Control device and control system for superconducting quantum processors based on frequency division multiplexing

Abstract

The application provides a control device and a control system of a superconducting quantum processor based on frequency division multiplexing. It relates to the field of superconducting quantum computing. The control device comprises a control module, which comprises a plurality of finite impulse response filters for pre-correcting a plurality of digital control signals; a superposition component adapted to linearly superimpose the pre-corrected digital control signals in the digital domain, and to obtain a composite signal carrying a plurality of control information according to the composite signal generated after linear superposition; and a frequency division multiplexing module adapted to obtain a plurality of analog control signals according to the composite signal. The frequency division multiplexing module comprises a plurality of transmission channels for transmitting the plurality of analog control signals to a plurality of control objects respectively, so as to control the plurality of control objects and realize high-fidelity transmission of control information.

Description

Technical Field

[0001] This invention relates to the fields of superconducting quantum computing and quantum control engineering, and particularly to a control device for a frequency division multiplexing-based superconducting quantum processor and a control system for the superconducting quantum processor. Background Technology

[0002] In the expansion of superconducting quantum computers, the number of control cables increases linearly or even exponentially with the increase in the number of qubits, leading to severe wiring space constraints and thermal load pressure within the dilution cooler. Frequency Division Multiplexing (FDM) technology can transmit signals carrying control information from multiple superconducting qubits or multiple tunable couplers on a single microwave transmission line, significantly reducing hardware resource consumption. However, in FDM, the control information suffers varying degrees of distortion as it passes through the physical channels including transmission lines and filters, resulting in a decrease in the fidelity of quantum gate operations. Summary of the Invention

[0003] In view of this, the present invention provides a control device and control system for a frequency division multiplexing-based superconducting quantum processor, wherein the control device includes:

[0004] The control module includes:

[0005] Multiple finite impulse response filters are used to pre-correct multiple digital control signals; each of the multiple digital control signals corresponds to a multiple controlled object, and the digital control signals carry control information for manipulating the corresponding controlled object; the controlled object is a superconducting quantum bit or a tunable coupler of a superconducting quantum processor.

[0006] The superposition component is suitable for linearly superimposing multiple pre-corrected digital control signals in the digital domain, and obtaining a composite signal carrying multiple control information based on the composite signal generated after linear superposition. The composite signal is an analog signal.

[0007] The frequency division multiplexing module is suitable for obtaining multiple analog control signals based on the above-mentioned synthesized signal. The multiple analog control signals correspond one-to-one with multiple controlled objects. The analog control signals carry control information for manipulating the corresponding controlled objects. The frequency division multiplexing module includes multiple transmission channels for transmitting the multiple analog control signals to the multiple controlled objects respectively, so as to use the multiple control information to manipulate the multiple controlled objects.

[0008] Among them, the pre-corrected digital control signal is used to compensate for the changes in control information caused by the transmission channel during the transmission of analog control signals.

[0009] According to an embodiment of the present invention, the control device further includes:

[0010] The processor is configured as follows:

[0011] Based on the transfer function values ​​of the i-th transmission channel at multiple frequency points, determine the time-domain impulse response sequence of the i-th finite impulse response filter;

[0012] Based on the time-domain impulse response sequence of the i-th finite impulse response filter, determine multiple tap coefficients of the i-th finite impulse response filter;

[0013] Multiple finite impulse response filters are configured to pre-correct multiple digital control signals using the following method:

[0014] The i-th digital control signal is pre-corrected by convolving multiple tap coefficients of the i-th finite impulse response filter, where i ≥ 1.

[0015] According to an embodiment of the present invention, the processor determines the time-domain impulse response sequence of the i-th finite impulse response filter by the following method:

[0016] Based on the magnitude values ​​of the transfer function of the i-th transmission channel at multiple frequency points, determine the frequency points whose magnitude values ​​are greater than a preset threshold from the multiple frequency points to obtain multiple target frequency points;

[0017] The pre-correction coefficients for each target frequency point are obtained by taking the reciprocal of the transfer function of the i-th transmission channel at each of the multiple target frequency points.

[0018] Perform an inverse discrete Fourier transform on the pre-correction coefficients at each target frequency point to obtain the time-domain impulse response sequence of the i-th finite impulse response filter.

[0019] According to an embodiment of the present invention, the processor determines multiple tap coefficients of the i-th finite impulse response filter by the following method:

[0020] Perform a cyclic shift operation on the time-domain impulse response sequence of the i-th finite impulse response filter so that zero time delay is at the center of the time-domain impulse response sequence, and obtain the shifted time-domain impulse response sequence;

[0021] Extract the target time-domain impulse response sequence from the shifted time-domain impulse response sequence, such that the center of the target time-domain impulse response sequence coincides with the center of the shifted time-domain impulse response sequence;

[0022] Windowing is applied to the target time-domain impulse response sequence to obtain multiple tap coefficients of the i-th finite impulse response filter.

[0023] According to an embodiment of the present invention, when the controlled object is a superconducting quantum bit, the superposition component includes:

[0024] The first digital adder is suitable for linearly superimposing multiple pre-corrected digital control signals in the digital domain to generate a composite signal.

[0025] A mixer is suitable for mixing the above-mentioned composite signal with the local oscillator signal to complete the up-conversion processing of the above-mentioned composite signal.

[0026] The first arbitrary waveform generator is suitable for performing digital-to-analog conversion on the up-converted composite signal to generate the above-mentioned composite signal.

[0027] According to an embodiment of the present invention, when the controlled object is an adjustable coupler, the superposition component includes:

[0028] The second digital adder is used to linearly superimpose multiple pre-corrected digital control signals in the digital domain to generate a composite signal.

[0029] The second arbitrary waveform generator is suitable for performing digital-to-analog conversion on composite signals to generate the aforementioned composite signal.

[0030] According to an embodiment of the present invention, the frequency division multiplexing module includes:

[0031] A power divider, suitable for dividing the above-mentioned synthesized signal into multiple sub-signals;

[0032] Multiple bandpass filters are one-to-one with multiple sub-signals. The multiple bandpass filters are used to filter the multiple sub-signals respectively to obtain multiple analog control signals.

[0033] Multiple transmission lines correspond one-to-one with multiple bandpass filters. These transmission lines and their corresponding bandpass filters form a transmission channel to transmit multiple analog control signals to multiple controlled objects.

[0034] As a second aspect of the invention, a control system is also provided, which is applied to a superconducting quantum processor. The superconducting quantum processor includes A groups of superconducting qubits and B groups of couplers. The superconducting qubit groups include multiple superconducting qubits, and the coupler groups include multiple couplers. The control system comprises:

[0035] There are M control devices, where A control devices correspond one-to-one with A superconducting qubit groups, and each of the A control devices generates multiple analog control signals to manipulate multiple superconducting qubits in its corresponding superconducting qubit group; B control devices correspond one-to-one with B coupler groups, and each of the B control devices generates multiple analog control signals to manipulate multiple tunable couplers in its corresponding coupler group, and A+B=M.

[0036] According to an embodiment of the present invention, the center frequency interval of the analog control signals of any two superconducting qubits within each superconducting qubit group is greater than or equal to half the sum of the bandwidths of the analog control signals of the two superconducting qubits.

[0037] According to an embodiment of the present invention, the center frequency interval of the analog control signals of any two adjustable couplers in each coupler group is greater than or equal to half the sum of the bandwidths of the analog control signals of the two adjustable couplers.

[0038] This invention offers the following technical advantages: By utilizing multiple finite impulse response (FIR) filters to independently pre-correct each digital control signal, it can compensate for the distortion characteristics of different operating frequencies and transmission channels corresponding to different controlled objects, ensuring high-fidelity transmission of control information. The pre-corrected digital control signal compensates for the amplitude-frequency response unevenness and phase-frequency response nonlinearity (including group delay fluctuations) introduced into the transmission channel, enabling the control information carried by the analog control signal (such as pulse envelope shape, pulse width, phase, and timing) to be faithfully reproduced after transmission, thereby ensuring the operational fidelity of single-bit gates and double-bit gates. Attached Figure Description

[0039] The above and other objects, features and advantages of the present invention will become more apparent from the following description of embodiments of the invention with reference to the accompanying drawings, in which:

[0040] Figure 1 A schematic diagram of a control device for a frequency division multiplexing-based superconducting quantum processor is shown according to an embodiment of the present invention.

[0041] Figure 2 A schematic diagram of a control device for a frequency division multiplexing-based superconducting quantum processor is shown according to another embodiment of the present invention.

[0042] Figure 3 A partial structural schematic diagram of a frequency division multiplexing module provided according to an embodiment of the present invention is shown.

[0043] Figure 4A The transfer function curve of a bandpass filter according to an embodiment of the present invention is shown.

[0044] Figure 4B Multiple tap coefficients of a finite impulse response filter obtained according to an embodiment of the present invention are shown.

[0045] Figure 5 The simulation comparison results before and after digital control signal pre-correction according to an embodiment of the present invention are shown.

[0046] Figure 6A schematic diagram of the control system of a superconducting quantum processor provided according to an embodiment of the present invention is shown.

[0047] Explanation of reference numerals in the attached figures

[0048] 10. Control module; 20. Frequency division multiplexing module; 11. Finite impulse response filter; 12. Superposition component; 121. First digital adder; 122. Mixer; 123. First arbitrary waveform generator; 124. Second digital adder; 125. Second arbitrary waveform generator; 21. Power divider; 22. Bandpass filter; 100. First control device; 200. Second control device; 300. Superconducting quantum processor. Detailed Implementation

[0049] In the process of realizing this invention, it was discovered that the gate types of analog control signals in superconducting quantum computing systems include single-qubit gates and two-qubit gates, as shown in Table 1. Single-qubit gates control superconducting qubits, typically operating in the microwave band (approximately 5 GHz, corresponding to the superconducting qubit resonance frequency), and their waveform is characterized by envelope-shaped microwave pulses. Two-qubit gates control tunable couplers, typically operating in the mid-to-low frequency band (approximately 200 MHz to 400 MHz, corresponding to the tunable coupler modulation frequency), and their waveform is characterized by low-frequency modulation, used for sideband excitation.

[0050] Table 1

[0051]

[0052] Existing general digital predistortion methods can be used to compensate for digital control signals in a single path or a single frequency band. However, in the frequency division multiplexing control scenario of superconducting quantum computing, the frequencies, passbands of bandpass filters, transmission lines, and output ports of the analog control signals corresponding to different superconducting qubits and tunable couplers are all different. If only the composite signal generated after linear superposition is uniformly corrected, it is difficult to compensate for the amplitude distortion and phase distortion caused by each transmission channel separately. Therefore, the control device of this invention can calculate the finite impulse response pre-correction kernel according to the transfer function corresponding to each controlled object before linearly superimposing the digital control signals in the digital domain, and apply the finite impulse response pre-correction kernel to the corresponding digital control signal.

[0053] Therefore, a pre-correction method for digital control signals specifically designed for frequency division multiplexing (FDM) technology is needed. This method should be able to simultaneously optimize the operational fidelity of single-bit and double-bit gates and support adaptive updates of pre-correction parameters based on changes in transmission channel characteristics, so as to meet the actual needs of FDM superconducting quantum computing systems.

[0054] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the invention. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the invention for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0055] Figure 1 A schematic diagram of a control device for a frequency division multiplexing-based superconducting quantum processor is shown according to an embodiment of the present invention.

[0056] Figure 2 A schematic diagram of a control device for a frequency division multiplexing-based superconducting quantum processor is shown according to another embodiment of the present invention.

[0057] like Figure 1 and Figure 2 As shown, the control device includes: a control module 10 and a frequency division multiplexing module 20.

[0058] The control module 10 includes multiple finite impulse response (FIR) filters 11 and a superposition component 12. The FIR filters 11 pre-correct multiple digital control signals in the digital domain; each digital control signal corresponds one-to-one with a control object, and the digital control signal carries control information for manipulating the corresponding control object; the control object is a superconducting quantum bit of a superconducting quantum processor or a tunable coupler. The superposition component 12 is suitable for linearly superimposing multiple pre-corrected digital control signals in the digital domain, and obtaining a composite signal carrying control information for multiple control objects based on the composite signal generated after linear superposition. The composite signal is an analog signal carrying multiple control information. The frequency division multiplexing (FDM) module 20 is suitable for obtaining multiple analog control signals from the composite signal. Each analog control signal corresponds one-to-one with a control object, and the analog control signal carries control information for manipulating the corresponding control object. The FDM module includes multiple transmission channels for transmitting the multiple analog control signals to the multiple control objects respectively, so as to manipulate the multiple control objects using the multiple control information; wherein, the pre-corrected digital control signals are used to compensate for changes in control information caused by the transmission channels during the transmission of analog control signals.

[0059] Multiple finite impulse response filters 11 independently pre-correct each digital control signal, compensating for the distortion characteristics of different operating frequencies and transmission channels corresponding to different controlled objects, thus ensuring high-fidelity transmission of control information. The pre-corrected digital control signal compensates for the amplitude-frequency response unevenness and phase-frequency response nonlinearity (including group delay fluctuation) introduced in the transmission channel, enabling the control information carried by the analog control signal (such as pulse envelope shape, pulse width, phase, and timing) to be restored with high fidelity after transmission, thereby ensuring the operational fidelity of single-bit gates and double-bit gates.

[0060] refer to Figure 1 When the controlled object is a superconducting quantum bit, the superposition component 12 includes: a first digital adder 121, a mixer 122, and a first arbitrary waveform generator 123. The first digital adder 121 is suitable for linearly superimposing multiple pre-corrected digital control signals in the digital domain to generate a composite signal. The mixer 122, for example, is an IQ mixer, suitable for mixing the composite signal with a local oscillator signal to complete the up-conversion processing of the composite signal. The first arbitrary waveform generator 123 is suitable for performing digital-to-analog conversion on the up-converted composite signal to generate a synthesized signal.

[0061] refer to Figure 2 When the controlled object is an adjustable coupler, the superposition component 12 includes: a second digital adder 124 and a second arbitrary waveform generator 125. The second digital adder 124 is used to linearly superimpose multiple pre-corrected digital control signals in the digital domain to generate a composite signal. The second arbitrary waveform generator 125 is suitable for performing digital-to-analog conversion on the composite signal to generate a synthesized signal.

[0062] Figure 3 A partial structural schematic diagram of a frequency division multiplexing module provided according to an embodiment of the present invention is shown.

[0063] like Figure 3As shown, the frequency division multiplexing module 20 includes a power divider 21, multiple bandpass filters 22, and multiple transmission lines. The power divider 21 is used to divide the synthesized signal into multiple sub-signals. Each bandpass filter 22 corresponds to one of the multiple sub-signals and is used to filter the multiple sub-signals respectively to obtain multiple analog control signals. Each transmission line corresponds to one of the multiple bandpass filters, and the transmission lines and their corresponding bandpass filters form transmission channels. The multiple transmission lines are used to transmit the multiple analog control signals to multiple controlled objects. The transmission lines can be, for example, physical cables, or on-chip metal interconnects, depending on the specific packaging method. In superconducting quantum computing frequency division multiplexing (FDM) control systems, bandpass filters (BPFs) are key devices for separating different analog control signals. However, the non-ideal characteristics of bandpass filters introduce two types of distortion: first, the uneven amplitude-frequency response within the passband causes waveform amplitude distortion in the analog control signal; second, the nonlinear phase-frequency response causes group delay fluctuations, which in turn lead to phase distortion in the analog control signal. These amplitude and phase distortions significantly reduce the fidelity of quantum gate operations, thus limiting the overall performance of quantum computing.

[0064] The control device of this invention overcomes the impact of amplitude distortion and phase distortion introduced by the non-ideal characteristics of the bandpass filter on the fidelity of the quantum gate. The following is a detailed description.

[0065] According to an embodiment of the present invention, the control device further includes a processor. The processor is configured to: determine the time-domain impulse response sequence of the i-th finite impulse response filter based on the function values ​​of the transfer function of the i-th transmission channel at multiple frequency points; and determine multiple tap coefficients of the i-th finite impulse response filter based on the time-domain impulse response sequence of the i-th finite impulse response filter.

[0066] According to an embodiment of the present invention, a plurality of finite impulse response filters are configured to pre-correct a plurality of digital control signals by convolving the i-th digital control signal with a plurality of tap coefficients of the i-th finite impulse response filter to pre-correct the i-th digital control signal.

[0067] According to an embodiment of the present invention, the processor determines the time-domain impulse response sequence of the i-th finite impulse response filter by the following method: based on the magnitude values ​​of the transfer function of the i-th transmission channel at multiple frequency points, the processor determines the frequency points whose magnitude values ​​are greater than a preset threshold from the multiple frequency points to obtain multiple target frequency points; the processor takes the reciprocal of the function values ​​of the transfer function of the i-th transmission channel at each of the multiple target frequency points to obtain the pre-correction coefficients of each target frequency point; and the processor performs an inverse discrete Fourier transform on the pre-correction coefficients of each target frequency point to obtain the time-domain impulse response sequence of the i-th finite impulse response filter.

[0068] According to an embodiment of the present invention, the processor determines multiple tap coefficients of the i-th finite impulse response filter by the following method: performing a cyclic shift operation on the time-domain impulse response sequence of the i-th finite impulse response filter, so that zero delay is at the center of the time-domain impulse response sequence, to obtain a shifted time-domain impulse response sequence; extracting a target time-domain impulse response sequence from the shifted time-domain impulse response sequence, so that the center of the target time-domain impulse response sequence coincides with the center of the shifted time-domain impulse response sequence; and performing windowing processing on the target time-domain impulse response sequence to obtain multiple tap coefficients of the i-th finite impulse response filter.

[0069] According to an embodiment of the present invention, in order to facilitate understanding of the above process, the present invention fully explains the logic of the processor calculating the time-domain impulse response sequence of the i-th finite impulse response filter and the tap coefficients of the i-th finite impulse response filter through the following formula derivation, and realizes the pre-correction of the i-th digital control signal. The specific steps are as follows.

[0070] Step 1: First, let N be the total number of discrete frequency points in the inverse discrete Fourier transform, and L be the tap length (number of taps) of the i-th finite impulse response filter. Where N ≥ 8L, and N can also be a fixed number of high-resolution points greater than the tap length L.

[0071] Construct a baseband frequency grid, where k is the index of the frequency point, k = 0, 1, ..., N-1, and the corresponding baseband grid frequency. Satisfy equations (1) to (2).

[0072] (1);

[0073] (2);

[0074] in, This indicates the digital sampling rate of the first or second arbitrary waveform generator; generally, the digital sampling rates of the two are equal.

[0075] For cases where the controlled object is a superconducting quantum bit, the baseband grid frequency needs to be upconverted, and the specific relationship is expressed as equation (3).

[0076] (3);

[0077] in, The local oscillator signal frequency, This indicates the baseband grid frequency. The frequency applied to the superconducting qubit after upconversion is used to achieve quantum gate manipulation.

[0078] It should be noted that, for the case where the controlled object is an adjustable coupler, there is no need for an up-conversion, and the frequency can be adjusted in equation (3). .

[0079] Step 2, the transfer function of the i-th transmission channel is , Used to characterize the non-ideal transmission characteristics of the i-th transmission channel; It can be determined through vector network analyzer measurements, electromagnetic simulation, system identification, or quantum gate calibration inversion. Simultaneously, it fully includes the amplitude response and phase response of the i-th transmission channel. The transfer function... Using amplitude components With phase components This gives us equation (4).

[0080] (4);

[0081] in, , .

[0082] This means eliminating the 2π jump caused by phase periodic folding in order to restore the true continuous phase trajectory. This indicates taking the principal argument of a complex number, usually located in the interval [-π, π] or [0, 2π).

[0083] Step 3, respectively for and After interpolation, we obtain equation (5).

[0084] (5);

[0085] This represents the value of the transfer function of the i-th transmission channel at the k-th frequency.

[0086] Step 4: To avoid excessive pre-correction gain and waveform noise amplification at transmission nulls, stopbands, or noise-dominant frequency points, a minimum amplitude protection threshold is set. , ,in, This is the decibel threshold, a fixed configuration parameter. Based on the protection threshold. The restricted frequency domain inverse response of the i-th channel is constructed, and this process is expressed as Equation (6).

[0087] (6);

[0088] Let represent the magnitude of the transfer function of the i-th transmission channel at the k-th frequency. According to equation (6), multiple target frequencies can be selected. This method can effectively shield the correction interference of invalid frequencies and ensure the stability and effectiveness of the pre-correction process. Take the reciprocal of the function values ​​of the transfer function of the i-th transmission channel at each of the multiple target frequencies to obtain the pre-correction coefficients of each target frequency. .

[0089] Step 5, for Performing an N-point inverse discrete Fourier transform yields the time-domain impulse response sequence of the i-th finite impulse response filter. . It is represented by equation (7).

[0090] (7).

[0091] Where n is the sampling time number, n=0,1,...,N-1.

[0092] To eliminate group delay offset and align the timing of quantum gate manipulation, the time-domain impulse response sequence of the i-th finite impulse response filter is... Perform a cyclic shift to adjust the zero-delay position to the center of the sequence, resulting in the shifted impulse response sequence. .

[0093] The impulse response sequence after shifting S points are extracted from the center position to obtain the target time-domain impulse response sequence. The target time-domain impulse response sequence is then windowed, i.e., multiplied by a window function. This yields multiple tap coefficients for the i-th finite impulse response (FIR) filter. These multiple tap coefficients can also be called the FIR pre-correction kernel (FIR pre-correction kernel) of the i-th FIR filter. It is represented by equation (8).

[0094] (8).

[0095] in, The shifted impulse response sequence The central index, For window functions, The window function sampling number has a value range of l=0, 1, ..., L-1. The window function can be a Kaiser window, Hamming window, Hanning window, or rectangular window. It is a complex finite impulse response pre-correction kernel that can simultaneously compensate for the amplitude distortion and nonlinear phase distortion of the i-th transmission channel.

[0096] Step 6, for the i-th digital control signal Linear convolution is performed using multiple tap coefficients of the i-th finite impulse response filter to complete the operation. Independent pre-correction yields pre-corrected digital control signals. The process of linear convolution operation is represented by equation (9).

[0097] (9).

[0098] It should be noted that the multiple tap coefficients of the i-th finite impulse response filter generated using the center truncation method... It can compensate for the inherent bias of convolution operations and remove approximately Group delay of each sampling point, alignment The timing of the quantum gate is synchronized with that of the original quantum gate, ensuring the accuracy of the quantum gate timing. The input digital adder is linearly superimposed in the digital domain with the pre-corrected digital control signals from other transmission channels to generate a composite signal.

[0099] It should be noted that in practical engineering applications, a single-channel digitally synthesized waveform can integrate multiple control frequencies to meet the needs of multi-device coordinated control. This invention supports differentiated FIR pre-correction core solution strategies, flexibly adapting to two working modes based on hardware resources. When processor (e.g., FPGA) and arbitrary waveform generator (AWG) hardware resources are sufficient, a dedicated FIR pre-correction core can be independently solved for each transmission channel, achieving precise and independent compensation for transmission distortion in each channel and maximizing the correction accuracy of the digital control signal. When hardware computing power and storage resources are limited, to reduce the computational resource overhead of the FIR pre-correction core, transmission channels with similar transmission characteristics can be divided into the same channel multiplexing group. By solving the transfer function of the channel multiplexing group, a unified shared FIR pre-correction core is generated, achieving multi-channel multiplexing correction and completing lightweight engineering deployment.

[0100] Assume that a multiplexed group contains R transmission channels, and the function values ​​of each transmission channel at the k-th frequency are as follows: , … … The transfer function of the channel multiplexing group at the k-th frequency point is solved based on the least squares criterion. , which is expressed as equation (10).

[0101] (10);

[0102] in express The complex conjugate of. If | |≤ If the denominator is zero, then let =0. Then, based on equations (7) and (8), the shared FIR pre-correction core of the channel multiplexing group is obtained. . | | represents the maximum magnitude of the transfer function of the R transmission channels in the channel multiplexing group at the k-th frequency.

[0103] It should be noted that when the control device experiences changes in operating conditions, such as wiring adjustments, bandpass filter replacement, low-temperature operating environment shifts, changes in control frequency parameters, or drift in quantum gate calibration characteristics, the frequency domain response characteristics of the transmission channel will change accordingly. In this case, the updated channel transfer function can be obtained through re-measurement or system identification estimation. Following the complete process of frequency interpolation, restricted frequency domain inverse response construction, time domain impulse response solving, and tap coefficient generation, the FIR pre-correction core of each transmission channel or channel multiplexing group is updated in real time. This update method supports offline calibration updates and can also be executed periodically through an automated calibration process during system operation, ensuring the pre-correction accuracy of digital control signals and the stability of quantum manipulation under long-term working conditions. Simultaneously, to prevent the amplitude of the pre-corrected digital control signal from exceeding the hardware dynamic range of the FPGA and arbitrary waveform generator, and to prevent waveform clipping distortion and hardware malfunctions, upper limit constraints can be added to the peak amplitude, sequence energy, and frequency domain correction gain of the FIR pre-correction core to achieve amplitude limiting protection of the correction parameters and improve the reliability of system engineering operation.

[0104] According to an embodiment of the present invention, two control devices can be integrated. One control device generates analog control signals to control the superconducting qubits, and the other control device generates analog control signals to control the tunable coupler. To more clearly illustrate the working principle of the integrated device, a specific application example is described in detail below. This example uses the concurrent gate operation of two superconducting qubits and two tunable couplers as a scenario to illustrate the operating principle of the control device of the present invention.

[0105] Suppose that concurrent single-bit gates need to be executed on the first superconducting qubit (hereinafter referred to as Qubit1) and the second superconducting qubit (hereinafter referred to as Qubit2), with resonant frequencies of 5.0 GHz and 5.2 GHz, respectively; at the same time, concurrent two-bit gates need to be executed on the first tunable coupler (hereinafter referred to as Coupler1) and the second tunable coupler (hereinafter referred to as Coupler2), with control frequency bands of 200 MHz and 300 MHz, respectively.

[0106] The control frequencies of Qubit1, Qubit2, Coupler1, and Coupler2 are obtained, with a local oscillator frequency of 4.8 GHz. The sampling rate for generating the sideband signals is 2 GSa / s, and the complex transfer function of each transmission channel is read or measured. Based on the set finite impulse response filter tap number L=200, a restricted frequency domain inverse response is constructed on the corresponding frequency grid. Then, through inverse discrete Fourier transform, center truncation, and windowing, the FIR pre-correction kernel corresponding to each controlled object is obtained. Specifically, superconducting quantum bit control uses the FIR pre-correction kernel convolved with the complex baseband waveform; adjustable coupler control uses the FIR pre-correction kernel convolved with the original digital control signal, depending on the specific implementation.

[0107] For scenarios where the controlled object is a superconducting qubit, digital control signals are generated to drive Qubit1 and Qubit2 respectively, and these signals are then passed through their respective finite impulse response (FIR) filters. The baseband frequencies of the control signals for Qubit1 and Qubit2 are 200MHz and 400MHz, respectively. Due to the different distortion characteristics of the transmission channels corresponding to Qubit1 and Qubit2, FIR pre-correction kernels with different compensation intensities are applied to the FIR filters for Qubit1 and Qubit2.

[0108] For scenarios where the controlled object is an adjustable coupler, digital control signals are generated to drive Coupler1 and Coupler2 to execute AC-CZ gates. These signals are then passed through the corresponding finite impulse response filters of Coupler1 and Coupler2 to compensate for the distortion caused by the transmission channels of Coupler1 and Coupler2.

[0109] The two pre-corrected digital control signals are linearly superimposed in the digital adder inside the FPGA. The superimposed composite signal is up-converted by a digital IQ mixer (local oscillator frequency of 4.8GHz) to generate a digital microwave signal containing two main frequency components of 5.0GHz and 5.2GHz, which is then output by an arbitrary waveform generator.

[0110] The pre-corrected digital control signal is directly fed into the arbitrary waveform generator and outputs as a flux bias voltage without undergoing up-conversion processing.

[0111] Figure 4A The transfer function curve of a bandpass filter according to an embodiment of the present invention is shown.

[0112] like Figure 4A As shown, the solid line represents the amplitude response, and the dashed line represents the phase response. This frequency domain response is the data source for the transfer function of the transmission channel. The nonlinear phase change within the passband of the bandpass filter will cause waveform distortion and timing shift in the analog control signal.

[0113] Figure 4B Multiple tap coefficients of a finite impulse response filter obtained according to an embodiment of the present invention are shown.

[0114] like Figure 4B As shown, the horizontal axis represents the tap index, and the solid and dashed lines correspond to the real and imaginary parts of each tap coefficient, respectively. The tap coefficient is generated by obtaining the time-domain impulse response sequence from the pre-correction coefficient through the N-point inverse discrete Fourier transform. The sequence is then centered, shifted, and windowed to extract the target time-domain impulse response sequence. This is used to synchronously compensate for the amplitude distortion and nonlinear phase distortion introduced by the bandpass filter in the transmission channel.

[0115] Figure 5 The simulation comparison results before and after digital control signal pre-correction according to an embodiment of the present invention are shown.

[0116] in, Figure 5 The transfer function of the transmission channel used in the simulation process is based on... Figure 4A The transfer function of the bandpass filter in [the context]. Figure 5 Part (a) in the diagram represents the digital intermediate frequency (IF) signal envelope, which includes the target IF envelope and the pre-corrected IF envelope. The horizontal axis represents time in nanoseconds (ns), and the vertical axis represents the signal amplitude. The solid line represents the target IF envelope, corresponding to... Figure 1 and Figure 2 The signal envelope of the digital control signal input to the finite impulse response filter is shown. The dashed line represents the pre-corrected intermediate frequency envelope, which is the signal envelope of the pre-corrected digital control signal output from the finite impulse response filter.

[0117] Figure 5 Part (b) in the diagram represents the demodulated output envelope, which includes the target intermediate frequency (IF) envelope, the uncorrected output envelope, and the precorrected output envelope. The horizontal axis represents time in nanoseconds (ns), and the vertical axis represents the signal amplitude. The target IF envelope and... Figure 5The target intermediate frequency (IF) envelope in part (a) is the same. The dotted dashed line represents the uncorrected output envelope, which is the signal envelope obtained by demodulating the analog control signal output from the transmission channel of the frequency division multiplexing module after the target IF envelope has passed through the transmission channel without pre-correction by the finite impulse response filter; the long dashed line represents the pre-corrected output envelope, which is obtained by... Figure 5 The pre-corrected intermediate frequency envelope of part (a) passes completely through the transmission channel of the frequency division multiplexing module, and is then demodulated by the analog control signal output from the transmission channel to obtain the signal envelope. This pre-corrected output envelope is superimposed with the amplitude distortion and phase distortion introduced by the transmission channel, and is used to verify the pre-correction compensation effect of the finite impulse response filter.

[0118] Figure 5 Part (c) is the residual amplitude curve after timing alignment. The horizontal axis represents time in nanoseconds (ns), and the vertical axis represents the error amplitude in decibels (dB). Figure 5 Part (c) is to Figure 5 The error amplitude curves are calculated after the two demodulated output envelopes in part (b) are time-aligned with the target intermediate frequency envelope; the dashed line corresponds to... Figure 5 The error magnitude caused by the uncorrected output envelope in part (b); the solid line corresponds to... Figure 5 The error amplitude generated by the pre-corrected output envelope in part (b) is used to quantitatively evaluate the degree of waveform distortion under the two operating conditions. Simulation results show that, under the premise of timing alignment, the uncorrected output envelope exhibits significant distortion and a large error amplitude. After pre-correction, the pre-corrected output envelope closely approximates the desired target intermediate frequency envelope, and the error amplitude is significantly reduced, verifying that the control device of this invention has a good compensation effect on signal distortion introduced by the transmission channel.

[0119] A superconducting quantum processor includes, for example, A sets of superconducting qubits and B sets of couplers. Each set of superconducting qubits includes multiple superconducting qubits, and each set of couplers includes multiple couplers. The control system includes M control devices. A control devices correspond one-to-one with A sets of superconducting qubits, and each of the A control devices generates multiple analog control signals to manipulate the multiple superconducting qubits in its corresponding set. Similarly, B control devices correspond one-to-one with B sets of couplers, and each of the B control devices generates multiple analog control signals to manipulate the multiple tunable couplers in its corresponding set. A + B = M.

[0120] Existing multiplexing schemes cannot simultaneously meet the transmission requirements of two types of signals. Spectrum resources and cryogenic wiring resources are difficult to optimize globally, failing to support the stable and high-precision operation of large-scale superconducting quantum processors. The control system of this invention distinguishes between two types of control objects—qubits and tunable couplers—and sets up independent multiplexing links. Based on the differences in frequency bands, bandwidths, crosstalk tolerances, concurrent gate relationships, and connection topologies of the analog control signals for each type of control object, the two types of control objects are grouped separately. This achieves global and coordinated allocation of spectrum resources, overcoming the disordered spectrum allocation and resource waste inherent in traditional single-multiplexing architectures, and adapting to multi-object collaborative control scenarios of superconducting quantum processors.

[0121] According to an embodiment of the present invention, A control devices are used to generate analog control signals for manipulating superconducting qubits; these A control devices can be referred to as A first control devices. B control devices are used to generate analog control signals for manipulating tunable couplers; these B control devices can also be referred to as B second control devices. In the control system of the superconducting quantum processor, the frequency division multiplexing modules of the A first control devices and the frequency division multiplexing modules of the B second control devices can be integrated together to form a transmission component. The power divider and bandpass filter of the frequency division multiplexing module form the frequency division multiplexing component.

[0122] Figure 6 A schematic diagram of the control system of a superconducting quantum processor provided according to an embodiment of the present invention is shown.

[0123] like Figure 6 As shown, the control system includes A first control devices 100 and B second control devices 200. The transmission lines corresponding to the first control devices 100 and the second control devices 200 are integrated and deployed. Multiple analog control signals generated by the A first control devices 100 are transmitted to the superconducting quantum processor 300 through the integrated transmission lines and act on the superconducting quantum bits. Multiple analog control signals generated by the B second control devices 200 are transmitted to the superconducting quantum processor 300 through the integrated transmission lines and act on the tunable coupler.

[0124] According to an embodiment of the present invention, the synthesized signal output by control device A is a multi-frequency microwave analog control signal comprising analog control signals for each of the multiple superconducting qubits and intermediate frequency sideband signals. The synthesized signal output by control device B is a multi-frequency AC flux analog control signal comprising second analog control signals for each of the multiple tunable couplers.

[0125] The center frequency of the bandpass filter of the A control device is matched with the frequency of the analog control signal of the corresponding superconducting quantum bit. Its passband completely covers the effective spectrum of the analog control signal of the corresponding quantum gate, and the stopband is used to suppress analog control signals of other frequencies. The output of the bandpass filter is connected to the corresponding quantum bit control port via a transmission component composed of integrated cables.

[0126] The center frequency of the bandpass filter of the second control device B is matched with the frequency of the analog control signal of the corresponding adjustable coupler. Its passband completely covers the effective spectrum of the corresponding coupled control signal, and the stopband is used to suppress analog control signals of other frequencies. The output of the bandpass filter is connected to the control port of the corresponding adjustable coupler via a transmission component composed of integrated cables.

[0127] According to embodiments of the present invention, superconducting qubits and tunable couplers are grouped based on the topology of the superconducting quantum chip in a practical superconducting quantum processor, the control frequency of the first analog control signal controlling the superconducting qubits, the modulation frequency of the tunable coupler, and the concurrent gate operation requirements. Grouping requires not only that no two frequencies exist within the same group, but also that sufficient guard bandwidth be maintained between adjacent frequencies, and that the controlled objects within the same group are allowed to share the same frequency division multiplexing input link in terms of quantum chip topology and concurrent gate timing.

[0128] The bandpass filter in this invention differs from filters used in conventional communication. The bandpass filter in this invention ensures that the quantum gate analog control signal, after being transmitted to the quantum processor, still meets the quantum gate fidelity requirements. Specifically, the center frequency of the bandpass filter matches the center frequency of the first analog control signal of the corresponding superconducting quantum bit, and the passband of the bandpass filter completely covers the effective bandwidth of the first analog control signal of the corresponding superconducting quantum bit. Amplitude fluctuations and group delay fluctuations within the passband are limited to a range that does not disrupt the gate waveform envelope of the analog control signal. The bandpass filter sets a high-attenuation stopband for the frequency range of other analog control signals, achieving sufficient crosstalk suppression. Simultaneously, the input and output ports of the bandpass filter are impedance-matched with the cryogenic control link, reducing signal reflection and transmission crosstalk.

[0129] According to embodiments of the present invention, the center frequency interval of the analog control signals of any two superconducting qubits within each superconducting qubit group is greater than or equal to half the sum of the bandwidths of the analog control signals of the two superconducting qubits. Similarly, the center frequency interval of the analog control signals of any two tunable couplers within each coupler group is greater than or equal to half the sum of the bandwidths of the analog control signals of the two tunable couplers. These embodiments of the present invention limit the center frequency interval of the analog control signals of two superconducting qubits to be greater than or equal to half the sum of the bandwidths of the analog control signals of the two superconducting qubits, and the center frequency interval of the analog control signals of two tunable couplers to be greater than or equal to half the sum of the bandwidths of the analog control signals of the two tunable couplers, thus avoiding spectral overlap of analog control signals within the same group. Combined with the high stopband suppression capability of the bandpass filter, attenuation and isolation of non-target analog control signals are achieved, reducing crosstalk between multiple analog control signals, reducing single-bit gate and double-bit gate errors caused by crosstalk, realizing global overall allocation of spectrum resources, and improving resource utilization.

[0130] Two superconducting qubits in the same superconducting qubit group satisfy the condition in equation (11).

[0131] (11);

[0132] in, This represents the center frequency of the analog control signal corresponding to the a-th superconducting qubit. This represents the center frequency of the analog control signal corresponding to the b-th superconducting qubit. This represents the effective bandwidth of the analog control signal corresponding to the a-th superconducting quantum bit. This represents the effective bandwidth of the analog control signal corresponding to the b-th superconducting qubit. This represents the protection bandwidth determined by combining the bandpass filter roll-off characteristics, system crosstalk index, and quantum gate error tolerance, where a≥1 and b≥1.

[0133] Any two adjustable couplers in the coupler group satisfy the condition in equation (12).

[0134] (12);

[0135] in, This represents the center frequency of the analog control signal corresponding to the c-th adjustable coupler. This represents the center frequency of the analog control signal corresponding to the d-th adjustable coupler. This represents the effective bandwidth of the analog control signal corresponding to the c-th adjustable coupler. This represents the effective bandwidth of the analog control signal corresponding to the d-th adjustable coupler. This represents the protection bandwidth determined by combining the bandpass filter roll-off characteristics, system crosstalk index, and quantum gate error tolerance, where c≥1 and d≥1.

[0136] The control unit can be deployed at any level of the dilution refrigerator, from room temperature to the lowest temperature level. The closer the control unit is to the quantum chip of the superconducting quantum processor, the fewer cryogenic control lines can be required; the closer the control unit is to room temperature, the more advantageous it is to reduce the complexity of the cryogenic module. In actual deployment, the installation temperature level can be determined by comprehensively considering the insertion loss of the control unit, system thermal load, packaging space, maintainability, and the reduction in the number of control lines.

[0137] Suppose that the quantum chip in the superconducting quantum processor includes four superconducting qubits (Q1, Q2, Q3, and Q4) and two tunable couplers (C12 and C34). Superconducting qubits Q1 through Q4 require microwave single-qubit gate control, while tunable couplers C12 and C34 require AC flux modulation to execute two-qubit gates. Superconducting qubits Q1 and Q2 can be assigned to the first superconducting qubit group, Q3 and Q4 to the second superconducting qubit group, and tunable couplers C12 and C34 to the coupler group.

[0138] Taking the first control device as an example, the control module of the first control device generates multi-frequency digital control signals carrying superconducting qubits Q1 and Q2. These signals are input to the input of the frequency division multiplexing module and then split to two bandpass filters via a power divider. The center frequency of one bandpass filter matches the frequency of the analog control signal controlling superconducting qubit Q1, and its output signal is connected to the control port of superconducting qubit Q1. The center frequency of the other bandpass filter matches the frequency of the analog control signal controlling superconducting qubit Q2, and its output signal is connected to the control port of superconducting qubit Q2.

[0139] The second control device operates similarly, receiving analog control signals carrying the modulation frequencies of tunable couplers C12 and C34, along with multi-frequency AC flux. Two bandpass filters within the frequency division multiplexing module filter out the modulation frequencies corresponding to tunable couplers C12 and C34, and output them to the control ports of the respective tunable couplers. This architecture ensures that the single-qubit gate control of the superconducting quantum bits and the two-qubit gate control of the couplers are independent of each other.

[0140] The embodiments of the present invention have been described above. However, these embodiments are merely illustrative and not intended to limit the scope of the invention. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of the invention, and all such substitutions and modifications should fall within the scope of the invention.

Claims

1. A control device for a superconducting quantum processor based on frequency division multiplexing, characterized in that, The control device includes: The control module includes: Multiple finite impulse response filters are used to pre-correct multiple digital control signals; each of the multiple digital control signals corresponds to a multiple controlled object, and the digital control signal carries control information for manipulating the corresponding controlled object; the controlled object is a superconducting quantum bit of a superconducting quantum processor or a tunable coupler. The superposition component is suitable for linearly superimposing multiple pre-corrected digital control signals in the digital domain, and obtaining a composite signal carrying multiple control information based on the composite signal generated after linear superposition, wherein the composite signal is an analog signal; The frequency division multiplexing module is suitable for obtaining multiple analog control signals from the synthesized signal. The multiple analog control signals correspond one-to-one with multiple controlled objects. The analog control signals carry control information for manipulating the corresponding controlled objects. The frequency division multiplexing module includes multiple transmission channels for transmitting the multiple analog control signals to the multiple controlled objects respectively, so as to use the multiple control information to manipulate the multiple controlled objects. Among them, the pre-corrected digital control signal is used to compensate for the changes in control information caused by the transmission channel during the transmission of analog control signals.

2. The control device according to claim 1, characterized in that, The control device further includes a processor configured to: Based on the transfer function values ​​of the i-th transmission channel at multiple frequency points, determine the time-domain impulse response sequence of the i-th finite impulse response filter; Based on the time-domain impulse response sequence of the i-th finite impulse response filter, determine multiple tap coefficients of the i-th finite impulse response filter; Multiple finite impulse response filters are configured to pre-correct multiple digital control signals using the following method: The i-th digital control signal is pre-corrected by convolving multiple tap coefficients of the i-th finite impulse response filter, where i ≥ 1.

3. The control device according to claim 2, characterized in that, The processor determines the time-domain impulse response sequence of the i-th finite impulse response filter using the following method: Based on the magnitude values ​​of the transfer function of the i-th transmission channel at multiple frequency points, determine the frequency points whose magnitude values ​​are greater than a preset threshold from the multiple frequency points to obtain multiple target frequency points; The pre-correction coefficients for each target frequency point are obtained by taking the reciprocal of the transfer function of the i-th transmission channel at each of the multiple target frequency points. Perform an inverse discrete Fourier transform on the pre-correction coefficients at each target frequency point to obtain the time-domain impulse response sequence of the i-th finite impulse response filter.

4. The control device according to claim 2, characterized in that, The processor determines multiple tap coefficients of the i-th finite impulse response filter using the following method: Perform a cyclic shift operation on the time-domain impulse response sequence of the i-th finite impulse response filter so that zero time delay is at the center of the time-domain impulse response sequence, and obtain the shifted time-domain impulse response sequence; Extract the target time-domain impulse response sequence from the shifted time-domain impulse response sequence, such that the center of the target time-domain impulse response sequence coincides with the center of the shifted time-domain impulse response sequence; Windowing is applied to the target time-domain impulse response sequence to obtain multiple tap coefficients of the i-th finite impulse response filter.

5. The control device according to claim 2, characterized in that, When the controlled object is a superconducting quantum bit, the superposition component includes: The first digital adder is suitable for linearly superimposing multiple pre-corrected digital control signals in the digital domain to generate a composite signal. A mixer is used to mix the composite signal with the local oscillator signal to complete the up-conversion processing of the composite signal. The first arbitrary waveform generator is suitable for performing digital-to-analog conversion on the up-converted composite signal to generate the composite signal.

6. The control device according to claim 2, characterized in that, When the controlled object is an adjustable coupler, the superposition component includes: The second digital adder is used to linearly superimpose multiple pre-corrected digital control signals in the digital domain to generate a composite signal. The second arbitrary waveform generator is suitable for performing digital-to-analog conversion on composite signals to generate the composite signal.

7. The control device according to claim 1, characterized in that, The frequency division multiplexing module includes: A power divider suitable for dividing the synthesized signal into multiple sub-signals; Multiple bandpass filters are one-to-one with multiple sub-signals. The multiple bandpass filters are used to filter the multiple sub-signals respectively to obtain multiple analog control signals. Multiple transmission lines are associated with multiple bandpass filters, and the transmission lines and corresponding bandpass filters form the transmission channels to transmit multiple analog control signals to multiple controlled objects.

8. A control system, characterized in that, This control system is applied to a superconducting quantum processor, which includes A groups of superconducting qubits and B groups of couplers. Each group of superconducting qubits comprises multiple superconducting qubits, and each group of couplers comprises multiple couplers. The control system includes: M control devices as described in any one of claims 1 to 7, wherein A control devices correspond one-to-one with A superconducting qubit groups, and the multiple analog control signals generated by each of the A control devices are used to manipulate the multiple superconducting qubits of the corresponding superconducting qubit group; B control devices correspond one-to-one with B coupler groups, and the multiple analog control signals generated by each of the B control devices are used to manipulate the multiple adjustable couplers of the corresponding coupler groups, and A+B=M.

9. The control system according to claim 8, characterized in that, The center frequency interval of the analog control signal of any two superconducting qubits within each superconducting qubit group is greater than or equal to half the sum of the bandwidths of the analog control signals of the two superconducting qubits.

10. The control system according to claim 8, characterized in that, The center frequency interval of the analog control signals of any two adjustable couplers within each coupler group is greater than or equal to half the sum of the bandwidths of the analog control signals of the two adjustable couplers.

Images