CN117808108B

CN117808108B - Quantum noise relieving method and device, storage medium and electronic equipment

Info

Publication number: CN117808108B
Application number: CN202410223282.5A
Authority: CN
Inventors: 于云龙; 李辰; 张新; 赵雅倩
Original assignee: Suzhou Metabrain Intelligent Technology Co Ltd
Current assignee: Suzhou Metabrain Intelligent Technology Co Ltd
Priority date: 2024-02-28
Filing date: 2024-02-28
Publication date: 2024-05-24
Anticipated expiration: 2044-02-28
Also published as: CN117808108A

Abstract

The embodiment of the application provides a method and a device for relieving quantum noise, a storage medium and electronic equipment, wherein the method comprises the following steps: acquiring a first measured value of a quantum circuit of the quantum equipment operated in a noise state through a preset measurement operator, and acquiring a plurality of second measured values of a target quantum circuit corresponding to the quantum circuit operated in different noise intensities through the preset measurement operator under the condition of carrying out multiple probability adjustment, wherein the target quantum circuit is generated after carrying out equivalent replacement on two bit gates on the quantum circuit, zero noise extrapolation is carried out according to probability average values corresponding to the first measured value and the plurality of second measured values, so as to obtain a target expected value, and a noise alleviation result corresponding to the quantum equipment is determined according to the target expected value. The problems that the accuracy of error alleviation of the quantum equipment is low, the resource cost of the quantum equipment is high, and the effect of noise alleviation is poor are solved.

Description

Quantum noise relieving method and device, storage medium and electronic equipment

Technical Field

The embodiment of the application relates to the technical field of artificial intelligence, in particular to a method and a device for relieving quantum noise, a storage medium and electronic equipment.

Background

In recent years, the quantum information industry has rapidly developed, and quantum computers have exhibited quantum advantages existing in quantum computers to varying degrees. But is limited by the experimental setup below, the number of qubits that can be achieved at present is not very large, and there is also much noise. These limit the current large-scale applications of quantum computers. And aiming at the noise problem, the currently adopted main methods are quantum error mitigation methods, including zero noise extrapolation and the like. According to the method, firstly, a series of replacement is carried out on the two-bit gates, then the result without noise is reversely deduced according to the calculated results before and after the replacement, but the noise exists to cause a certain gap between the reversely deduced result and an ideal value. Therefore, the current mainstream method has low accuracy in error mitigation for the quantum computer and poor effect in noise mitigation.

Aiming at the problems of low accuracy of error alleviation on quantum equipment (such as a quantum computer), high resource expense of the quantum equipment and poor effect of noise alleviation in the related art, no effective solution has been proposed at present.

Content of the application

The embodiment of the application provides a method and a device for relieving quantum noise, a storage medium and electronic equipment, which are used for solving the problems that the accuracy of error relieving on the quantum equipment is low, the resource cost of the quantum equipment is high, and the effect of noise relieving is poor in the related technology.

According to an embodiment of the present application, there is provided a method of mitigating quantum noise, including: acquiring a first measured value of a quantum circuit of a quantum device operated in a noise state through a preset measurement operator, and acquiring a plurality of second measured values of a target quantum circuit corresponding to the quantum circuit operated in different noise intensities through the preset measurement operator under the condition of carrying out multiple probability adjustment, wherein the target quantum circuit is generated after carrying out equivalent replacement on two bit gates on the quantum circuit; zero noise extrapolation is carried out according to probability average values corresponding to the first measured value and the second measured values, a target expected value is obtained, and a noise alleviation result corresponding to the quantum equipment is determined according to the target expected value.

In an exemplary embodiment, before zero noise extrapolation from the probability average corresponding to the first measurement and the second measurement, the method further includes sending target data to a computing device connected to the quantum device, where the target data includes at least: a first measurement value, a second measurement value; and under the condition that a loss function is constructed through the first measured value and the second measured value on the computing equipment and the target data is present in the computing equipment, the quantum circuits on the quantum equipment are determined to be replaced by two bit gates with different numbers, a plurality of target quantum circuits are obtained, and the preset measuring operators are instructed to measure the plurality of target quantum circuits.

In one exemplary embodiment, determining to perform a different number of two-bit gate substitutions on the quantum wire on the quantum device includes: determining the sampling number of the number of two-bit gates to be subjected to two-bit gate replacement on the quantum circuit; and determining a target quantum circuit according to the sampling number.

In an exemplary embodiment, the above method further comprises: acquiring a connection relationship between the quantum device and the computing device; and under the condition that the connection relation is determined to be effective, establishing a data channel of the quantum equipment and the computing equipment, wherein the data channel is used for efficiently transmitting interaction data of the quantum equipment and the computing equipment.

In an exemplary embodiment, the above method further comprises: determining a replacement distribution for performing equivalent replacement on the quantum circuit; under the condition that P first types of quantum bit gates existing in the quantum circuit are identified according to the replacement distribution, replacing each first type of quantum bit gate by using 2n+1 second types of two bit gates, and repeating the replacement R times to obtain R replacement results corresponding to each first type of quantum bit gate, wherein P, R and n are positive integers, and the first type of quantum bit gate is a target two-bit gate to be subjected to equivalent replacement in the quantum circuit; and generating a target quantum circuit corresponding to the quantum circuit based on R replacing results corresponding to each of the quantum bit gates of the P first types.

In an exemplary embodiment, after replacing each of the first type of qubit gates with 2n+1 second type of two bit gates, the method further comprises: comparing a target replacement value with a preset replacement value, wherein the target replacement value is a quantity value of all two-bit gates in a quantum circuit after the replacement operation is completed; determining that the quantum circuit is too deep and prohibiting the quantum circuit from running on the quantum device when the target replacement value is greater than the preset replacement value; and under the condition that the target replacement value is smaller than or equal to the preset replacement value, determining that the quantum circuit meets operation conditions, and allowing the quantum circuit to operate on the quantum device.

In an exemplary embodiment, before replacing each of the first type of qubit gates with 2n+1 second type of two bit gates, the method further comprises: determining a first error rate corresponding to the two-bit gates of the first type and a second error rate corresponding to the whole of the two-bit gates of the 2n+1 second types; determining the magnitude relation between the first error rate, the second error rate and the maximum error rate in a preset error rate range respectively; and determining whether the replacement is a valid two-bit gate replacement according to the size relationship.

In one exemplary embodiment, determining whether the replacement is a valid two-bit gate replacement according to the size relationship comprises: determining that a 2n+1 second type two-bit gate replacement is currently used for each first type of qubit gate with a valid two-bit gate replacement if the first error rate is less than or equal to the maximum error rate and the second error rate is less than or equal to the maximum error rate; in the event that the second error rate is greater than the maximum error rate, it is determined that a replacement is currently performed for each of the first type of qubit gates using 2n+1 second type of two-bit gates with an invalid two-bit gate replacement.

In an exemplary embodiment, before replacing each of the first type of qubit gates with 2n+1 second type of two bit gates, the method further comprises: determining a number of substitutions the quantum device allows for performing the substitution in parallel; grouping the target two-bit gates of the first type according to the replacement number to obtain a plurality of parallel replacement groups; and simultaneously replacing the first type of target qubit gates in the plurality of parallel replacement groups on the quantum device.

In an exemplary embodiment, before determining the substitution distribution for equivalent substitution of the quantum wire, the method further includes: determining parameters to be optimized, which are input by a target object on the computing device, wherein the parameters to be optimized have the following relation with a loss function: Wherein said/> For a target expected value corresponding to a parameter to be optimized, f is a loss function, the target expected value is an amount which has an exact physical meaning, is related to a noiseless state in the quantum device and is easy to process by the computing device.

In one exemplary embodiment, determining a substitution distribution for equivalent substitution of the quantum wire includes: acquiring probability distribution corresponding to the parameter to be optimized; and under the condition that the number of substitutions corresponding to the probability distribution is smaller than the preset allowable number of substitutions, sampling the number of substitutions of each two-bit gate on the quantum circuit for R times according to the probability distribution to obtain first sub-distribution, and under the condition that Q two-bit gates coexist on the quantum circuit, determining the substitution distribution for carrying out equivalent substitution on the quantum circuit based on the Q first sub-distribution, wherein R, Q is a positive integer.

In an exemplary embodiment, the method further includes obtaining, by a preset measurement operator, a first measurement value of a quantum device operating a quantum wire in a noise state, the method further including: in the case of determining that a plurality of quantum wires are simultaneously operated and measured in the quantum device, determining a preset number of quantum wires that the quantum device is allowed to be operated and measured in parallel; controlling the preset measuring operators to measure the quantum circuits of the quantum equipment running in the noise state to obtain a plurality of first sub-measured values and/or second sub-measured values; summarizing the plurality of first sub-measured values to obtain the first measured value, and/or summarizing the plurality of second sub-measured values to obtain the second measured value.

In an exemplary embodiment, after zero noise extrapolation is performed according to the probability average values corresponding to the first measurement value and the plurality of second measurement values to obtain a target expected value, so as to determine a noise mitigation result corresponding to the quantum device according to the target expected value, the method further includes: instructing the computing device to confirm convergence of a loss function if the target expected value is present on the computing device; and under the condition that a confirmation instruction sent by the computing device is received, determining whether noise adjustment of the quantum device is completed or not according to the confirmation instruction.

In an exemplary embodiment, the above method further comprises: determining that noise adjustment of the quantum device is completed under the condition that the confirmation instruction indicates that the current target expected value meets the convergence requirement of the loss function; and under the condition that the confirmation instruction indicates that the current target expected value does not meet the convergence requirement of the loss function, determining that the noise adjustment of the quantum device is not completed, and indicating the computing device to continuously perform iterative optimization on the loss function.

In an exemplary embodiment, after receiving the target data sent by the computing device connected to the quantum device, the method further includes: analyzing the target data through a computing device configured in the quantum device; and determining gradient information of the loss function to be optimized on the quantum equipment according to the analysis result.

In an exemplary embodiment, after zero noise extrapolation is performed according to the probability average values corresponding to the first measurement value and the second measurement values, the method further includes: acquiring an optimal target expected value of the computing equipment and the quantum equipment for converging a loss function; acquiring key information generated in the process of determining the optimal target expected value, wherein the key information at least comprises one of the following: gradient change information, probability change information, target expected value change information; storing the critical information in a memory preconfigured for the computing device and the quantum device.

In an exemplary embodiment, the above method further comprises: determining an output structure corresponding to the quantum equipment; and outputting a measurement result of a preset measurement operator after noise alleviation according to the output structure.

In an exemplary embodiment, before zero noise extrapolation according to the probability average value corresponding to the first measurement value and the second measurement values, the method further includes: determining a linear extrapolation formula for performing the zero noise extrapolation; wherein the linear extrapolation formula isSaidFor measuring the measurement value of a measuring operator under different noise states on a quantum computer,For the first measurement,Is the probability mean of the second measurement value,The parameters to be optimized carry probability characteristics.

In an exemplary embodiment, the above method further comprises: under the condition that a target expected value corresponding to the quantum equipment is determined and the target expected value is continuously optimized, acquiring a new target expected value of the quantum equipment after the quantum equipment is optimized; and comparing the difference between the target expected value and the new target expected value to determine whether the optimization of the target expected value is effective according to the difference.

According to another embodiment of the present application, there is provided a quantum noise mitigation device including: the device comprises an acquisition module, a detection module and a control module, wherein the acquisition module is used for acquiring a first measured value of a quantum circuit operated by quantum equipment in a noise state through a preset measurement operator, and acquiring a plurality of second measured values of a target quantum circuit corresponding to the quantum circuit operated by the quantum equipment in different noise intensities through the preset measurement operator under the condition of carrying out multiple probability adjustment, wherein the target quantum circuit is a circuit generated after equivalent replacement of a two-bit gate on the quantum circuit; and the extrapolation module is used for carrying out zero noise extrapolation according to probability average values corresponding to the first measured value and the plurality of second measured values to obtain a target expected value so as to determine a noise alleviation result corresponding to the quantum equipment according to the target expected value.

According to a further embodiment of the application, there is also provided a computer readable storage medium having stored therein a computer program, wherein the computer program is arranged to perform the steps of any of the method embodiments described above when run.

According to a further embodiment of the application there is also provided an electronic device comprising a memory having stored therein a computer program and a processor arranged to run the computer program to perform the steps of any of the method embodiments described above.

According to a further embodiment of the application, there is also provided a computer program product comprising a computer program which, when executed by a processor, implements the steps of any of the method embodiments described above.

According to the application, the corresponding loss function is set in the computing device, the quantum device and the computing device are used for executing the operation of optimizing the specific loss function to weaken the influence caused by noise, namely, the variable component sub-algorithm is combined with the random congruent insertion zero-noise extrapolation method, and the advantages of the computing device and the quantum device are fully exerted by designing the loss function to be optimized and the corresponding optimization flow, so that the noise in the running process of the quantum device is relieved, and the problems of lower accuracy of error relief on the quantum device, higher resource expenditure of the quantum device and poorer effect of noise relief are solved, the influence of noise is weakened in the running process of the quantum device, the running effect of higher fidelity is obtained, and the quantum device can support the quantum algorithm with higher running accuracy.

Drawings

FIG. 1 is a block diagram of a hardware architecture of a quantum computer of a method of mitigating quantum noise according to an embodiment of the present application;

FIG. 2 is a flow diagram of a method of mitigating quantum noise according to an embodiment of the application;

FIG. 3 is a flow chart of a zero noise extrapolation optimization method in accordance with an embodiment of the present application;

FIG. 4 is a schematic diagram of a quantum circuit considered for numerical computation according to an embodiment of the present application;

FIG. 5 is a graph of noise intensity at different levels in a quantum circuit by two methods according to an embodiment of the application Schematic of (2);

Fig. 6 is a block diagram of a quantum noise mitigation device according to an embodiment of the application.

Detailed Description

Embodiments of the present application will be described in detail below with reference to the accompanying drawings in conjunction with the embodiments.

It should be noted that the terms "first," "second," and the like in the description and the claims of the present application and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order.

The method embodiments provided in the embodiments of the present application may be performed in a quantum computer or similar computing device. Taking the example of running on a quantum computer, fig. 1 is a hardware block diagram of a quantum computer of a method for alleviating quantum noise according to an embodiment of the present application. As shown in fig. 1, a quantum computer may include one or more processors 102 (only one is shown in fig. 1) (the processor 102 may include, but is not limited to, a microprocessor MCU, a programmable logic device FPGA, or the like) and a memory 104 for storing data, wherein the quantum computer may further include a transmission device 106 for communication functions and an input-output device 108. It will be appreciated by those of ordinary skill in the art that the structure shown in fig. 1 is merely illustrative and is not intended to limit the structure of the quantum computer described above. For example, a quantum computer may also include more or fewer components than shown in fig. 1, or have a different configuration than shown in fig. 1.

The memory 104 may be used to store a computer program, for example, a software program of application software and a module, such as a computer program corresponding to a method for mitigating quantum noise in an embodiment of the present application, and the processor 102 executes the computer program stored in the memory 104, thereby performing various functional applications and data processing, that is, implementing the method described above. Memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, memory 104 may further include memory remotely located relative to processor 102, which may be connected to the quantum computer through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The transmission device 106 is used to receive or transmit data via a network. Specific examples of the network described above may include wireless networks provided by communication providers of quantum computers. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, simply referred to as a NIC) that can connect to other network devices through a base station to communicate with the internet. In one example, the transmission device 106 may be a Radio Frequency (RF) module, which is configured to communicate with the internet wirelessly.

In this embodiment, a method for mitigating quantum noise is provided, and fig. 2 is a flowchart of a method for mitigating quantum noise according to an embodiment of the present application, as shown in fig. 2, where the flowchart includes the following steps:

Step S202, a first measured value of a quantum circuit operated by quantum equipment in a noise state is obtained through a preset measurement operator, and a plurality of second measured values of a target quantum circuit corresponding to the quantum circuit operated by the quantum equipment in different noise intensities are obtained through the preset measurement operator under the condition of carrying out multiple probability adjustment, wherein the target quantum circuit is a circuit generated after equivalent replacement of a two-bit gate on the quantum circuit;

and S204, performing zero noise extrapolation according to probability average values corresponding to the first measured value and the second measured values to obtain a target expected value, so as to determine a noise alleviation result corresponding to the quantum equipment according to the target expected value.

As an alternative embodiment, the above-mentioned process of determining the target expected value may further be as follows:

step one, under the condition of carrying out probability adjustment for a plurality of times, acquiring a plurality of first measured values of a quantum circuit operated by quantum equipment in a noise state through a preset measurement operator; acquiring a plurality of second measured values of a target quantum circuit corresponding to the quantum circuit operated by the quantum equipment under different noise intensities through the preset measuring operators;

And step two, zero noise extrapolation is carried out according to the first probability average value corresponding to the plurality of first measurement values and the second probability average value corresponding to the plurality of second measurement values, so as to obtain a target expected value.

Optionally, the extrapolation formula for zero noise extrapolation by the first probability average and the second probability average is as follows:， Is the probability mean of the first measurement value,/> Is the probability mean of the second measurement value,The parameters to be optimized carry probability characteristics.

Through the steps, the corresponding loss function is set in the computing equipment, the quantum equipment and the computing equipment are utilized to execute the operation of optimizing the specific loss function to weaken the influence caused by noise, namely, a variable component sub-algorithm is combined with a random congruent insertion zero-noise extrapolation method, and the advantages of the computing equipment and the quantum equipment are fully exerted through designing the loss function to be optimized and the corresponding optimization flow, so that the noise in the running process of the quantum equipment is relieved, and the problems that the accuracy of error relief on the quantum equipment is lower, the resource cost of the quantum equipment is higher, and the effect of noise relief is poor are solved, the influence of noise is weakened in the running process of the quantum equipment, the running effect of higher fidelity is obtained, and the quantum equipment can support the quantum algorithm with higher running accuracy.

The main body of execution of the above steps may be a quantum computer, a device including a quantum computer, or the like, but is not limited thereto.

It can be understood that the corresponding sampling probability and the loss function are set on the computing device, then the computing task is provided for the quantum device through the computing device, and the quantum device runs the corresponding quantum circuit under the condition that the quantum device acquires the target data, and the optimal state of the current quantum device on the target data is known in real time through measurement.

In summary, through the above embodiment, the idea of a variable component sub-algorithm is adopted, and the proposed random congruent insertion zero noise extrapolation optimization method combines different tasks executed on the quantum device and the computing device, so as to realize the alleviation of the quantum noise existing in the quantum device, that is, the algorithm divides the computing task into two parts, the computing which can be efficiently completed by a classical computer (equivalent to the computing device in the above embodiment) is delivered to the classical computer for processing, and the task which can be efficiently processed by the quantum computer (equivalent to the quantum device in the above embodiment) is delivered to the quantum computer for processing, thereby reducing the dependence on quantum resources, and enabling the quantum computer to be maximally used for processing complex tasks.

Alternatively, the second type of two-bit gate described above may comprise a different two-bit gate, for example, the replacement of one CNOT gate may be done using 3 CNOT gates, i.e. 1 cnot= CNOT CNOTCNOT, or the replacement of one CNOT gate may be done using two CZ gates and one CNOT gate, i.e. 1 cnot=cz cnot=cnot CZ. Wherein CZ is contorolled-Z gate, controlled Z gate, if the target bit isThen a Z operation is performed on the control bits.

The EM method includes probabilistic error cancellation (Probabilistic error cancellation, error mitigation, abbreviated as EM), congruent insertion (Identity Insertion), and the like. In the probabilistic error elimination method, the noise generated quantum channelRandomly inserting the Brix arithmetic in the front and back of the channel, converting the quantum channel containing unknown noise into the quantum channel containing Brix noise, and measuring to obtain the noise characteristic information/>, of the new quantum channelCharacteristic information hereCorresponding to the probability of occurrence of each bubble error. With this information, one can extrapolateIn the form of (a). AtThe "probability" of each bubble appearing is a quasi-probability. In general, an exponential measurement is required to obtain noise characteristic informationThis means that this may not be an efficient method.

In contrast, the EM method of congruent insertion (Identity Insertion) does not require knowledge of the specific information of the noise. This method mainly focuses on a two-qubit gate in a quantum circuit, taking a CNOT gate (Control NOT, simply referred to as a CNOT gate, equivalent to a two-bit gate in this embodiment) as an example. Assuming that Q CNOT gates are contained in the quantum circuit in total, taking the ith CNOT gate into consideration, and recording the noise intensity asR=2n+1 (n is a natural number) CNOT gates can be used in place of the original CNOT gates, so that the noise of the quantum line can be amplified without affecting the logic of the original quantum line. If the operator to be measured is M, at noise intensitySmaller, under first order approximation, ideally noise-free time operatorMeasured valueIn linear relation to the number r of CNOT gates to replace. The operator/>, under different noisy conditions, can be obtained by setting r=1, 3Measured valueThen, the linear interpolation fitting is adopted to obtain. But this approach requires 3Q CNOT gates, which also presents challenges to current NISQ (Noisy Intermediate-Scale Quantum, NISQ) devices. This method is also called fixed congruent insertion FIIM (Fixed Identity Insertion Method, fixed congruent insertion method, FIIM for short).

In order to reduce the number of CNOT gates required, one solution is to consider n in the above scheme as a random variable, i.e. for the firstIndividual CNOT gate usageThe number of CNOT gates to replace is then in this case on average only Q+2 CNOT gates are required, which greatly reduces the number of quantum gates that need to be inserted compared to the last congruent insertion method. The random variable/>, can be chosen according to a certain probability distributionFor example as per mean valueRandomly choose，Then in the sense of statistical averaging, one can get the sumLowerAverage valueAnd linearly extrapolates the noiseless result by altering the probability of sampling the poisson distribution. The poisson distributed sampling method not only maintains the characteristics of a small number of CNOT gates, but also has a convenient data post-processing process of zero-noise extrapolation.

Alternatively, this method is also referred to as random congruent insertion RIIM (Random Identity Insertion Method, random congruent insertion method, RIIM). However, in this method, it is required to go from 0 toPairSampling is performed, but there is always a truncation in the actual process, which requires the probability characteristic parameterThere is a precise control so that the truncation error is as small as possible. Meanwhile, if the number of sampling times for single CNOT gate replacement is denoted as R, in order to satisfy poisson distribution as much as possible, the larger the sampling times are, the better, which results in a rapid increase in the total sampling times with an increase in Q.

Alternatively, all pairs obeying the poisson distribution are setMaximum value reached by samplingAnd the number of samples R required for the Poisson distribution, and generates samples. For noisy quantum circuits, consider the CNOT gate (control NOT gate) therein, for the ith CNOT gate, one can useThe CNOT gates replace it, and the process is repeated until all CNOT gates are replaced. Quantum line measurement operator/>, after CNOT gate replacementObtain statistical average. Then the original noisy quantum line, i.e. the line without CNOT gate replacement, is run and the expected value/>, of operator M is measured. Then according toAndOperator/>, when noiseless, by linear interpolationDesired value. The/>, obtained at this timeAnd correspondingReturning to the classical computer and calculating the loss function: /(I)。

F is minimized by classical computer optimization v. Note that the calculationThe task of (2) is on quantum computers, while the search for an optimized direction for v is done on classical computers.

In summary, through the above embodiment, the number of CNOT gates to be replaced is calculated according to the Poisson distributionSampling is performed, and the influence of quantum noise is counteracted to a certain extent by using sampling errors of limited sampling, so that the algorithm can be improved in a mode of optimization.

It will be appreciated that the number of two-bit gates used to replace a single two-bit gate cannot be used in excess of two-bit gates to avoid causing the quantum circuit to be too deep to run on NISQ devices, thereby ensuring the effectiveness of the operation.

In summary, through the above embodiment, the number of the two-bit gates of the first type to be replaced is determined in real time, so as to avoid the situation that the dependent quantum algorithm cannot normally operate.

That is, the error rate is not too high for the two-bit gates in the quantum wire, resulting in meaningless measurements due to the presence of noise when no or only a small number of two-bit gate substitutions are made.

It should be noted that, for two-bit gates in a quantum circuit, the sequential operation of a plurality of two-bit gates should be logically equivalent to the operation of only one two-bit gate, that is, before and after the replacement of two-bit gates, only one two-bit gate is replaced with an odd number of two-bit gates that are continuously executed, and the logic of the corresponding two-bit gates is unchanged before and after the replacement.

In summary, by determining the error rate of the two-bit gates of different types, it is ensured that an effective measurement value can be obtained after the replacement operation is performed.

In short, there should be some parallelism for the process of replacing a single two-bit gate, thereby reducing the runtime on a quantum computer.

In summary, through the above embodiment, by grouping in parallel, the replacement efficiency of the two-bit gates on the quantum circuit is improved, so that when the replacement operation is performed, the parallel replacement can be rapidly performed, and the running time on the quantum computer is reduced.

For example, assuming that there are 3 two-bit gates on the current quantum wire, the number of substitutions for each two-bit gate can be determined by treating the optimized parametersSampling the corresponding probability distribution, further determining a replacement sub-distribution corresponding to the replacement number of each two-bit gate, and summarizing the replacement sub-distribution of the replacement number of each two-bit gate to obtain a replacement distribution of the equivalent replacement of 3 two-bit gates;

In addition, it should be noted that when the equivalent replacement of the two-bit gates is performed according to the replacement distribution, the 3 two-bit gates may perform the same number of types of replacement of the two-bit gates, for example, 3 two-bit gates are performed for each of the 3 two-bit gates, 5 two-bit gates and 7 two-bit gates are equivalently replaced, or different types of two-bit gate replacement may be performed, for example, the first two-bit gate in the 3 two-bit gates is equivalently replaced by 3 two-bit gates, 5 two-bit gates and 7 two-bit gates, respectively; the second two-bit gate in the 3 two-bit gates is equivalently replaced by 3 two-bit gates and 9 two-bit gates respectively; and the third two-bit gate in the 3 two-bit gates is equivalently replaced by 5 two-bit gates and 7 two-bit gates respectively.

That is, in the process of performing the equivalent substitution, logic consistency of each two-bit gate before and after the substitution needs to be ensured, but whether the two-bit gates to be substituted adopt the same equivalent substitution is only related to sampling of the probability distribution, and the logic consistency can be flexibly changed according to the sampling result.

As an optional implementation manner, a flowchart of a random congruent insertion zero-noise extrapolation optimization method for noise mitigation provided by the present application is shown in fig. 3, and fig. 3 is a schematic flow chart of a zero-noise extrapolation optimization method according to an embodiment of the present application, where a specific implementation manner is as follows:

s1: fixed poisson distribution cut-off And the number of samples R of the poisson distribution.

S2: setting a certain average value of poisson distribution (namely parameters to be optimized) Is set to be a constant value.

S3: for a noisy quantum computer actually executing a quantum algorithm, for an ith CNOT gate, according to probabilityRandom generation of R differentValueThe ith CNOT gate is replaced by/>, respectivelyAnd CNOT gates, and the same operation is performed on all CNOT gates.

S4: probability means for the quantum wire measurement operator M obtained in step S3；

S5: reconsidering the original noisy quantum circuit for running the quantum algorithm, and obtaining the expected value of the operator M without performing any CNOT gate replacement measurement。

S6: estimating operators using zero noise extrapolationExpected value in the absence of noise。

S7: results obtained by Quantum computersI.e. correspondingOutputting to classical computer, completing the task by classical computer according to flow, calculating/>, on classical computerAnd records the current v. The next value of v is determined by optimizing f.

S8: if the optimization in S7 reaches convergence, then output at this timeAnd ending the flow, completing the noise relieving process of the quantum computer, otherwise returning to S2 to optimize again. And (4) carrying out noise alleviation on the computer, otherwise, returning to S2 to carry out optimization again.

In summary, the noise in the actual system is compensated by utilizing the truncation error of the poisson distribution, so that the optimization can be carried out. The algorithm feasibility and noise robustness are improved by utilizing the data to be processed of the classical computer in the variation process; the method is applied to NISQ equipment to weaken the influence of noise, and a result with higher fidelity is obtained.

In an exemplary embodiment, before zero noise extrapolation according to the probability average value corresponding to the first measurement value and the second measurement values, the method further includes: determining a linear extrapolation formula for performing the zero noise extrapolation; wherein the linear extrapolation formula is，For measuring the measurement value of a measuring operator under different noise states on a quantum computer,For the first measurement,Is the probability mean of the second measurement value,The parameters to be optimized carry probability characteristics.

In conclusion, by performing error mitigation on the quantum computer, a higher quality result is obtained. The probability of random congruent insertion is optimized on a classical computer, and corresponding states (namely the target quantum circuits in the embodiment) are prepared on the quantum computer according to the optimized result, so that a better error mitigation result is obtained.

It will be apparent that the embodiments described above are merely some, but not all, embodiments of the application. In order to better understand the above method for alleviating quantum noise, the following description will explain the above process with reference to the embodiments, but is not intended to limit the technical solution of the embodiments of the present application, specifically:

for a better understanding of embodiments of the present application, a brief description of the field of application and implementation of the present application will now be presented.

In dealing with specific mathematical problems, quantum computers and quantum algorithms can achieve exponential acceleration, also known as quantum dominance, compared to classical supercomputers and classical algorithms. Quantum computers and quantum algorithms have been hot problems of industry research due to the existence of quantum advantages.

The current mainstream way to perform quantum computation is to apply a series of quantum gate operations on the qubit, and this process can be represented by an abstract quantum circuit model consisting of dotted lines. Similar to 0 and 1 in classical bits, quantum states are usedAndTo mark the qubits. For quantum gates, one can divide into single quantum gates acting on single qubits, e.g., NOT gate,And a double quantum gate acting on a double qubit such as a Control Not (CNOT) gate if the control bit isThen do/>, for the target bitAnd (3) operating. From the above analysis, it can be seen that if one qubit is acted upon twice not gateThe original state of the CNOT gate is not changed by the operation, and similarly, the original state of the CNOT gate is not changed if the CNOT gate is operated twice on two quantum bits. This is an important property to be used in the present application.

To realize quantum algorithms with definite quantum advantages, it requires quantum bits andThe magnitude of gate fidelity is used for detecting and correcting errors generated in the operation process, and the requirement on the quantity and quality of qubits is far more than the capability of quantum equipment in the NISQ times at present.

For the error correction process on quantum computers, the core idea is to encode a logical qubit with a series of physically implemented qubits, while ensuring that the error rate of the quantum gate needs to be below a certain threshold, so that any error can be detected and corrected with the help of a quantum error correction code. But for the current NISQ equipment, it is possible to implement approximatelyMagnitude qubits, while the average fidelity of a two-bit quantum gate is probablyLarge scale quantum error correction cannot be achieved using such devices, which also means that errors in the actual quantum algorithm execution are not avoided, which presents challenges to the reliability of current quantum devices. /(I)

In order to reduce the influence of noise as much as possible and improve the reliability of NISQ devices, researchers have proposed a class of Error Mitigation (EM) methods, which do not rely on the standard flow of detecting and correcting errors, which is applied to quantum Error correction, and the core idea is that by running noisy quantum circuits and obtaining noisy results, the results are processed to reversely derive the results in the case of no noise, and the EM methods greatly improve the feasibility of running quantum algorithms on NISQ devices, and provide a powerful means for weakening the influence of noise and realizing quantum advantages, which is also one of the main purposes of the present application.

In order to solve the problem, the application refers to the idea of a variable component sub-algorithm and provides a random congruent insertion zero-noise extrapolation optimization method for noise alleviation, which weakens the influence of quantum noise by using a cut-off error of limited sampling, and simultaneously, sets the optimization processCan reduce the total sampling times toCompared with the congruent insertion method, the method can also greatly improve the result of the zero noise extrapolation method under the condition of processing a larger error rate. The method also provides a new thought for the quantum algorithm with increased NISQ equipment reliability and higher operation accuracy. Meanwhile, as the variable component sub-algorithm is one of the main stream sub-algorithms running on NISQ equipment, the method provided by the application can also directly run on NISQ equipment, so that a result with higher fidelity is obtained on NISQ equipment.

The application mainly comprises the following points: the first point is that a variable component sub-algorithm is combined with a random congruent insertion zero-noise extrapolation method, a random congruent insertion zero-noise extrapolation optimization method for noise alleviation is provided, and the method can be directly applied to NISQ equipment to weaken the influence of noise, and a result with higher fidelity is obtained; the second point is that a data to be processed of a classical computer for realizing a variation process is designed, so that the feasibility and noise robustness of the method are improved; third, a method is proposed that uses limited sampling errors to attenuate the effects of quantum noise to further improve the feasibility of the method.

Specifically, the application provides a random congruent insertion zero-noise extrapolation optimization method for noise alleviation, which can be directly applied to the current NISQ equipment for noise alleviation. Noise inevitably exists in an actual quantum computer, and when a certain quantum algorithm is operated on the noisy quantum computer or a noisy quantum circuit, the expected value of a certain measured operator M is changed due to the existence of the noise, so that the influence brought by the noise can be relieved by adopting the random congruent insertion zero noise extrapolation optimization method for noise alleviation.

The flow of the random congruent insertion zero noise extrapolation optimization method provided by the application comprises a classical computer and a quantum computer, and all the methods obeying the poisson distribution are firstly setMaximum value reached by samplingAnd the number of samples R required for the Poisson distribution, and generates samples. For noisy quantum circuits, consider the CNOT gate therein for theA CNOT gate, which can useThe CNOT gates replace it, and the process is repeated until all CNOT gates are replaced. Obtaining a statistical average value/>, for expected values of quantum line measurement operators M after CNOT gate replacement. Then the original noisy quantum line, i.e. the line without CNOT gate replacement, is run and the expected value/>, of operator M is measured. Then according toAndObtaining the expected value/>, of the operator M when noiseless by linear interpolation. The/>, obtained at this timeAnd correspondingReturning to the classical computer and calculating the loss function: /(I); F is minimized by classical computer optimization v. Note that calculationIs a task on quantum computers, seeking pairsThe optimization direction of (c) is done on a classical computer.

After the optimization converges, the result isThe expected value of the measurement operator under the noiseless condition is obtained by a random congruent interpolation zero noise extrapolation optimization method. The systematic flow provided by the application can be used for treating noise encountered in the running process of the quantum computer to a certain extent, and the effect is superior to the existing methods of fixed congruent insertion and the like.

It should be noted that the random congruent insertion zero noise extrapolation optimization method proposed here adopts the idea of a variable component sub-algorithm, which is a classical-quantum hybrid algorithm. The mixed algorithm divides the calculation task into two parts, the calculation which can be efficiently completed by the classical computer is handed to the classical computer for processing, and the task which can be efficiently processed by the quantum computer is handed to the quantum computer for processing, so that the dependence on quantum resources is reduced, and the complex task can be processed by using the quantum computer to the greatest extent. In general, the preparation and measurement operations of the parameter-containing sub-states are carried out on a quantum computer, while the process of adjusting the parameters and optimizing the measured values is carried out on a classical computer. The parameter to be optimized on classical computer is the probability mean of poisson distributionWhile the loss function that needs to be optimized isTo obtainMeasuring the expected value/>, of operator M at different noise intensities on a quantum computerThereby obtaining/>, by zero noise extrapolation。

As an alternative embodiment, for ease of analysis, it is assumed that the measured operator M is a brix operator and the eigenvalues are located atIs within the interval of (2). This assumption is reasonable because on current quantum computers, the operator to be measured is typically chosen to be the Brix operator. And can always be rescaled by multiplying a constant in front of the measurement operator so that its eigenvalue is atBetween them. The presence of noise always reduces the absolute value of the measurement of M, here exemplified by depolarization noise, where the noise strength isUnder the effect of depolarization noise for the input density matrixThe expected value of the measurement operator M is: /(I); It should be noted that the assumption of depolarization noise is also reasonable here, since random krifords can always be added sequentially to some arbitrary quantum channel, thus converting an arbitrary noise into depolarization noise. This means that it is desirable to make/>, by EM methods such as zero noise extrapolationThe larger the better. It should be noted, however, that since the eigenvalue of M is located atWithin the interval, it is therefore desirable to obtain a/>, closer to 1Therefore, the loss function to be optimized is set to。

To explain why it can be determined by interpolationThe following analysis was performed. LetThe CNOT gate acts on theOn a quantum bit, the CNOT gate can be used with the operator/>, without noiseExpressed and the noise of the CNOT gate can be expressed as depolarization noiseTo describe, i.e. for input statesThrough the noisy CNOT gate, there are: ; wherein/> To generate the probability of depolarization noise, the intensity of depolarization noise may be expressed as "ForIdentity matrix on individual qubits,Is a reduced density matrix. Then consider the fixed congruent insertion method if/>, is usedThe CNOT gate replaces theThe individual CNOT gates will get:

；

When (when) When very small, there is; If consider replacing all Q CNOT gates in the quantum wire and measure the expected value/>, of operator MWill be obtained under first order approximation,; WhereinRepresenting the expected value of the operator M after depolarization noise has occurred at only the ith CNOT gate, andRepresenting the operator measurement without noise. Similarly, consider random congruent insertion, i.e. for theIndividual CNOT gate usageThe CNOT gates are replaced, then theWill become:

；

Here, the Is a random variable. If according to Poisson distributionSelecting random variablesThen in the sense of statistical averaging/>, one can getThe average value of (2) is:

；

This means that noiseless results can be extrapolated by altering the probability of sampling the poisson distribution. That is to say, ; Thereby determining a corresponding linear extrapolation formula.

In addition, for each CNOT gate, it is subjected toProbability sampling of sub-poisson distribution, so that intuitively, considering all CNOT gates requires samplingOnce, the total number of samples increases exponentially with the number of CNOT gates, which is clearly not an efficient method. In practice, thisSubsampling is reduced toAnd twice. Can setIn the optimization interval of the optimization process, the probability of n=0 is far greater than the probability of n=1, and is also far greater than other probabilities. In this way, under the first approximation, for the original quantum circuit, only the situation that each CNOT gate is not replaced and only one CNOT gate is replaced exists, only Q+1 quantum circuits are needed to be measured, and then the result obtained by multiplying the Q+1 quantum circuits by the probability used in sampling can be obtained to be executed with the original requirementThe secondary quantum wires have the same result.

Due to the cut-off value of poisson distribution onlyToward infinity, the relationship holds that,If a larger cut-off value/>, is setAnd a larger sample number R, then when the noise is smaller,AndCan be precisely guaranteed, in which case the optimizationAnd does not bring about a resultant improvement. On an actual quantum computer, noise is not a variable which can be easily controlled, so that in order to obtain improvement on optimization and enable an error mitigation method to be suitable for a wider error rate scene, a large cutoff value/>, is not setAnd the number of samples R, but still useAnd the theoretical result when R tends to infinity, namely a linear extrapolation formula, is used for zero noise extrapolation, and the purpose of the zero noise extrapolation is to make up the noise in an actual system by using the truncation error of the Poisson distribution, so that the optimization can be carried out.

Alternatively, the number of CNOT gates to be replaced is according to the Poisson distributionSampling is performed, but is not required to be performedCut-off valueIs set large and does not need approximationTowards endless results, because sampling errors of limited sampling are required to counteract the effects of quantum noise to some extent. This ensures that the behaviour of the algorithm can be improved in an optimised manner. Meanwhile, in the case, by setting the upper and lower limits of v in the optimization process, the total quantum circuit number of the system is Q+2 when CNOT gate replacement is performed each time, and only the repeated quantum circuitThe times can be obtainedThis also reduces the overhead of quantum resources.

It should be noted that the feasibility of the variable component sub-algorithm as an algorithm running on the current noisy mesoscale quantum computer has been confirmed by a plurality of teams. Meanwhile, error mitigation means are also means for mitigating noise effects commonly used on quantum computers at present, and the feasibility of the error mitigation means is also verified by a plurality of teams. The method provided by the application combines the variable component sub-algorithm and error mitigation, does not relate to operations which cannot be realized on the current quantum computer, and can be realized on the current quantum computer.

For a better understanding of the above-described cancellation of noise, an example of this is now illustrated by an alternative quantum wire diagram, fig. 4 is a schematic diagram of a quantum wire considered for numerical computation according to an embodiment of the present application, wherein,Is a noise channel in front of the CNOT gate, which is represented by a circle with solid dot connections. The noise channels are typically selected as depolarization noise, and the noise parameters are/>, respectively. Consider. In order to better compare the random congruent insertion zero noise extrapolation optimization method provided by the application, theThe values are 0.05,0.06,0.07,0.08,0.09,0.1,0.2 and 0.3,0.4,0.5 respectively, and the measuring arithmetic is binary Paulownia arithmetic。

Further, FIG. 5 is a graph of noise intensity at different levels in a quantum circuit by two methods according to an embodiment of the present applicationSchematic of (2); in FIG. 5, the two methods are compared for/>, given at different noise intensitiesIn the two different quantum circuits,Is-1. Wherein ZEN represents the random congruent insertion zero noise extrapolation optimization method proposed by the present application, and VZEN represents the original fixed congruent insertion method. It can be seen that the proposed method can tolerate more noise, gives better results than fixed congruent insertion at different noise intensities, and even reproduces theoretical values without noise at certain noise intensities.

It should be noted that, the method for attenuating noise on a quantum computer by using a truncated error with a limited sampling number can obtain better results by using the method provided by the application under the same noise intensity compared with the original fixed congruent insertion method.

In summary, by the above embodiment, the proposed zero-noise extrapolation method based on random congruent insertion of the variation can effectively reduce the calculation cost and reduce the resource overhead of the quantum computer, aiming at the problem of error mitigation on the quantum computer. Specifically, the idea of a variable component sub-algorithm is combined, probability required by random congruent insertion is used as physical quantity to be optimized, the inserted congruent arithmetic is truncated, the truncated error in the process is used for counteracting the influence of part of noise, and the probability is optimized through a classical computer, so that a better result after error relief is obtained. That is, the above embodiment combines the variable component sub-algorithm and the random congruent insertion method, and solves the problem in error mitigation on the quantum computer by using the idea of the variable component sub-algorithm. The noise is further mitigated by a combination of optimizing the probability to which random congruent insertion is subject and employing a limited truncation error to cancel out some of the effects of the noise. In summary, the above-mentioned alternative embodiments, in view of many shortcomings of the error mitigation method on the current quantum computer, propose the zero-noise extrapolation method of random congruent insertion based on the variation, through optimizing the probability of random congruent insertion, and counteract some effects of quantum noise with limited truncation error when random congruent insertion, thus get better results. The application has wide engineering development and scientific research application prospect.

From the description of the above embodiments, it will be clear to a person skilled in the art that the method according to the above embodiments may be implemented by means of software plus the necessary general hardware platform, but of course also by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to perform the method according to the embodiments of the present application.

The embodiment also provides a device for alleviating quantum noise, which is used for implementing the above embodiment and the preferred implementation, and is not described in detail. As used below, the term "module" may be a combination of software and/or hardware that implements a predetermined function. While the means described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and contemplated.

Fig. 6 is a block diagram of a quantum noise mitigating apparatus according to an embodiment of the present application, as shown in fig. 6, the apparatus including:

The obtaining module 62 is configured to obtain, by using a preset measurement operator, a first measurement value of a quantum device running a quantum circuit in a noise state, and obtain, by using the preset measurement operator, a plurality of second measurement values of a target quantum circuit corresponding to the quantum device running the quantum circuit in different noise intensities, where the target quantum circuit is a circuit generated by equivalently replacing a two-bit gate on the quantum circuit;

And an extrapolation module 64, configured to perform zero-noise extrapolation according to the probability average values corresponding to the first measurement values and the second measurement values, so as to obtain a target expected value, and determine a noise mitigation result corresponding to the quantum device according to the target expected value.

By the device, the corresponding loss function is set in the computing equipment, the quantum equipment and the computing equipment are used for executing the operation of optimizing the specific loss function to weaken the influence caused by noise, namely, a variable component sub-algorithm is combined with a random congruent insertion zero-noise extrapolation method, and the advantages of the computing equipment and the quantum equipment are fully exerted by designing the loss function to be optimized and the corresponding optimization flow, so that the noise in the running process of the quantum equipment is relieved, and the problems that the accuracy of error relief on the quantum equipment is lower, the resource cost of the quantum equipment is higher, and the effect of noise relief is poor are solved, the influence of noise is weakened in the running process of the quantum equipment, the running effect of higher fidelity is obtained, and the quantum equipment can support the quantum algorithm with higher running accuracy.

In an exemplary embodiment, the above apparatus further includes: the control module is used for sending target data by the computing device connected with the quantum device, wherein the target data at least comprises: a first measurement value, a second measurement value; and under the condition that a loss function is constructed through the first measured value and the second measured value on the computing equipment and the target data is present in the computing equipment, the quantum circuits on the quantum equipment are determined to be replaced by two bit gates with different numbers, a plurality of target quantum circuits are obtained, and the preset measuring operators are instructed to measure the plurality of target quantum circuits.

In an exemplary embodiment, the control module further includes a determining unit, configured to determine a number of samples of a number of two-bit gates to be replaced by two-bit gates on the quantum line; and determining a target quantum circuit according to the sampling number.

In an exemplary embodiment, the control module further includes: the establishing unit is used for acquiring the connection relation between the quantum equipment and the computing equipment; and under the condition that the connection relation is determined to be effective, establishing a data channel of the quantum equipment and the computing equipment, wherein the data channel is used for efficiently transmitting interaction data of the quantum equipment and the computing equipment.

In an exemplary embodiment, the above apparatus further includes: a substitution module for determining a substitution distribution for performing equivalent substitution on the quantum circuit; under the condition that P first types of quantum bit gates existing in the quantum circuit are identified according to the replacement distribution, replacing each first type of quantum bit gate by using 2n+1 second types of two bit gates, and repeating the replacement R times to obtain R replacement results corresponding to each first type of quantum bit gate, wherein P, R and n are positive integers, and the first type of quantum bit gate is a target two-bit gate to be subjected to equivalent replacement in the quantum circuit; and generating a target quantum circuit corresponding to the quantum circuit based on R replacing results corresponding to each of the quantum bit gates of the P first types.

In an exemplary embodiment, the replacing module further includes: a first determining unit, configured to compare a target replacement value with a preset replacement value after each of the first type of quantum bit gates is replaced with 2n+1 second type of two bit gates, where the target replacement value is a value of the number of all two bit gates in the quantum circuit after the replacement operation is performed; determining that the quantum circuit is too deep and prohibiting the quantum circuit from running on the quantum device when the target replacement value is greater than the preset replacement value; and under the condition that the target replacement value is smaller than or equal to the preset replacement value, determining that the quantum circuit meets operation conditions, and allowing the quantum circuit to operate on the quantum device.

In an exemplary embodiment, the replacing module further includes: a second determining unit, configured to determine, before replacing each quantum bit gate of the first type with 2n+1 two-bit gates of the second type, a first error rate corresponding to the two-bit gates of the first type and a second error rate corresponding to the two-bit gates of the 2n+1 second type as a whole; determining the magnitude relation between the first error rate, the second error rate and the maximum error rate in a preset error rate range respectively; and determining whether the replacement is a valid two-bit gate replacement according to the size relationship.

In an exemplary embodiment, the second determining unit is further configured to determine that, when the first error rate is less than or equal to the maximum error rate and the second error rate is less than or equal to the maximum error rate, a 2n+1 second type two-bit gates are currently used for replacing each first type of qubit gate with a valid two-bit gate replacement; in the event that the second error rate is greater than the maximum error rate, it is determined that a replacement is currently performed for each of the first type of qubit gates using 2n+1 second type of two-bit gates with an invalid two-bit gate replacement.

In an exemplary embodiment, the replacing module further includes: a third determining unit configured to determine, before replacement of each of the first type of qubit gates with 2n+1 second type of two bit gates, a number of replacements that the quantum device allows for performing the replacement in parallel; grouping the target two-bit gates of the first type according to the replacement number to obtain a plurality of parallel replacement groups; and simultaneously replacing the first type of target qubit gates in the plurality of parallel replacement groups on the quantum device.

In an exemplary embodiment, the replacing module further includes: a fourth determining unit, configured to determine, before the replacing distribution for performing equivalent replacement on the quantum wire, the method further includes: determining parameters to be optimized, which are input by a target object on the computing device, wherein the parameters to be optimized have the following relation with a loss function: Wherein said/> For a target expected value corresponding to a parameter to be optimized, f is a loss function, the target expected value is an amount which has an exact physical meaning, is related to a noiseless state in the quantum device and is easy to process by the computing device.

In an exemplary embodiment, the replacing module is further configured to obtain a probability distribution corresponding to the parameter to be optimized; and under the condition that the number of substitutions corresponding to the probability distribution is smaller than the preset allowable number of substitutions, sampling the number of substitutions of each two-bit gate on the quantum circuit for R times according to the probability distribution to obtain first sub-distribution, and under the condition that Q two-bit gates coexist on the quantum circuit, determining the substitution distribution for carrying out equivalent substitution on the quantum circuit based on the Q first sub-distribution, wherein R, Q is a positive integer.

In an exemplary embodiment, the obtaining module is further configured to, in a case where it is determined that a plurality of quantum wires are simultaneously operated and measured in the quantum device, determine a preset number of quantum wires that the quantum device is allowed to operate and measure in parallel; controlling the preset measuring operators to measure the quantum circuits of the quantum equipment running in the noise state to obtain a plurality of first sub-measured values and/or second sub-measured values; summarizing the plurality of first sub-measured values to obtain the first measured value, and/or summarizing the plurality of second sub-measured values to obtain the second measured value.

In an exemplary embodiment, the above apparatus further includes: the instruction module is used for carrying out zero noise extrapolation according to probability average values corresponding to the first measurement value and the plurality of second measurement values to obtain a target expected value, so that after a noise alleviation result corresponding to the quantum equipment is determined according to the target expected value, the computing equipment is instructed to carry out convergence confirmation on a loss function under the condition that the target expected value exists on the computing equipment; and under the condition that a confirmation instruction sent by the computing device is received, determining whether noise adjustment of the quantum device is completed or not according to the confirmation instruction.

In an exemplary embodiment, the instruction module is further configured to determine that noise adjustment of the quantum device has been completed if the confirmation instruction indicates that the current target expected value meets a convergence requirement of a loss function; and under the condition that the confirmation instruction indicates that the current target expected value does not meet the convergence requirement of the loss function, determining that the noise adjustment of the quantum device is not completed, and indicating the computing device to continuously perform iterative optimization on the loss function.

In an exemplary embodiment, the control module further includes an parsing unit, configured to parse, after receiving target data sent by a computing device connected to the quantum device, the target data through configuration in the quantum device; and determining gradient information of the loss function to be optimized on the quantum equipment according to the analysis result.

In an exemplary embodiment, the above apparatus further includes: the storage module is used for carrying out zero noise extrapolation according to probability average values corresponding to the first measured value and the plurality of second measured values, and obtaining an optimal target expected value for converging a loss function after obtaining a target expected value; acquiring key information generated in the process of determining the optimal target expected value, wherein the key information at least comprises one of the following: gradient change information, probability change information, target expected value change information; storing the critical information in a memory preconfigured for the computing device and the quantum device.

In an exemplary embodiment, the above apparatus further includes: the output module is used for determining an output structure corresponding to the quantum equipment; and outputting a measurement result of a preset measurement operator after noise alleviation according to the output structure.

In an exemplary embodiment, the above apparatus further includes: the determining module is configured to, before performing zero noise extrapolation according to the probability average values corresponding to the first measurement value and the second measurement values, further include: determining a linear extrapolation formula for performing the zero noise extrapolation; wherein the linear extrapolation formula isSaidFor measuring the measurement value of a measuring operator under different noise states on a quantum computer,For the first measurement,Is the probability mean of the second measurement value,The parameters to be optimized carry probability characteristics.

In an exemplary embodiment, the above apparatus further includes: the difference module is used for acquiring a new target expected value of the quantum equipment after optimization under the condition that the target expected value corresponding to the quantum equipment is determined and the target expected value is continuously optimized; and comparing the difference between the target expected value and the new target expected value to determine whether the optimization of the target expected value is effective according to the difference.

In the present embodiment, the term "module" or "unit" refers to a computer program or a part of a computer program having a predetermined function and working together with other relevant parts to achieve a predetermined object, and may be implemented in whole or in part by using software, hardware (such as a processing circuit or a memory), or a combination thereof. Also, a processor (or multiple processors or memories) may be used to implement one or more modules or units. Furthermore, each module or unit may be part of an overall module or unit that incorporates the functionality of the module or unit.

It should be noted that, for simplicity of description, the foregoing method embodiments are all described as a series of acts, but it should be understood by those skilled in the art that the present application is not limited by the order of acts described, as some steps may be performed in other orders or concurrently in accordance with the present application. Further, those skilled in the art will also appreciate that the embodiments described in the specification are all preferred embodiments, and that the acts and modules referred to are not necessarily required for the present application.

Embodiments of the present application also provide a computer readable storage medium having a computer program stored therein, wherein the computer program is arranged to perform the steps of any of the method embodiments described above when run.

In one exemplary embodiment, the computer readable storage medium may include, but is not limited to: a usb disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory RAM), a removable hard disk, a magnetic disk, or an optical disk, or other various media capable of storing a computer program.

An embodiment of the application also provides an electronic device comprising a memory having stored therein a computer program and a processor arranged to run the computer program to perform the steps of any of the method embodiments described above.

In an exemplary embodiment, the electronic device may further include a transmission device connected to the processor, and an input/output device connected to the processor.

Embodiments of the application also provide a computer program product comprising a computer program which, when executed by a processor, implements the steps of any of the method embodiments described above.

Embodiments of the present application also provide another computer program product comprising a non-volatile computer readable storage medium storing a computer program which, when executed by a processor, implements the steps of any of the method embodiments described above.

Embodiments of the present application also provide a computer program comprising computer instructions stored on a computer-readable storage medium; the processor of the computer device reads the computer instructions from the computer readable storage medium and the embedder executes the computer instructions to cause the computer device to perform the steps of any of the method embodiments described above.

Specific examples in this embodiment may refer to the examples described in the foregoing embodiments and the exemplary implementation, and this embodiment is not described herein.

It will be appreciated by those skilled in the art that the modules or steps of the application described above may be implemented in a general purpose computing device, they may be concentrated on a single computing device, or distributed across a network of computing devices, they may be implemented in program code executable by computing devices, so that they may be stored in a storage device for execution by computing devices, and in some cases, the steps shown or described may be performed in a different order than that shown or described herein, or they may be separately fabricated into individual integrated circuit modules, or multiple modules or steps of them may be fabricated into a single integrated circuit module. Thus, the present application is not limited to any specific combination of hardware and software.

The above description is only of the preferred embodiments of the present application and is not intended to limit the present application, but various modifications and variations can be made to the present application by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the principle of the present application should be included in the protection scope of the present application.

Claims

1. A method of mitigating quantum noise, comprising:

Acquiring a first measured value of a quantum circuit of a quantum device operated in a noise state through a preset measurement operator, and acquiring a plurality of second measured values of a target quantum circuit corresponding to the quantum circuit operated in different noise intensities through the preset measurement operator under the condition of carrying out multiple probability adjustment, wherein the target quantum circuit is generated after carrying out equivalent replacement on two bit gates on the quantum circuit;

zero noise extrapolation is carried out according to probability average values corresponding to the first measured value and the plurality of second measured values, a target expected value is obtained, and a noise alleviation result corresponding to the quantum equipment is determined according to the target expected value;

the method further comprises the steps of:

Determining a replacement distribution for performing equivalent replacement on the quantum circuit;

under the condition that P first types of quantum bit gates existing in the quantum circuit are identified according to the replacement distribution, replacing each first type of quantum bit gate by using 2n+1 second types of two bit gates, and repeating the replacement R times to obtain R replacement results corresponding to each first type of quantum bit gate, wherein P, R and n are positive integers, and the first type of quantum bit gate is a target two-bit gate to be subjected to equivalent replacement in the quantum circuit;

And generating a target quantum circuit corresponding to the quantum circuit based on R replacing results corresponding to each of the quantum bit gates of the P first types.

2. The method of claim 1, wherein prior to zero-noise extrapolation from the probability average for the first measurement and the second plurality of measurements, the method further comprises:

Transmitting target data to a computing device connected to the quantum device, wherein the target data comprises at least: a first measurement value, a second measurement value;

And under the condition that a loss function is constructed through the first measured value and the second measured value on the computing equipment and the target data is present in the computing equipment, the quantum circuits on the quantum equipment are determined to be replaced by two bit gates with different numbers, a plurality of target quantum circuits are obtained, and the preset measuring operators are instructed to measure the plurality of target quantum circuits.

3. The method of claim 2, wherein determining to perform a different number of two-bit gate substitutions on the quantum wire on the quantum device comprises:

determining the sampling number of the number of two-bit gates to be subjected to two-bit gate replacement on the quantum circuit;

and determining a target quantum circuit according to the sampling number.

4. The method of quantum noise mitigation according to claim 2, wherein the method further comprises:

acquiring a connection relationship between the quantum device and the computing device;

And under the condition that the connection relation is determined to be effective, establishing a data channel of the quantum equipment and the computing equipment, wherein the data channel is used for efficiently transmitting interaction data of the quantum equipment and the computing equipment.

5. The method of claim 1, wherein after replacing each of the first type of qubit gates with 2n+1 second type of two bit gates, the method further comprises:

Comparing a target replacement value with a preset replacement value, wherein the target replacement value is a quantity value of all two-bit gates in a quantum circuit after the replacement operation is completed;

Determining that the quantum circuit is too deep and prohibiting the quantum circuit from running on the quantum device when the target replacement value is greater than the preset replacement value;

And under the condition that the target replacement value is smaller than or equal to the preset replacement value, determining that the quantum circuit meets operation conditions, and allowing the quantum circuit to operate on the quantum device.

6. The method of claim 1, wherein prior to replacing each of the first type of qubit gates with 2n+1 second type of two bit gates, the method further comprises:

Determining a first error rate corresponding to the two-bit gates of the first type and a second error rate corresponding to the whole of the two-bit gates of the 2n+1 second types;

Determining the magnitude relation between the first error rate, the second error rate and the maximum error rate in a preset error rate range respectively;

and determining whether the replacement is a valid two-bit gate replacement according to the size relationship.

7. The method of claim 6, wherein determining whether the substitution is a valid two-bit gate substitution based on the magnitude relation comprises:

Determining that a 2n+1 second type two-bit gate replacement is currently used for each first type of qubit gate with a valid two-bit gate replacement if the first error rate is less than or equal to the maximum error rate and the second error rate is less than or equal to the maximum error rate;

In the event that the second error rate is greater than the maximum error rate, it is determined that a replacement is currently performed for each of the first type of qubit gates using 2n+1 second type of two-bit gates with an invalid two-bit gate replacement.

8. The method of claim 1, wherein prior to replacing each of the first type of qubit gates with 2n+1 second type of two bit gates, the method further comprises:

determining a number of substitutions the quantum device allows for performing the substitution in parallel;

grouping the target two-bit gates of the first type according to the replacement number to obtain a plurality of parallel replacement groups;

and simultaneously replacing the first type of target qubit gates in the plurality of parallel replacement groups on the quantum device.

9. The method of claim 1, wherein prior to determining the substitution distribution for equivalent substitutions on the quantum wires, the method further comprises:

and determining a parameter to be optimized, which is input by a target object on the computing device, wherein the parameter to be optimized has the following relation with a loss function, wherein the target expected value corresponding to the parameter to be optimized is f, the target expected value is a loss function, and the target expected value is an amount which has exact physical meaning, is related to a noiseless state in the quantum device and is easy to process by the computing device.

10. The method of claim 9, wherein determining a substitution distribution for equivalent substitutions on the quantum wires comprises:

acquiring probability distribution corresponding to the parameter to be optimized;

And under the condition that the number of substitutions corresponding to the probability distribution is smaller than the preset allowable number of substitutions, sampling the number of substitutions of each two-bit gate on the quantum circuit for R times according to the probability distribution to obtain first sub-distribution, and under the condition that Q two-bit gates coexist on the quantum circuit, determining the substitution distribution for carrying out equivalent substitution on the quantum circuit based on the Q first sub-distribution, wherein R, Q is a positive integer.

11. The method of claim 1, wherein the first measurement of the quantum device operating the quantum circuit in the noisy state is obtained by a preset measurement operator, the method further comprising:

In the case of determining that a plurality of quantum wires are simultaneously operated and measured in the quantum device, determining a preset number of quantum wires that the quantum device is allowed to be operated and measured in parallel;

Controlling the preset measuring operators to measure the quantum circuits of the quantum equipment running in the noise state to obtain a plurality of first sub-measured values and/or second sub-measured values;

summarizing the plurality of first sub-measured values to obtain the first measured value, and/or summarizing the plurality of second sub-measured values to obtain the second measured value.

12. The method of claim 1, wherein after zero-noise extrapolation is performed according to a probability average value corresponding to the first measurement value and the plurality of second measurement values to obtain a target expected value, so as to determine a noise mitigation result corresponding to the quantum device according to the target expected value, the method further comprises:

instructing the computing device to confirm convergence of a loss function if the target expected value exists on the computing device;

and under the condition that a confirmation instruction sent by the computing device is received, determining whether noise adjustment of the quantum device is completed or not according to the confirmation instruction.

13. The method of quantum noise mitigation according to claim 12, wherein the method further comprises:

Determining that noise adjustment of the quantum device is completed under the condition that the confirmation instruction indicates that the current target expected value meets the convergence requirement of the loss function;

And under the condition that the confirmation instruction indicates that the current target expected value does not meet the convergence requirement of the loss function, determining that the noise adjustment of the quantum device is not completed, and indicating the computing device to continuously perform iterative optimization on the loss function.

14. The method of claim 2, wherein after receiving target data sent by a computing device connected to the quantum device, the method further comprises:

analyzing the target data through a computing device configured in the quantum device;

And determining gradient information of the loss function to be optimized on the quantum equipment according to the analysis result.

15. The method of claim 1, wherein after zero-noise extrapolation is performed according to a probability average value corresponding to the first measurement value and the plurality of second measurement values to obtain a target expected value, the method further comprises:

obtaining an optimal target expected value of the computing equipment and the quantum equipment for converging a loss function;

Acquiring key information generated in the process of determining the optimal target expected value, wherein the key information at least comprises one of the following: gradient change information, probability change information, target expected value change information;

storing the critical information in a memory preconfigured for the computing device and the quantum device.

16. The method of quantum noise mitigation according to claim 1, wherein the method further comprises:

determining an output structure corresponding to the quantum equipment;

And outputting a measurement result of a preset measurement operator after noise alleviation according to the output structure.

17. The method of claim 1, wherein prior to zero-noise extrapolation from the probability average for the first measurement and the second plurality of measurements, the method further comprises:

Determining a linear extrapolation formula for performing the zero noise extrapolation;

wherein the linear extrapolation formula is ，For measuring the measurement value of a measuring operator under different noise states on a quantum computer,For the first measurement,Is the probability mean of the second measurement value,The parameters to be optimized carry probability characteristics.

18. The method of quantum noise mitigation according to claim 1, wherein the method further comprises:

under the condition that a target expected value corresponding to the quantum equipment is determined and the target expected value is continuously optimized, acquiring a new target expected value of the quantum equipment after the quantum equipment is optimized;

and comparing the difference between the target expected value and the new target expected value to determine whether the optimization of the target expected value is effective according to the difference.

19. A quantum noise mitigation device, comprising:

The device comprises an acquisition module, a detection module and a control module, wherein the acquisition module is used for acquiring a first measured value of a quantum circuit operated by quantum equipment in a noise state through a preset measurement operator, and acquiring a plurality of second measured values of a target quantum circuit corresponding to the quantum circuit operated by the quantum equipment in different noise intensities through the preset measurement operator under the condition of carrying out multiple probability adjustment, wherein the target quantum circuit is a circuit generated after equivalent replacement of a two-bit gate on the quantum circuit;

the extrapolation module is used for carrying out zero noise extrapolation according to probability average values corresponding to the first measured value and the plurality of second measured values to obtain a target expected value so as to determine a noise alleviation result corresponding to the quantum equipment according to the target expected value;

the apparatus further comprises: a substitution module for determining a substitution distribution for performing equivalent substitution on the quantum circuit;

20. A computer readable storage medium, characterized in that a computer program is stored in the computer readable storage medium, wherein the computer program, when being executed by a processor, implements the steps of the method according to any of the claims 1 to 18.

21. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the method of any one of claims 1 to 18 when the computer program is executed.

22. A computer program product comprising a computer program, characterized in that the computer program, when being executed by a processor, implements the steps of the method as claimed in any one of claims 1 to 18.