zehnder interferometer), the real-time reading of a temperature sensor deployed at a critical location on the chip, or the quantum logic gate fidelity estimated through specific quantum operations (such as random benchmarking).

[0020] In this embodiment, the key modes refer to the main modes that contribute the most to the overall thermal crosstalk effect after modal decomposition (such as eigenvalue decomposition or singular value decomposition) of the thermal crosstalk matrix. These modes typically correspond to the low-frequency spatial components in the thermal diffusion process, and their cumulative contribution rate exceeds a preset percentage (such as 90%), representing the main source and most critical influence path of thermal crosstalk.

[0021] In this embodiment, the subset to be updated refers to a small set of units selected from all the thermo-optical control units of the quantum chip according to preset rules. The units in this set are considered to be the main contributors to the current phase drift exceeding the standard. Therefore, in the adaptive calibration process, only the units in this subset need to be recalibrated, instead of measuring all units, thereby significantly reducing calibration overhead.

[0022] This invention provides a packaging method for optical quantum chips to address the technical problem of phase drift caused by thermal crosstalk during the operation of large-scale optical quantum chips after packaging, due to the dynamic non-uniform thermal field generated by the simultaneous operation of multiple on-chip thermo-optical control units. This thermal crosstalk effect affects the fidelity of quantum computing. Existing technologies typically employ one-time calibration or periodic full-array recalibration. The former cannot cope with dynamic drift, while the latter becomes impractical due to excessive calibration overhead as chip scale increases. The method proposed in this application achieves continuous suppression of phase drift through a closed-loop adaptive calibration mechanism, while minimizing calibration overhead.

[0023] Example 1 The method provided in this application's embodiments is based on a continuously running closed-loop calibration process. The following describes... Figure 1 The execution method of each step is shown.

[0024] First, for step A, a thermal crosstalk matrix is ​​established between multiple thermo-optical control units on the optical quantum chip. The dimension of the thermal crosstalk matrix is ​​determined by the third threshold, and the coupling strength distribution of the thermal crosstalk matrix is ​​determined by the first threshold. The purpose of this step is to obtain a baseline model of the thermal interactions between all thermo-optical control units on the chip. By establishing this matrix, it is possible to mathematically describe the extent to which the heating behavior of any control unit affects the phase of all other optical paths, which forms the basis for subsequent precise compensation.

[0025] After obtaining the thermal crosstalk matrix, the method proceeds to step B, where the compensation values ​​of the driving parameters of each thermo-optical control unit are obtained based on the thermal crosstalk matrix and the second threshold. The second threshold here represents the preset phase control accuracy requirement. To counteract the thermal crosstalk effect, the compensation value of the driving parameters to be applied to each thermo-optical control unit needs to be calculated based on the thermal crosstalk matrix and the second threshold. This step actively applies a pre-distorted driving signal to counteract the impending thermal crosstalk in advance, thereby maintaining the stability of the target phase macroscopically. Through this step, the theoretical model can be transformed into actual control commands, solving the problem of how to actively compensate for thermal crosstalk using a thermal crosstalk model.

[0026] Next, the method proceeds to step C, where the photonic quantum chip is run under the condition of applying the compensation value of the driving parameters, and the state quantity characterizing the phase drift is monitored. When the state quantity exceeds the drift threshold corresponding to the second threshold, steps D to G are executed; when the state quantity does not exceed the drift threshold, the state quantity characterizing the phase drift continues to be monitored. After applying the compensation values ​​for the driving parameters of each thermo-optical control unit obtained in step B above, the quantum chip is allowed to operate normally, and during this period, one or more state variables that can characterize phase drift are continuously monitored. The principle behind this step is that even with pre-compensation, phase drift will still slowly accumulate due to factors such as ambient temperature fluctuations, laser power jitter, and the approximation of the model itself. Therefore, a real-time feedback loop is necessary to monitor the actual operating status of the system. Through continuous monitoring, the dynamics of phase drift can be grasped in real time, providing data input for subsequent decision-making.

[0027] During the monitoring process, a crucial judgment is made: whether the monitored state quantity exceeds a preset drift threshold. This drift threshold is determined by a second threshold, which is essential for the fidelity of quantum computing and characterizes the required phase control accuracy. The drift threshold is a critical phase drift value not exceeding the second threshold. When the state quantity representing the phase drift reaches this critical value, recalibration is triggered to bring the phase error back to the tolerance range before it exceeds the second threshold. If the state quantity does not exceed the drift threshold, it indicates that the current compensation strategy is still effective and the system is stable. In this case, no intervention is required, and the monitoring of the state quantity representing the phase drift continues, forming a stable operating cycle. The purpose of this design is to avoid unnecessary calibration operations, save computation time and energy resources, and solve the problem of numerous overcalibrations in existing periodic calibration schemes.

[0028] When the state quantity exceeds the drift threshold corresponding to the second threshold, steps D to G are executed. If the state quantity exceeds the drift threshold, it indicates that the current compensation model is insufficient to suppress the actual phase drift. At this time, the system triggers the adaptive update mechanism.

[0029] Step D involves determining the key modes in the thermal crosstalk matrix that require continued update validity based on the first and third thresholds. Specifically, based on a preset first threshold characterizing inter-channel crosstalk tolerance and a third threshold characterizing the channel size of the chip, the key modes in the thermal crosstalk matrix that contribute the most to the current drift are identified. This step is based on the physical properties of thermal diffusion, where thermal crosstalk effects are typically dominated by a few major, wide-ranging modes. By identifying these key modes, the focus of the problem can be narrowed down from the entire complex N×N matrix to a few critical factors.

[0030] Step E involves extracting the thermo-optical modulation units corresponding to the important modes as a subset to be updated. Specifically, after identifying the important modes, a few thermo-optical modulation units corresponding to these important modes are extracted to form a subset to be updated. The significance of this step lies in reducing the calibration problem of the entire array to a "miniature" calibration problem targeting only a few key units. This significantly reduces the number of units that need to be measured, thereby directly reducing the time and data volume required for calibration.

[0031] Step F involves recalibrating the subset to be updated and updating the elements in the thermal crosstalk matrix corresponding to that subset based on the measurement results, thus obtaining an updated thermal crosstalk matrix. Specifically, this step only recalibrates the subset to be updated, which is much smaller than the entire set of cells, and updates only the elements in the thermal crosstalk matrix related to that subset based on the latest measurement results, thereby obtaining an updated thermal crosstalk matrix. This step significantly improves calibration efficiency by replacing global reconstruction with precise local updates. By updating only a portion of the matrix, the validity of the model for the current system state is ensured, while avoiding the huge overhead of remeasuring the entire N×N matrix.

[0032] Step G: Based on the updated thermal crosstalk matrix, re-execute step B to update the compensation values ​​of the driving parameters of each thermo-optical control unit; then continue to execute step C.

[0033] Specifically, based on this newly updated and more accurate thermal crosstalk matrix, the process returns to step B to calculate a new and more effective set of driving parameter compensation values, which are then applied to the chip. Step C, involving operation and monitoring, then continues. This completes a full closed-loop adaptive calibration process. Through this closed loop of monitoring, judgment, updating, and compensation, this method ensures that the phase of the optical quantum chip remains locked within an extremely high precision range during long-term operation, thus providing crucial packaging support technology for achieving large-scale, high-fidelity optical quantum computing.

[0034] Furthermore, in a preferred embodiment, the structure of the thermal crosstalk matrix is ​​specifically defined. Please refer to [link / reference]. Figure 1 The quantum chip 10 integrates a large number of thermo-optical control units 11, each of which may have a thermal effect on other units. The rows of the thermal crosstalk matrix correspond to the various waveguide channels on the chip, while the columns correspond to the various thermo-optical control units 11. Each element in the matrix, for example... The physical meaning of is the phase influence coefficient generated on the i-th waveguide channel when the j-th thermo-optical control unit 11 applies a unit driving amount. The dimension of this matrix is ​​N×N, where N is the number of encapsulated channels, and its size is limited by the third threshold. Through this explicit definition, a clear and standardized mathematical framework is provided for the modeling and analysis of thermal crosstalk effects, enabling subsequent compensation calculations and modal analysis to be carried out accurately based on this framework.

[0035] In another preferred embodiment, the process of establishing the thermal crosstalk matrix is ​​optimized. Specifically, the complete measurement operation for establishing the thermal crosstalk matrix in step A is performed only once during the first power-on of the photonic quantum chip 10. Although this initial full matrix measurement is time-consuming, its result is stored as a comprehensive baseline model. During subsequent chip operation, when an update is triggered, the system no longer rebuilds the entire N×N matrix, but only performs local measurements and updates on the subset to be updated. This strategy significantly reduces the average calibration overhead during operation by amortizing the high initial calibration cost over the entire chip lifecycle, solving the problem of low time utilization caused by periodic full matrix calibration in the prior art.

[0036] In one optional implementation, to suppress thermal crosstalk at the physical level, a crosstalk suppression structure 12 is also provided between the waveguides of the photonic quantum chip 10. The crosstalk suppression structure 12 can take various forms, such as a subwavelength grating structure, a trench isolation structure etched between waveguides, or a multilayer dielectric structure formed by stacking materials with different refractive indices. The purpose of these structures is to create a thermally resistive or optically insulating band between adjacent or closely spaced waveguides, hindering the lateral diffusion of heat. The geometric parameters of the crosstalk suppression structure 12, such as the grating period, trench depth, or material refractive index, are collaboratively designed based on a first threshold and a third threshold. By introducing these physical structures during the chip design stage, the values ​​of off-diagonal elements in the thermal crosstalk matrix can be effectively reduced, making the matrix more diagonally dominant. This not only directly improves the chip's performance but also allows subsequent subset updates to operate under more favorable conditions, achieving a hardware-software co-optimization effect.

[0037] Furthermore, the specific form of the state quantity can be chosen in various ways. In a preferred embodiment, the state quantity is the output spectral drift of the on-chip reference interferometer. Specifically, one or more Mach-Zehnder interferometers that do not participate in quantum computing and are specifically used as sensors can be additionally integrated on the photonic quantum chip 10. By injecting stable probe light into the reference interferometer and monitoring the drift of the interference fringes in its output spectrum, the local average phase drift of the chip can be accurately deduced. The advantage of this method is high measurement accuracy and strong correlation with the actual phase drift.

[0038] In another implementation, the state variable can be a reading from an on-chip temperature sensor. By deploying a miniature temperature sensor array at a critical location on the chip, the dynamic changes in the thermal field can be directly measured, and the phase drift can then be estimated using a thermo-optic coefficient model. This approach is simple to implement and has a fast response time. In yet another implementation, the state variable can also be a fidelity estimate of the quantum operation result. By periodically executing standardized quantum algorithms (such as random benchmarks) and evaluating the fidelity of their outputs, the final impact of phase drift on the computational result can be directly measured. This method is the most direct, but also has the highest measurement overhead. In practice, one or more combinations of these state variables can be selected based on the specific application's requirements for accuracy and overhead.

[0039] In one specific implementation, the precision requirements for phase control are quantified. A second threshold is set at a phase control error of less than π / 100. This value is not arbitrarily set but originates from the error rate requirements of physical qubits in quantum error correction theory. To achieve fault-tolerant quantum computing, the error rate of logic gates needs to be controlled at an extremely low level, which directly translates into stringent requirements for the precision of the underlying physical phase control. By setting the second threshold to π / 100, the technical objective of the method of this invention is to meet the practical application requirements of large-scale fault-tolerant optical quantum computers.

[0040] In a preferred embodiment, a specific technical solution is provided for determining the subset to be updated. This solution is achieved by performing mode decomposition on the thermal crosstalk matrix. Specifically, when an update is needed, the current thermal crosstalk matrix C is subjected to singular value decomposition (SVD) or eigenvalue decomposition to obtain a series of modes, typically singular values ​​and their singular vectors or eigenvalues ​​and their eigenvectors. Then, the contribution rates of these modes are accumulated from largest to smallest until the cumulative contribution rate exceeds a preset percentage (e.g., 90%). The modes with the highest contribution rates are identified as important modes. Finally, the thermo-optical control units 11 corresponding to these important modes are extracted. These thermo-optical control units are typically the units with larger components in the singular vectors or eigenvectors, forming the subset to be updated. Furthermore, the preset percentage can be dynamically adjusted according to a first threshold; the more stringent the crosstalk requirement, the higher the percentage, to include more modes to ensure accuracy. Through this mode decomposition method, the main physical source causing the current drift can be found mathematically in an optimized manner, achieving efficient utilization of calibration resources.

[0041] In another alternative implementation, a computationally simpler method is provided for determining the subset to be updated. This method is achieved by identifying and excluding thermo-optical control units 11 whose influence is below a preset threshold. Specifically, for each column of the thermal crosstalk matrix C, its norm (e.g., L2 norm) is calculated, which can be considered as a measure of the influence of the corresponding thermo-optical control unit 11 on the entire chip. Then, all units with influence below a preset threshold are identified as minor contributors and excluded from the subset to be updated, retaining only those units with high influence. Compared to mode decomposition, this influence screening method has less computation and is simpler to implement. Similarly, the preset threshold can also be dynamically adjusted according to a first threshold; the stricter the crosstalk requirement, the lower the threshold is set to retain more potential contributors. These two methods offer different options for the trade-off between accuracy and overhead.

[0042] To further reduce the overall power consumption and computational burden of the system, this invention also provides an adaptive monitoring optimization strategy. In a preferred embodiment, when the monitored state variables do not exceed the drift threshold after returning to step C multiple times (e.g., 20 times), the system is determined to have entered a thermally stable state. At this time, the time interval for performing monitoring operations can be adaptively extended (e.g., from once every 10 milliseconds to once every 100 milliseconds), or the monitoring frequency can be directly reduced (e.g., from 100Hz to 10Hz). The magnitude of this adjustment can be preset according to a first threshold and a third threshold. By slowing down when the system is stable and speeding up when the system is dynamically changing, intelligent management of monitoring resources is achieved, significantly reducing the average power consumption during long-term operation without sacrificing system responsiveness.

[0043] Example 2 The following embodiment uses a silicon-based photonic quantum chip as an example. The chip integrates N thermo-optical modulation units (N=64), and phase modulation is achieved using the thermo-optical effect. The heater material is platinum, and the thermo-optical coefficient dn / dT≈1.8×10 -4 Phase modulation is achieved using / K. The chip inputs and outputs optical signals through an optical fiber array, and a mode converter is used to achieve low-loss coupling between the fiber and the chip. The chip is mounted on a package substrate, which integrates a thermoelectric cooler (TEC) to maintain the overall temperature stability of the chip and provides an electrical interface to connect to the heater drive circuit. The controller (FPGA or MCU) is responsible for executing the calibration algorithm of this invention.

[0044] In this embodiment, the optical quantum chip is configured to meet the following conditions: inter-channel crosstalk is lower than a first threshold (-30dB in this embodiment), phase control error is lower than a second threshold (π / 100 in this embodiment), and the number of packaged channels is greater than a third threshold (N≥64 in this embodiment).

[0045] It should be noted that the second threshold (π / 100) is the maximum phase error boundary allowed for the operation of the quantum chip, determined by the quantum fault tolerance requirements; the drift threshold is a phase drift critical value set based on the second threshold to trigger recalibration, and its value is not greater than the second threshold. In this embodiment, the mapping relationship between the drift threshold and the second threshold is: the drift threshold equals the second threshold multiplied by a preset coefficient, which is a real number between 0.8 and 1.0, i.e., drift threshold = coefficient × π / 100. When the coefficient is 1.0, the drift threshold equals the second threshold, and recalibration is triggered when the phase drift reaches the quantum fault tolerance limit; when the coefficient is 0.8, the drift threshold = 0.8 × π / 100 = π / 125, and recalibration is triggered in advance when the phase drift is close to but has not yet reached the quantum fault tolerance limit, reserving a safety margin for the system. In this embodiment, when the coefficient is 1.0, the drift threshold equals π / 100.

[0046] Step A: Establish the thermal crosstalk matrix between multiple thermo-optical control units on the optical quantum chip; Specifically, the controller sequentially applies a preset voltage sequence (such as 0V, 1V, 2V, ..., 10V) to N heaters. At each drive level, the controller reads the output optical power of all N waveguide channels through an on-chip monitoring detector and uses the Mach-Zehnder interferometer transfer function relationship to convert the power change into a phase change.

[0047] Specifically, the Mach-Zehnder interferometer transfer function is:

[0048] in, This indicates the optical power at the output of the interferometer. This indicates the optical power at the input of the interferometer. This represents the phase difference between the two arms of the interferometer. The measurement is performed for the i-th waveguide channel and the j-th heater when the m-th drive level is applied. ,in This indicates the voltage applied to the j-th heater. The phase of the i-th channel, This indicates that the j-th heater is subjected to an initial voltage. The phase of the i-th channel.

[0049] Therefore, the phase influence of the j-th heater on the i-th channel at the m-th driving level is:

[0050] in, This indicates the voltage applied to the j-th heater. The measured output optical power of the i-th channel. This indicates that the j-th heater is subjected to an initial voltage. The measured output optical power of the i-th channel.

[0051] For each heater j, its phase influence at multiple drive levels For drive power Linear fitting of the rate of change yields the phase influence coefficient per unit driving quantity:

[0052] in, This represents the element in the i-th row and j-th column of the thermal crosstalk matrix C (i.e., the phase influence coefficient of the j-th heater on the i-th waveguide channel). This indicates the incremental driving power applied to the j-th heater. The phase change generated by the i-th channel is used to initially calibrate the thermal crosstalk matrix. The unit is rad / mW.

[0053] After the above measurements and calculations, the N×N thermal crosstalk matrix is ​​obtained:

[0054] Where N represents the number of encapsulated channels (N=64 in this embodiment), the row index i (i=1,2,…,N) of the matrix corresponds to the waveguide channel number, and the column index j (j=1,2,…,N) corresponds to the thermo-optical control unit number. The matrix dimension N×N is determined by a third threshold (the number of encapsulated channels is greater than the third threshold, which is N≥64 in this embodiment). The amplitude distribution of each element in the matrix reflects the thermal coupling strength distribution: the lower the crosstalk tolerance required by the first threshold (crosstalk between channels is less than -30dB), the larger the proportion of weakly coupled elements between distant channels in the matrix, and the denser the matrix.

[0055] This operation is performed once when the chip is first powered on, and the measurement result C is stored in the controller's non-volatile memory. On subsequent power-ups, it is directly loaded from the non-volatile memory without the need for re-measurement.

[0056] Step B: Based on the thermal crosstalk matrix and the second threshold, obtain the compensation value of the driving parameters of each thermo-optical control unit; The controller obtains the N-dimensional target phase configuration vector required for the current quantum operation. ,in , Let represent the target phase value that the i-th waveguide channel needs to achieve, and let the superscript T denote the transpose of the matrix or vector. Using the N×N thermal crosstalk matrix C established in step A, solve the linear equation:

[0057] Where P represents the N-dimensional compensation power vector to be solved. , This represents the compensation power vector that needs to be applied to the j-th heater. Elements of the thermal crosstalk matrix C. This represents the phase effect on the i-th waveguide channel when the j-th heater applies unit power, therefore matrix multiplication... This represents the total phase change generated in each waveguide channel when all heaters work together.

[0058] Among them, equation This means that by selecting a suitable heater power vector P, the total phase change generated by the heater array across all waveguide channels is exactly equal to the target phase configuration vector. This cancels out the phase error caused by thermal crosstalk.

[0059] When the thermal crosstalk matrix C is a full-rank matrix and invertible, the controller can be solved directly:

[0060] in, Let represent the inverse of matrix C.

[0061] When the thermal crosstalk matrix C is a singular matrix (i.e., the determinant of the matrix is ​​zero and it is not invertible) or close to singular (i.e., the condition number is greater than a preset threshold and the matrix is ​​ill-conditioned), the following method is used to solve it: (1) Least squares solution:

[0062] in, The pseudo-inverse of matrix C is represented by the Moore-Penrose generalized inverse.

[0063] (2) Tikhonov regularized least squares solution:

[0064] in, Describes the transpose of matrix C. This represents the regularization parameter (preset as a positive real number, used to control the balance between solution stability and accuracy; in this embodiment...). ), This represents an N×N identity matrix (a square matrix with diagonal elements of 1 and off-diagonal elements of 0), and the superscript -1 indicates the inverse operation of the matrix.

[0065] The Tikhonov regularization method described above introduces a regularization term. Improvement Matrix The condition number is used to avoid the solution being overly sensitive to measurement noise due to matrix ill-conditioning.

[0066] In this embodiment, the target phase configuration vector The system is 64-dimensional (N=64), and the thermal crosstalk matrix C is 64×64. The controller calculates the 64-dimensional compensation power vector. ,in This represents the compensation power vector that needs to be applied to the j-th heater. The calculated 64 compensation power vectors are output to the respective heater drive circuits via a digital-to-analog converter (DAC). The drive voltage amplitude is determined by the power resolution requirement corresponding to the second threshold (phase control error less than π / 100)—π / 100 phase accuracy requires the DAC to have at least 12 effective bits to ensure that the power control accuracy of each heater is sufficient to meet the phase adjustment resolution requirements. Step C: The photonic quantum chip is operated under the condition of applying the compensation values ​​of the drive parameters, and the state variables characterizing phase drift are monitored; the controller applies the compensation power vector... Under certain conditions, quantum operations are initiated. During operation, the controller reads the output spectral shift at a preset sampling frequency (100Hz in this embodiment) through the on-chip reference interferometer.

[0067] In this embodiment, an on-chip integrated unbalanced Mach-Zehnder interferometer is used as the phase reference unit. The two arms of the interferometer have a preset optical path difference ΔL (ΔL=100μm in this embodiment) and input continuous probe light with a wavelength λ0=1550nm.

[0068] Let the output light intensity of the interferometer be The input light intensity is The relationship between the output light intensity and the phase difference between the two arms of the interferometer satisfies:

[0069] in, This indicates the optical power at the output of the interferometer. ΔL represents the optical power at the input of the interferometer, ΔL represents the optical path difference between the two arms of the interferometer (unit: μm), and λ represents the real-time wavelength of the probe light (unit: nm). The state quantity representing the additional phase shift caused by temperature change (unit: rad) is the state quantity characterizing the phase shift. The specific manifestation of the reference interferometer unit monitoring scheme in this embodiment.

[0070] When the chip temperature changes, the two arms of the interferometer experience an additional phase shift due to the thermo-optical effect; this phase shift is a state variable. The relationship between the temperature change ΔT and the temperature change is as follows:

[0071] in, The state variable representing phase drift (unit: rad) and λ represent the wavelength of the probe light (unit: nm). The thermo-optic coefficient of the waveguide material is represented (in this embodiment, the thermo-optic coefficient of silicon material is 1.8 × 10⁻⁶). -4 / K), ΔT represents the temperature change (unit: K). This indicates the optical length of a single arm of the interferometer (unit: μm).

[0072] The controller monitors fluctuations in output light intensity and calculates the phase drift state variables in real time. Specifically, the phase drift state variables are solved by inversely from the above output light intensity formula:

[0073] in, The state quantity representing the phase drift calculated in real time (unit: rad). This represents the interferometer output light intensity measured at the current moment. Given the known input light intensity, ΔL is the fixed optical path difference between the two arms of the interferometer, and λ0 is the nominal wavelength of the input probe light (1550 nm). The first term on the right-hand side of the equation... The second term represents the total phase difference between the two arms of the interferometer at the current moment. This represents the initial phase difference between the two arms of the interferometer when there is no temperature drift. Subtracting the two gives the state quantity of the phase drift caused by temperature change.

[0074] In one specific implementation, the drift threshold is π / 100 (corresponding to the second threshold). When When the phase drift is less than π / 100, the controller determines that the current phase drift is within the quantum fault-tolerant range, does not perform any calibration operation, and continues to operate and monitor at the current sampling frequency. When the phase drift is greater than or equal to π / 100, the state variable of the phase drift exceeds the drift threshold π / 100 corresponding to the second threshold. The controller determines that the current phase drift is close to or has reached the quantum fault tolerance limit, triggers the recalibration process, and jumps to steps D to G. Step D: Determine the important modes in the thermal crosstalk matrix that need to maintain update validity based on the first threshold and the third threshold. The controller performs eigenvalue decomposition on the current N×N thermal crosstalk matrix C:

[0075] Where C represents the current N×N hot crosstalk matrix (N=64 in this embodiment), U represents the N×N eigenvector matrix (each column is an eigenvector of C), Σ represents the N×N diagonal eigenvalue matrix (the diagonal elements are the eigenvalues ​​of C, and the off-diagonal elements are zero), and U^T represents the transpose of U.

[0076] After eigenvalue decomposition, N eigenvalues ​​are obtained, which are arranged in descending order as follows:

[0077] in, (i=1,2,…,N) represents the i-th eigenvalue. The largest eigenvalue, The smallest eigenvalue is denoted as . The magnitude of the eigenvalue reflects the importance of the corresponding characteristic mode in the thermal crosstalk matrix; the larger the eigenvalue, the greater the contribution of the characteristic mode to the thermal crosstalk coupling effect.

[0078] A preset energy threshold η is determined based on a first threshold (inter-channel crosstalk tolerance). In this embodiment, the first threshold is -30dB, and the corresponding preset energy threshold is η=90%. The controller calculates the cumulative contribution rate of the first k feature values. :

[0079] in, The numerator represents the cumulative contribution rate of the first k eigenvalues. The denominator represents the sum of the first k eigenvalues. It represents the sum of all N eigenvalues ​​(i.e., the total energy of the matrix). This reflects the extent to which the first k characteristic modes cover the total coupling effect of the thermal crosstalk matrix.

[0080] The controller found that the condition was met. The minimum k value ≥ η is the number of important modes that need to be retained. In this embodiment, k = 8, meaning the first 8 characteristic modes are identified as important modes, which contribute more than 90% of the total coupling effect of the thermal crosstalk matrix. The value of k is also constrained by a third threshold (number of encapsulation channels N ≥ 64). When the number of channels N = 64, the ratio of k to N is approximately 1 / 8; when the number of channels increases to N = 128, the value of k increases accordingly (k ≈ 14-16), but the ratio further decreases (approximately 1 / 8 to 1 / 9), reflecting a further improvement in calibration efficiency during large-scale integration.

[0081] It should be noted that the mapping relationship between the preset energy threshold η and the first threshold (inter-channel crosstalk tolerance) is as follows: When the crosstalk tolerance is stringent (e.g., the first threshold is increased to -40dB), long-distance weak coupling cannot be ignored, so the preset energy threshold needs to be increased to η=95%. In this case, the number of modes that need to be retained increases (k≈10-12) to ensure the modeling accuracy of long-distance weak coupling. When the crosstalk tolerance is more lenient (e.g., the first threshold is relaxed to -20dB), some weakly coupled modes can be appropriately discarded to reduce calibration overhead, and the preset energy threshold is reduced to η=80%. In this case, the number of modes that need to be retained decreases (k≈5-6). The value of η is uniquely determined by the first threshold, and the mapping relationship between the two is pre-calibrated experimentally and stored in the controller's non-volatile memory.

[0082] Step E: Extract the thermo-optical modulation units corresponding to the important modes as the subset to be updated; Based on the number of important modes k determined in step D (k=8 in this embodiment), the controller extracts the first k eigenvectors from the eigenvector matrix U, denoted as... ,in Let j represent the j-th eigenvector (j=1,2,…,k), which is an N-dimensional column vector. Each component of the vector corresponds to the participation weight of a thermo-optical control unit in that eigenmode.

[0083] For the j-th feature vector The controller finds the row index corresponding to the component with the largest absolute value, i.e.:

[0084] in, The row index of the largest component in the j-th eigenvector. Represents the j-th eigenvector The i-th component (i.e., the participation weight of the i-th thermo-optical control unit in the j-th characteristic mode), |·| represents the absolute value operation. The value of is between 1 and N, corresponding to the physical number of one of the N thermo-optical control units on the quantum chip.

[0085] Perform the above operation on j=1,2,…,k respectively to obtain k row indices:

[0086] The thermo-optical control units corresponding to these row indices are the k units that contribute the most to the total coupling effect of the thermal crosstalk matrix. The controller uses these k thermo-optical control units to form a subset to be updated: ; in, This represents the physical number of the j-th selected thermo-optical control unit ( = S represents the subset to be updated, consisting of these unit numbers, with a subset size of M=k. In this embodiment, k=8, meaning that the subset S to be updated contains 8 thermo-optical control units.

[0087] Step F: Perform recalibration measurement on the subset to be updated, and update the elements in the thermal crosstalk matrix corresponding to the subset to be updated according to the measurement results to obtain the updated thermal crosstalk matrix; The controller performs recalibration measurements only on the M thermo-optical control units (M=8 in this embodiment) in the subset S to be updated. The specific process of the recalibration measurement is as follows: the controller sequentially drives each thermo-optical control unit in the subset to be updated, applying a preset test voltage sequence (e.g., 0V, 0.5V, 1.0V, ..., 5.0V) to each unit. At each drive level, the controller records the output optical power changes of all N waveguide channels through an on-chip monitoring detector, and converts the power changes into phase changes according to the method described in step A. A total of M×N measurements are performed; in this embodiment, this is 8×64=512 measurements.

[0088] Update the column elements in the thermal crosstalk matrix C corresponding to the subset S to be updated based on the measurement results. Specifically, for each unit in subset S, let its physical number be... (j=1,2,…,M), update the phase influence coefficient obtained from the recalibration measurement of this unit to the thermal crosstalk matrix C. The columns (i.e., those corresponding to the physical number of the unit) remain unchanged, while the remaining NM columns not in subset S remain unchanged. After the update, the updated thermal crosstalk matrix is ​​obtained. .

[0089] As an alternative to the mode decomposition method (steps D to F), the subset to be updated can also be determined using the influence screening method. This method is independent of the mode decomposition method and can be used alone for subset determination in steps D to F. Specifically: The controller calculates the L2 norm of each column of the thermal crosstalk matrix C:

[0090] Where C(:,j) represents the j-th column of the thermal crosstalk matrix C (i.e., the phase influence coefficient vector of the j-th thermo-optical control unit on all N waveguide channels). The L2 norm (Euclidean norm) of the column vector is represented, i represents the index of the waveguide channel (i=1,2,…,N), and j represents the index of the thermo-optical control unit (j=1,2,…,N). This represents the phase influence coefficient of the j-th thermo-optical modulation unit on the i-th waveguide channel. This norm value characterizes the overall phase influence intensity of the j-th thermo-optical modulation unit on all N waveguide channels—the larger the norm value, the greater the influence of this unit in the thermal crosstalk matrix.

[0091] An influence threshold δ is set. The influence threshold δ is dynamically determined based on a first threshold (inter-channel crosstalk tolerance). In this embodiment, when the first threshold is -30dB, δ = 0.05 × ||C(:,j)||2, where This means taking the maximum value for all j (j=1,2,…,N), i.e., δ is 5% of the maximum norm value.

[0092] Controller identifies column norm The thermo-optical control units corresponding to columns with influence thresholds δ are excluded from the subset to be updated, and the remaining units constitute the subset S to be updated. In this embodiment, 12 out of 64 units are identified as low-influence units and excluded, and the remaining 52 units constitute the subset S to be updated (i.e., M=52).

[0093] The mapping relationship between the influence threshold δ and the first threshold is as follows: when the crosstalk requirement is -30dB, δ is set to 5% of the maximum Leveno value; when the crosstalk requirement is increased to -40dB, δ is reduced to 2% of the maximum Leveno value (to retain more weakly influential elements to ensure the modeling accuracy of long-distance weak coupling); when the crosstalk requirement is relaxed to -20dB, δ is increased to 10% of the maximum Leveno value (to exclude more low-influence elements and reduce calibration overhead). The mapping relationship between δ and the first threshold is pre-calibrated and stored in the controller's non-volatile memory.

[0094] The controller is based on the updated N×N thermal crosstalk matrix. Repeat step B. Specifically, the controller obtains the N-dimensional target phase configuration vector required for the current quantum operation. Solve the linear equation:

[0095] in, This represents the updated N×N thermal crosstalk matrix obtained in step F. This represents the updated N-dimensional compensation power vector to be solved. , This represents the compensation power vector that needs to be applied to the j-th heater after the update. Represents an N-dimensional target phase configuration vector. , This represents the target phase value that the i-th waveguide channel needs to achieve.

[0096] The updated N-dimensional compensation power vector is obtained by solving the problem. Then, the controller will assign each compensated power vector The parameters are applied to the corresponding j-th heater drive circuit. The controller then jumps back to step C and continues to monitor the phase drift state at the current sampling frequency, forming a continuous closed loop.

[0097] If the state quantity is determined not to exceed the drift threshold in step C, the controller returns to step C and continues to monitor the state quantity that represents the phase drift.

[0098] Preferably, the controller maintains a triggerless counter cnt, initially set to 0. Each time step C is executed and the state quantity is determined not to exceed the drift threshold, cnt is incremented by 1. When the state quantity does not exceed the drift threshold for Q consecutive monitoring cycles (i.e., cnt reaches Q, in this embodiment Q=20), the controller reduces the monitoring frequency.

[0099] The adjustment range of the monitoring frequency is jointly determined by a first threshold (characterizing the tolerance of crosstalk between channels) and a third threshold (characterizing the chip channel size): when the crosstalk tolerance is strict (first threshold < -30dB) and the number of channels is large (N=64), the risk of phase drift is high, and the frequency reduction is limited to 50% (from 100Hz to 50Hz); when the crosstalk tolerance is lenient (first threshold > -20dB), the frequency reduction can reach 90% (from 100Hz to 10Hz). In this embodiment, the first threshold is -30dB, the third threshold is N=64, and the controller's frequency control strategy is: when cnt=20, the monitoring frequency is reduced from 100Hz to 50Hz; when cnt=50, it is further reduced to 10Hz.

[0100] when When the drift threshold is exceeded again, cnt returns to zero, and the monitoring frequency returns to 100Hz.

[0101] Preferably, a crosstalk suppression structure is disposed between the waveguides of the photonic quantum chip. This structure is formed in the region between adjacent waveguides during the chip photolithography manufacturing stage to reduce static optical evanescent wave coupling. The crosstalk suppression structure is at least one of the following: (1) Subwavelength grating structure: Subwavelength gratings are etched between adjacent silicon waveguides. The grating period Λ = 200 nm, which is less than the operating wavelength. =1550nm is half the equivalent wavelength in the SiO2 cladding, that is ,in Indicates the effective refractive index of the waveguide (in this embodiment) ), The etching depth h = 220 nm, equal to the thickness of the silicon core layer. The duty cycle of the grating (the ratio of ridge width to period) is 0.5.

[0102] (2) Trench isolation structure: Deep trenches are etched between adjacent waveguides. The trench depth H = 10 μm, which is greater than the sum of the silicon core layer thickness (220 nm) and the buried oxide layer thickness (2 μm) (i.e. The trench width W = 2 μm, and the trench is filled with SiO2.

[0103] (3) Multilayer dielectric structure: Alternating stacked SiO2 / Si3N4 Bragg reflector layers are deposited between the waveguides. The thickness of each layer satisfies the quarter-wavelength condition. Where d represents the thickness of a single-layer film, The wavelength is represented by λ, and λ represents the refractive index of the material in this layer (the refractive index of SiO2 is approximately 1.45, and the refractive index of Si3N4 is approximately 2.0). A total of 5 cycles are stacked, with the SiO2 layer being approximately 228 nm thick and the Si3N4 layer being approximately 152 nm thick. The evanescent wave lateral diffusion is suppressed through the Bragg reflection effect.

[0104] The geometric parameters of the crosstalk suppression structure (grating period Λ, etching depth h, trench width W, number of multilayer film periods) are determined based on the first threshold (inter-channel crosstalk below -30dB) and the third threshold (number of packaged channels N≥64): when the crosstalk requirement is -30dB and the number of channels is 64, a subwavelength grating with Λ=200nm can meet the requirements; when the crosstalk requirement is increased to -40dB, Λ is reduced to 180nm and a trench isolation structure is added; when the number of channels increases to 256 and the waveguide spacing is compressed from 10μm to 4μm, a combination of grating and trench isolation scheme is required.

[0105] Preferably, as an alternative to the reference interferometric unit monitoring method, a distributed Bragg grating temperature sensor array (4 sensors per square millimeter) is deployed in the critical area of ​​the chip. The controller reads the center wavelength drift of each sensor. The local temperature change is calculated using the following formula:

[0106] in, This indicates the amount of temperature change (unit: K). This indicates the amount of wavelength shift at the center of the Bragg grating (unit: nm). The nominal center wavelength of the Bragg grating (unit: nm). The coefficient of thermal expansion of optical fiber material (unit: 1 / K). This represents the thermo-optic coefficient of the optical fiber material (unit: 1 / K).

[0107] The controller then calculates the state variables of phase drift using a pre-calibrated temperature and phase mapping relationship:

[0108] in, The state quantity representing the phase drift estimated based on temperature sensor readings (unit: rad) is a state quantity characterizing the phase drift. The specific manifestations of temperature sensor monitoring solutions. Indicates the operating wavelength (unit: nm). The thermo-optic coefficient of the waveguide material (unit: 1 / K). Indicates the effective optical length of the waveguide (unit: μm). This indicates the amount of temperature change (unit: K).

[0109] Preferably, as an alternative, the controller periodically performs standard quantum state tomography measurements. Specifically, the controller prepares a known initial state, passes it through a quantum circuit containing a phase shifter under test, and then performs projection measurements to reconstruct the output state density matrix. Then, compared with the theoretical state... Calculate fidelity:

[0110] In this preferred implementation, F represents the fidelity (dimensionless, ranging from 0 to 1), and Tr represents the trace of the matrix (i.e., the sum of the diagonal elements of the matrix). This represents the output density of states matrix under the theoretical error-free condition. This represents the output density of states matrix reconstructed from experimental measurements. Represents the density matrix Perform matrix square root operation (i.e., find the matrix that satisfies the given conditions) (The matrix). The calibration process is triggered when the fidelity F drops beyond a preset threshold (e.g., F < 0.95).

[0111] Specifically, the controller can also convert the fidelity F into an equivalent phase error through the fidelity-phase error mapping relationship:

[0112] in, The equivalent phase error (unit: rad) is a state quantity that characterizes the phase drift. Specific manifestations under the quantum state tomography scheme; f ( The expression represents the fidelity-equivalent phase error mapping function, which is obtained in advance through theoretical calculations or experimental calibration, reflecting the monotonically decreasing relationship between fidelity and equivalent phase error. This function represents the state quantity (equivalent phase error) when there is phase drift. ≥ π When the value is / 100 (corresponding to the second threshold), it indicates that the phase drift has reached an unacceptable level, and the calibration process is triggered at this time.

[0113] Example 3 Please refer to Figure 2 This invention provides a photonic quantum chip packaging control device 30, which includes: a matrix establishment module 31, a compensation calculation module 32, a monitoring module 33, a judgment module 34, a mode determination module 35, a subset extraction module 36, and a matrix update module 37; wherein... The matrix establishment module 31 is used to establish a thermal crosstalk matrix between multiple thermo-optical control units on the optical quantum chip. The dimension of the thermal crosstalk matrix is ​​determined by a third threshold, and the coupling strength distribution of the thermal crosstalk matrix is ​​determined by a first threshold. The first threshold characterizes the crosstalk tolerance between channels, and the third threshold characterizes the channel size of the chip. The compensation calculation module 32 is used to obtain the compensation value of the driving parameters of each thermo-optical control unit based on the thermal crosstalk matrix and the second threshold, wherein the second threshold represents the preset phase control accuracy requirement. Monitoring module 33 is used to monitor state quantities characterizing phase drift when the photonic quantum chip is operated under the condition of applying the compensation value of the driving parameters; The judgment module 34 is used to determine whether the state quantity exceeds the drift threshold corresponding to the second threshold. The mode determination module 35 is used to determine, based on the first threshold and the third threshold, important modes in the thermal crosstalk matrix that need to maintain update validity when the judgment module determines that the state quantity exceeds the drift threshold. Subset extraction module 36 is used to extract the thermo-optical modulation units corresponding to the important modes as the subset to be updated; The matrix update module 37 is used to perform recalibration measurement on the subset to be updated, and update the elements in the thermal crosstalk matrix corresponding to the subset to be updated according to the measurement results, so as to obtain the updated thermal crosstalk matrix. The compensation calculation module 32 is also used to recalculate the compensation value of the driving parameters of each thermo-optical control unit based on the updated thermal crosstalk matrix. The monitoring module 33 is also used to continue operating the photonic quantum chip under the condition of applying a recalculated compensation value and to monitor the state quantities characterizing phase drift.

[0114] In one specific embodiment, the matrix establishment module 31 is used to establish a thermal crosstalk matrix among multiple thermo-optical control units 11 on the photonic quantum chip 10. Upon initial power-on, the matrix establishment module 31 executes a complete step A, sequentially driving all 64 thermo-optical control units 11 through the driving circuit and measuring the response of all channels to establish a 64×64 initial thermal crosstalk matrix C. The rows of matrix C correspond to waveguides, the columns correspond to thermo-optical control units, and the matrix elements... This represents the phase influence coefficient on the i-th waveguide channel when the j-th thermo-optical control unit applies a unit drive. The dimension of the matrix is ​​determined by the third threshold (64 packaged channels), and the distribution of coupling strength in the matrix is ​​determined by the first threshold (inter-channel crosstalk tolerance). This initial matrix C is stored in the FPGA's block RAM. This process takes approximately 2 seconds and is a one-time initialization overhead.

[0115] The compensation calculation module 32 is used to obtain the compensation values ​​of the driving parameters of each thermo-optical control unit based on the thermal crosstalk matrix and the second threshold. During normal operation, the compensation calculation module 32 calculates the compensation values ​​based on the currently stored matrix C and the target phase configuration vector required by the quantum algorithm. By solving the system of linear equations To calculate the compensation power vector P in real time. In practice, since C may be close to singular, the compensation calculation module 32 uses a pseudo-inverse calculation method to solve the problem.

[0116] Where P is the compensation power vector to be solved (dimension 64×1). This represents the pseudo-inverse matrix (Moore-Penrose generalized inverse) of matrix C. Configure a vector (64×1) for the target phase. The goal of this module is to maintain the phase control error of all channels below the second threshold π / 100.

[0117] Monitoring module 33 is used to monitor state quantities characterizing phase drift when the photonic quantum chip is operated under conditions of applied compensation. Monitoring module 33 continuously monitors the output spectral drift of an on-chip reference Mach-Zehnder interferometer at a frequency of 100 Hz and converts it into state quantities representing phase drift. This is used as a state quantity to characterize phase drift.

[0118] The judgment module 34 is used to determine whether the state quantity exceeds the drift threshold corresponding to the second threshold. Suppose that at a certain moment, due to the execution of a complex quantum algorithm, the power of multiple thermo-optical control units 11 changes drastically, the thermal field redistributes, and this causes the state quantity output by the monitoring module 33 to show phase drift. It exceeds the drift threshold π / 100, which satisfies ≥π / 100. The judgment module 34 outputs a result exceeding the threshold, triggering the subsequent calibration process.

[0119] The mode determination module 35 is used to determine the important modes in the thermal crosstalk matrix that need to maintain update validity based on the first threshold and the third threshold when the judgment module 34 determines that the state quantity exceeds the drift threshold. The mode determination module 35 performs singular value decomposition (or eigenvalue decomposition) on the currently stored matrix C, and then calculates the cumulative contribution rate of the eigenvalues. :

[0120] in, The cumulative contribution rate of the first k eigenvalues. For the i-th eigenvalue, sorted in descending order: λ1≥λ2≥…≥λ 64 The mode determination module 35 determines a preset energy threshold η (η=90% in this embodiment) based on a first threshold (inter-channel crosstalk tolerance), and finds the condition that satisfies the threshold. The minimum k value ≥ η (e.g., k = 8). These first 8 eigenmodes are identified as important modes in the thermal crosstalk matrix that need to maintain update effectiveness.

[0121] The subset extraction module 36 is used to extract the thermo-optical control units corresponding to the important modes as the subsets to be updated. The subset extraction module 36 determines the thermo-optical control units 11 corresponding to the first 8 important modes as the subset S to be updated, with a subset size M=8.

[0122] The matrix update module 37 is used to perform recalibration measurements on the subset to be updated and update the elements in the thermal crosstalk matrix corresponding to the subset to be updated based on the measurement results. The matrix update module 37 performs a fast recalibration measurement only on 8 cells in the subset S to be updated, and updates the corresponding 8 columns of data in matrix C using the measurement results, while leaving the other 56 columns unchanged, thus obtaining the updated thermal crosstalk matrix C'. This update process is much faster than full matrix measurement.

[0123] After the update is complete, the compensation calculation module 32 recalculates and applies the compensation value P' based on the updated and more accurate matrix C', thereby quickly bringing the phase drift back within the tolerance. The entire process from exceeding the tolerance to updating and then correcting can be completed in milliseconds, minimizing the impact on upper-level quantum computing.

[0124] Furthermore, if the judgment module 34 determines the state quantity of phase drift in 200 consecutive monitoring sessions (approximately 2 seconds), then... Stabilize within the drift threshold π / 100 (i.e. If the frequency is less than π / 100, the judgment module 34 determines that the system has entered the thermal stability period and sends a frequency adjustment command to the monitoring module 33. The monitoring module 33 adaptively reduces the monitoring frequency from 100Hz to 10Hz to save power. Once a low-frequency monitoring detects an increasing trend of drift, the monitoring frequency will immediately return to 100Hz.

[0125] Through this adaptive closed-loop control strategy that combines hardware and software collaboration with dynamic and static approaches, the system described in this embodiment can maintain extremely high phase fidelity during large-scale, long-term quantum computing tasks, while significantly reducing the time and energy overhead associated with calibration.

[0126] Example 4 Please refer to Figure 3 This invention provides a photonic quantum chip packaging system, which includes a photonic quantum chip 10, a packaging substrate 20, and a control device 30.

[0127] The photonic quantum chip 10 is the core component, on which multiple thermo-optical control units 11 are integrated and configured to meet the following conditions: thermal crosstalk between its channels is lower than a first threshold, its phase control error is lower than a second threshold, and the number of packaged channels is greater than a third threshold. In this embodiment, the photonic quantum chip 10 adopts a silicon-based photonic quantum chip integrating 64 channels, and its waveguides are pre-fabricated with trench isolation crosstalk suppression structures 12 through a deep etching process to maximize thermal isolation physically.

[0128] The packaging substrate 20 is used to carry the photonic quantum chip 10, providing it with stable mechanical support, electrical connection and thermal management, and stabilizing the overall operating temperature of the chip through an integrated semiconductor cooler (TEC).

[0129] The control device 30 is the core control unit of the entire system, configured to execute the packaging method described in the first embodiment of the aforementioned method. The control device 30 is implemented using a high-performance FPGA and is connected to the thermo-optical modulation unit 11 on the quantum chip 10 and various sensors via an electrical interface. The control device 30 internally includes a matrix establishment module 31, a compensation calculation module 32, a monitoring module 33, a judgment module 34, a mode determination module 35, a subset extraction module 36, and a matrix update module 37. Each module interacts with the others via the internal bus of the control device 30, and the state machine within the FPGA coordinates the sequential execution of each module to jointly achieve complete closed-loop control of the packaging calibration.

[0130] It should be noted that the optical quantum chip packaging system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the optical quantum chip packaging system and the optical quantum chip packaging method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0131] The packaging method, apparatus, and system provided by this invention have a wide range of applications. For example, in building large-scale, universal quantum computers based on photons, the phase of tens of thousands of qubits must be precisely and stably controlled; the packaged chip provided by this invention can serve as its core quantum processing unit (QPU). In quantum communication networks, the chip packaged by this invention can serve as a high-performance quantum router or switch node, performing complex optical path selection and phase manipulation to achieve high-fidelity quantum state transmission and distribution. In dedicated optical quantum simulators, such as those used for new material discovery or drug design, it is necessary to stably simulate complex Hamiltonian evolution over long periods; the continuous self-calibration capability of this invention ensures the long-term validity and reliability of the simulation results. In all these cutting-edge applications, effective suppression of phase drift is one of the key bottlenecks in moving from the laboratory to practical applications, and this invention provides a feasible engineering solution to this bottleneck.

[0132] Embodiments of this application also provide a computer device, please refer to... Figure 4 The computer device includes a processor and a memory, the memory storing at least one instruction, at least one program, code set or instruction set, the at least one instruction, at least one program, code set or instruction set being loaded and executed by the processor to implement the photonic quantum chip packaging method provided in the above-described method embodiments.

[0133] Embodiments of this application also provide a computer-readable storage medium storing at least one instruction, at least one program, code set, or instruction set. This instruction, program, code set, or instruction set is loaded and executed by a processor to implement the photonic quantum chip packaging method provided in the above-described method embodiments. For ease of description, the above systems or devices are described by functionally as various modules or units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware components.

[0134] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0135] Finally, it should be noted that in this document, relational terms such as first, second, third, and fourth are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0136] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for packaging a photonic quantum chip, wherein the photonic quantum chip is configured to satisfy the following conditions: inter-channel crosstalk is lower than a first threshold, phase control error is lower than a second threshold, and the number of packaged channels is greater than a third threshold, characterized in that, include: Step A: Establish a thermal crosstalk matrix among multiple thermo-optical control units on the optical quantum chip. The dimension of the thermal crosstalk matrix is ​​determined by the third threshold, and the coupling strength distribution of the thermal crosstalk matrix is ​​determined by the first threshold. The first threshold characterizes the crosstalk tolerance between channels, and the third threshold characterizes the channel size of the chip. Step B: Based on the thermal crosstalk matrix and the second threshold, obtain the compensation value of the driving parameter of each thermo-optical control unit, wherein the second threshold represents the preset phase control accuracy requirement; Step C: Run the photonic quantum chip under the condition of applying the compensation value of the driving parameters, and monitor the state quantity representing the phase drift. When it is determined that the state quantity exceeds the drift threshold corresponding to the second threshold, proceed to steps D to G; when it is determined that the state quantity does not exceed the drift threshold, continue to monitor the state quantity representing the phase drift. Step D: Perform eigenvalue decomposition on the thermal crosstalk matrix to obtain N eigenvalues ​​arranged in descending order; determine a preset energy threshold η based on the first threshold, and calculate the cumulative contribution rate of the first k eigenvalues. Find satisfaction The minimum k value ≥ η is used to determine the first k characteristic modes as important modes in the thermal crosstalk matrix that need to maintain update effectiveness; wherein, the ratio of the third threshold constraint k to N is further reduced as N increases; Step E: Extract the thermo-optical modulation units corresponding to the important modes as the subset to be updated; Step F: Perform recalibration measurement on the subset to be updated, and update the elements in the thermal crosstalk matrix corresponding to the subset to be updated according to the measurement results to obtain the updated thermal crosstalk matrix; Step G: Based on the updated thermal crosstalk matrix, re-execute step B to update the compensation values ​​of the driving parameters of each thermo-optical control unit; then continue to execute step C.

2. The method according to claim 1, characterized in that, The subset to be updated is determined by performing mode decomposition on the thermal crosstalk matrix and extracting the thermo-optical control units corresponding to modes whose cumulative contribution rate exceeds a preset percentage as the subset to be updated; wherein, the preset percentage is dynamically adjusted according to the first threshold.

3. The method according to claim 1, characterized in that, The subset to be updated is determined by identifying thermo-optical control units in the thermal crosstalk matrix whose influence is lower than a preset threshold, and excluding the units from the subset to be updated; wherein the preset threshold is dynamically adjusted according to the first threshold.

4. The method according to claim 1, characterized in that, The state variables characterizing the phase drift are obtained through at least one of the following methods: The output spectral drift of the on-chip reference interferometer is monitored, and the spectral drift is converted into a phase drift state quantity. Read the readings from the on-chip temperature sensor and convert the temperature readings into a state quantity with phase drift through the temperature-phase mapping relationship; The fidelity estimate of the quantum operation result is measured, and the fidelity estimate is converted into an equivalent phase error through the mapping relationship between fidelity and phase error, which is used as the state quantity of the phase drift.

5. The method according to claim 1, characterized in that, The second threshold is when the phase control error is less than π / 100.

6. The method according to claim 1, characterized in that, The rows of the thermal crosstalk matrix correspond to waveguide channels, and the columns correspond to thermo-optical control units. The matrix elements represent the phase influence coefficient under a unit driving amount. The dimension of the thermal crosstalk matrix is ​​N×N, where N is the number of the encapsulation channels, which is limited by the third threshold.

7. The method according to claim 1, characterized in that, The operation of establishing the thermal crosstalk matrix in step A is performed once when the photonic quantum chip is first powered on, and the full matrix is ​​not re-established when the subset to be updated is subsequently updated.

8. The method according to claim 1, characterized in that, If the state quantity does not exceed the drift threshold after multiple consecutive returns to execute step C, the time interval for executing the monitoring operation in step C is extended or the monitoring frequency is reduced; wherein, the adjustment range of the time interval or monitoring frequency is determined according to the first threshold and the third threshold.

9. The method according to claim 1, characterized in that, The quantum chip has a crosstalk suppression structure between its waveguides. The crosstalk suppression structure is at least one of the following: a subwavelength grating structure, a trench isolation structure, or a multilayer dielectric structure with different refractive index contrasts. The geometric parameters of the crosstalk suppression structure are determined based on the first threshold and the third threshold.

10. A photonic quantum chip packaging and control device, characterized in that, include: A matrix establishment module is used to establish a thermal crosstalk matrix between multiple thermo-optical control units on a photonic quantum chip. The dimension of the thermal crosstalk matrix is ​​determined by a third threshold, and the coupling strength distribution of the thermal crosstalk matrix is ​​determined by a first threshold. The first threshold characterizes the crosstalk tolerance between channels, and the third threshold characterizes the channel size of the chip. The compensation calculation module is used to obtain the compensation value of the driving parameters of each thermo-optical control unit based on the thermal crosstalk matrix and the second threshold, wherein the second threshold represents the preset phase control accuracy requirement. A monitoring module is used to monitor state quantities characterizing phase drift when the photonic quantum chip is operated under conditions where a compensation value of the driving parameters is applied. The judgment module is used to determine whether the state quantity exceeds the drift threshold corresponding to the second threshold; The mode determination module is used to perform eigenvalue decomposition on the thermal crosstalk matrix to obtain N eigenvalues ​​arranged in descending order when the judgment module determines that the state variable exceeds the drift threshold; determine a preset energy threshold η based on the first threshold; and calculate the cumulative contribution rate of the first k eigenvalues. Find satisfaction The minimum k value ≥ η is used to determine the first k characteristic modes as important modes in the thermal crosstalk matrix that need to maintain update effectiveness; wherein, the ratio of the third threshold constraint k to N is further reduced as N increases; The subset extraction module is used to extract the thermo-optical modulation units corresponding to the important modes as the subsets to be updated; The matrix update module is used to perform recalibration measurements on the subset to be updated, and update the elements in the thermal crosstalk matrix corresponding to the subset to be updated according to the measurement results, so as to obtain the updated thermal crosstalk matrix. The compensation calculation module is also used to recalculate the compensation value of the driving parameters of each thermo-optical control unit based on the updated thermal crosstalk matrix. The monitoring module is also used to continue operating the quantum chip under the condition of applying a recalculated compensation value and to monitor the state quantities characterizing phase drift.

11. A photonic quantum chip packaging system, characterized in that, include: A photonic quantum chip integrates multiple thermo-optical control units. The photonic quantum chip is configured to meet the following conditions: inter-channel crosstalk is lower than a first threshold, phase control error is lower than a second threshold, and the number of packaged channels is greater than a third threshold. The packaging substrate carries the optical quantum chip; The control device, as described in claim 10, is electrically connected to the packaging substrate and is used to perform packaging calibration on the quantum chip.

12. A computer-readable storage medium, characterized in that, The device contains a computer program that, when executed by a processor, implements the packaging method for the photonic quantum chip according to any one of claims 1 to 9.

Images