Quantum data generation method and system, storage medium, computer device and terminal

Abstract

The application belongs to the technical field of quantum computing and quantum artificial intelligence, and discloses a quantum data generation method, system, storage medium, computer device and terminal. The designed quantum U-shaped network is trained by using existing limited quantum data. The quantum U-shaped network is capable of converting a random noise quantum data into quantum data of the same type as the training data, that is, generating a brand new quantum data of the same type. The number of convolution network layers and pooling layers in the quantum U-shaped network structure can be automatically adjusted according to the size of specific quantum data and the success rate of generated data. The quantum gate used in the convolution network layer of the quantum U-shaped network structure is set to other types of quantum gates such as three-bit gates according to actual conditions. Any quantum data can be generated. The application can efficiently generate any required type of quantum data, and the number of generated data is not limited.

Description

Technical Field

[0001] This invention belongs to the field of quantum computing and quantum artificial intelligence technology, and in particular relates to a quantum data generation method, system, storage medium, computer equipment and terminal. Background Technology

[0002] Currently, quantum computing, as a technology that deeply integrates quantum physics and computer science, is emerging as a promising new frontier field with broad application prospects. Quantum systems have demonstrated information processing capabilities far exceeding those of existing digital information processing systems and are considered a key component of the next generation of information technology. With the rapid development of materials science, hardware manufacturing, and disciplines such as error correction and compilation, large-scale, universal, and fault-tolerant quantum computing is gradually becoming possible.

[0003] The physical realization of quantum systems is subject to noise and decoherence, meaning that the coherence of a quantum system gradually disappears over time, eventually degenerating into a classical system. Currently operational quantum devices are of medium scale (50-100 qubits), contain noise, and are expensive. Acquiring quantum data is also a costly task. Quantum computers are progressing towards large-scale and practical applications, and quantum machine learning is at the forefront of quantum computing. The generation of large-scale quantum data will be one of the most important and challenging aspects of quantum machine learning, quantum data processing, and quantum complexity simulation.

[0004] Based on the above analysis, the problems and shortcomings of the existing technology are as follows:

[0005] 1. Scale limitation: Currently operational quantum devices are relatively limited in size, mostly ranging from 50 to 100 qubits. This limits the ability of quantum computers to solve larger-scale problems.

[0006] 2. Noise Issues: Existing quantum devices are affected by noise, leading to decoherence in quantum systems. This impacts the performance and accuracy of quantum algorithms and may make it difficult to realize quantum advantage in some applications.

[0007] 3. High cost: The manufacturing, maintenance and operation costs of quantum computers are relatively high, which may limit their popularization and application in commercial and scientific research fields.

[0008] 4. Data Acquisition Challenges: Acquiring quantum data is a costly and challenging task. This may slow down the development of fields such as quantum machine learning, quantum data processing, and quantum complex simulation.

[0009] 5. Error correction and fault tolerance: Existing quantum computers have not yet achieved full fault tolerance, which means that errors in the quantum computing process may be difficult to correct effectively, affecting the accuracy of the calculation results.

[0010] 6. Maturity of Algorithms and Applications: Although much progress has been made in the theoretical research of quantum computing, the quantum algorithms and applications in reality still need further development and improvement in order to better leverage the advantages of quantum computing in practical problems.

[0011] In summary, existing quantum technologies still face certain challenges and limitations in terms of scale, noise control, cost, data acquisition, and application maturity. To overcome these challenges, researchers need to conduct in-depth research on quantum computing hardware, quantum algorithms, and quantum applications to continuously advance the development of quantum computing technology. Summary of the Invention

[0012] To address the problems existing in the prior art, this invention provides a quantum data generation method, system, storage medium, computer device, and terminal.

[0013] The present invention is implemented as follows: a quantum data generation method, wherein the quantum data generation method trains a designed quantum U-shaped network using existing finite quantum data; thereby enabling the quantum U-shaped network to transform a random noise quantum data into quantum data of the same type as the training data, that is, to generate a completely new quantum data of the same type.

[0014] Furthermore, the quantum data generation method employs a denoising diffusion generation model during the training process; the quantum U-shaped network consists of two bridge-connected multi-scale entangled renormalization networks; and the quantum U-shaped network adopts a U-shaped network structure Q-Unet composed of downsampling-upsampling full quantum convolutional networks.

[0015] Furthermore, the quantum data generation method specifically includes the following steps:

[0016] Step 1: Collect N types of quantum data that need to be generated, and construct various quantum circuits to collect them on the existing NISQ quantum computer;

[0017] Step two: Construct a quantum U-shaped network. The quantum U-shaped network includes downsampling and upsampling parts. The downsampling part is implemented by n-fold quantum convolutional layers and pooling layers, while the upsampling part is implemented by n-fold quantum convolutional layers. Each quantum convolutional layer consists of a two-qubit quantum gate, and the pooling layer is implemented by a single-qubit quantum gate and a controlled NOT gate to compress the entangled information of the two qubits into a single qubit. n is a positive integer, determined according to the size of the generated quantum data.

[0018] Step 3: Gaussian noise is gradually added to the N quantum data generated in Step 1 to make them noisy data; this process is achieved by applying a random rotation operation Ry(θ) to each bit of the quantum data, where θ follows a normal distribution.

[0019] Step four: Using the noisy data and corresponding noise-free data pairs generated in step three, train the diffusion model of the quantum U-shaped network built in step two to make it converge.

[0020] Step five: Generate a fully acoustic quantum data set and repeatedly input it into the quantum U-shaped network trained in step four. This will generate a new set of quantum data of the same type as the quantum data collected in step one. Repeating this step multiple times will generate multiple new sets of quantum data.

[0021] Furthermore, the 3-bit GHZ type state of the quantum data generation method is as follows:

[0022]

[0023]

[0024] The 8 bits of information are encoded onto a classical graph. An 8-bit quantum state is encoded onto a 16×16 graph. The pixel with graph coordinates (0, 0) corresponds to the quantum computing basis |00000000>, and the pixel with graph coordinates (16, 16) corresponds to the quantum computing basis |11111111>. Each pixel on the graph corresponds to the computing basis of this 8-bit quantum state, and the height of the pillar on each pixel corresponds to the probability of the quantum state in that computing basis.

[0025] Furthermore, in the quantum data generation method, the success rate of quantum state generation is measured by the sum of the probabilities of the |000> and |111> components of the quantum state in the data generated in the quantum U-shaped network; whether the generated quantum state belongs to the same class as the GHZ-type quantum states in the training network is determined by detecting the noise level P of the generated data. noise Measure; if P noise If the threshold is less than ε, the generated data belongs to the GHZ class. ε is a set threshold that can be set according to the usage scenario; noise level P noise The definition of P is as follows: noise =1-P0-P 255 Here (P0, P) 255 ) represent the probabilities that the generated quantum state is in the computation basis (|00000000>, |11111111>).

[0026] Another object of the present invention is to provide a computer device including a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the quantum data generation method.

[0027] Another object of the present invention is to provide a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the quantum data generation method.

[0028] Another objective of this invention is to provide an information data processing terminal for implementing the quantum data generation method.

[0029] Another object of the present invention is to provide a quantum data generation system based on the aforementioned quantum data generation method, the quantum data generation system comprising:

[0030] The data training module is used to train the designed quantum U-shaped network using existing limited quantum data;

[0031] The data generation module enables the quantum U-shaped network to transform random noise quantum data into quantum data of the same type as the training data, that is, to generate a completely new quantum data of the same kind.

[0032] Another object of the present invention is to provide a quantum device comprising the aforementioned quantum data generation system.

[0033] Based on the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solution to be protected by this invention are as follows:

[0034] First, the number of convolutional network layers and pooling layers in the quantum U-shaped network structure of this invention can be automatically adjusted according to the size of the specific quantum data and the success rate of data generation; the quantum gates used in the convolutional network layers of the quantum U-shaped network structure are not limited to the two-qubit gates and single-qubit gates shown in the example, and other types of quantum gates, such as three-qubit gates, can be set according to actual conditions; arbitrary quantum data can be generated, such as the first type ( Figure 5 ) and the second GHZ type ( Figure 6 The success rates were 93.77% and 90.72% respectively; P noise The accuracy of the generated quantum data can be determined by setting ε according to the usage scenario.

[0035] Second, considering the technical solution as a whole or from the perspective of the product, the technical effects and advantages of the technical solution to be protected by this invention are specifically described as follows: This invention can efficiently generate any type of quantum data, and the amount of data generated is unlimited.

[0036] Third, as supplementary evidence of the inventive step of the claims of this invention, it is also reflected in the following important aspects:

[0037] (1) The expected benefits and commercial value of the technical solution of this invention after transformation are as follows:

[0038] This invention is versatile and practical, applicable to the preparation of quantum data on arbitrary quantum processors in the NIST era, including quantum superconducting chips, optical chips, and ion traps. This invention addresses an important aspect of quantum information processing—quantum state preparation—and can be incorporated into existing quantum information processing and quantum computing software packages to enhance its practicality.

[0039] (2) The technical solution of this invention fills a technical gap in the industry both domestically and internationally:

[0040] Existing quantum data generation / preparation schemes all involve designing corresponding quantum circuits for different quantum states. The quantum data generation scheme proposed in this invention has universality and is applicable to generating and preparing arbitrary quantum states. Attached Figure Description

[0041] Figure 1 This is a flowchart of the quantum data generation method provided in an embodiment of the present invention;

[0042] Figure 2 This is a schematic diagram of the process for generating 8-qubit data provided in an embodiment of the present invention;

[0043] Figure 3 This is a schematic diagram of the 8-qubit Q-Unet quantum circuit structure provided in an embodiment of the present invention: the left side is the downsampling part, and the right side is the upsampling part;

[0044] Figure 4 This is a schematic diagram of MERA corresponding to the quantum DDPM network Q-Unet provided in this embodiment of the invention; the square is the disentangler, the triangle is the isometry, and the endpoints with the same number in the circles on the left and right sides are the data communication diagrams in the Q-Unet upsampling and downsampling processes;

[0045] Figure 5 This is an illustration of the quantum data generation process of the quantum U-shaped network provided in this embodiment of the invention: (a) the initial noisy quantum state, (b) the quantum state generated in the middle of the process, and (c) the first type of GHZ state generated at the end; the xy axis of the figure encodes the ground state of the 8-bit quantum state, and the z axis encodes the probability of the corresponding quantum state on the computation basis; the bars in the figure are set to distinguish the components of each ground state and have no specific meaning.

[0046] Figure 6This is a schematic diagram showing the success rate of generating the first type of GHZ quantum data using the quantum U-shaped network provided in this embodiment of the invention; the horizontal axis represents the 100 quantum states generated by this invention, and the vertical axis represents the sum of the probabilities of the corresponding quantum states being in the computational basis (|00000000>, |11111111>), with an average success rate of 93.77%.

[0047] Figure 7 This is a schematic diagram showing the success rate of generating the second type of GHZ quantum data using the quantum U-shaped network provided in this embodiment of the invention. The horizontal axis represents the 100 quantum states generated by this invention, and the vertical axis represents the sum of the probabilities of the corresponding quantum states being in the computational basis (|00000000>, |11111111>). The average success rate is 90.72%. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0049] like Figure 1 As shown, the quantum data generation method provided in this embodiment of the invention includes the following steps:

[0050] S101: Train the designed quantum U-shaped network using existing limited quantum data;

[0051] S102: Enables the quantum U-shaped network to transform random noise quantum data into quantum data of the same type as the training data, that is, to generate a completely new quantum data of the same type.

[0052] This invention trains a designed quantum U-shaped network using existing, limited quantum data, enabling the network to transform random, noisy quantum data into quantum data of the same type as the training data—that is, generating entirely new quantum data of the same kind. The training process employs a denoising diffusion generation model (DDPM). The quantum U-shaped network consists of two bridged multi-scale entanglement renormalization networks (MERA). Inspired by the classical U-shaped network, the quantum U-shaped network adopts a U-shaped network structure, Q-Unet, composed of downsampling-upsampling fully quantum convolutional networks, as shown in the diagram. Figure 2 As shown.

[0053] The quantum data generation method provided in this embodiment of the invention specifically includes the following steps:

[0054] Step 1: Collect N types of quantum data that need to be generated. Various quantum circuits can be constructed on the existing NISQ quantum computer to collect the data.

[0055] Step two: Construct a quantum U-shaped network. The quantum U-shaped network consists of downsampling and upsampling parts. The downsampling part is implemented using n-fold quantum convolutional layers and pooling layers, while the upsampling part is implemented using n-fold quantum convolutional layers. Each quantum convolutional layer consists of a two-qubit quantum gate, and the pooling layer is implemented using a single-qubit quantum gate and a controlled NOT gate to compress the entangled information of the two qubits into a single qubit. Here, n is a positive integer, determined based on the size of the generated quantum data.

[0056] Step 3: Gradually add Gaussian noise to the N quantum data generated in Step 1 until they become noisy data. This step is achieved by applying a random rotation operation Ry(θ) to each bit of the quantum data, where θ follows a normal distribution at each step.

[0057] Step four: Using the fully acoustic data and corresponding noise-free data pairs generated in step three, train the diffusion model of the quantum U-shaped network built in step two to make it converge.

[0058] Step five: Generate a fully acoustic quantum data set and repeatedly input it into the quantum U-shaped network trained in step four. This will generate a new set of quantum data of the same type as the quantum data collected in step one. Repeating this step multiple times will generate multiple new sets of quantum data.

[0059] Example 1:

[0060] This invention illustrates its feasibility and effectiveness using the generation of a 3-bit GHZ class as an example. Here, the 3-bit GHZ class is:

[0061]

[0062]

[0063] Figure 5 This diagram illustrates the process of a GHZ1 state being generated from a random noise quantum state through a trained quantum U-shaped network. For data visualization, this invention encodes 8 bits of information onto a classical graph. Specifically, an 8-bit quantum state is encoded onto a 16×16 graph. The pixel at coordinates (0, 0) corresponds to the quantum computing basis |00000000>, and the pixel at coordinates (16, 16) corresponds to the quantum computing basis |11111111>. Each pixel on the graph corresponds to the computing basis of this 8-bit quantum state, and the height of the bars on each pixel corresponds to the probability of the quantum state in that computing basis, such as... Figure 5 As shown in the figure, the |000> component of the quantum state is retained throughout the training process, while the |111> component gradually increases.

[0064] In this embodiment of the invention, the success rate of quantum state generation is measured by the sum of the probabilities of the |000> and |111> components of the quantum state in the data generated by the quantum U-shaped network. A higher success rate indicates that the generated data is closer to the desired data type. Whether the generated quantum state belongs to the same class as the GHZ-type quantum states in the training network is determined by detecting the noise level P of the generated data. noise To measure. Specifically, if P noise If the threshold is less than ε, the generated data belongs to the GHZ class. Here, ε is a set threshold that can be adjusted according to the usage scenario. Noise level P noise The definition of P is as follows: noise =1-P0-P 255 Here (P0, P) 255 ) represent the probabilities that the generated quantum state lies within the computational basis (|00000000>, |11111111>). Here, the average success rate of the quantum GHZ1 state reaches 93.77%, as shown below. Figure 6 As shown.

[0065] The quantum data generation system provided in this embodiment of the invention includes:

[0066] The data training module is used to train the designed quantum U-shaped network using existing limited quantum data;

[0067] The data generation module enables the quantum U-shaped network to transform random noise quantum data into quantum data of the same type as the training data, that is, to generate a completely new quantum data of the same kind.

[0068] It should be noted that embodiments of the present invention can be implemented in hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by a suitable instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the above-described devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuitry such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field-programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of the above-described hardware circuitry and software, such as firmware.

[0069] The effect of this example is... Figure 5-7The diagram illustrates the quantum state generation process and its accuracy. Currently, there are no comparable technologies available.

[0070] The following are four specific examples illustrating how to apply the above quantum data generation method:

[0071] Example 1: Generating quantum state data

[0072] In this embodiment, the present invention aims to generate a set of quantum state data using a quantum data generation method. To this end, the present invention first collects data on N quantum states and constructs the corresponding quantum circuits on an existing NISQ quantum computer. Then, the present invention builds a quantum U-shaped network and sets the network parameters according to the size of the quantum state data. Next, the present invention converts the original quantum data into fully acoustic data by adding Gaussian noise, and uses this data to train the quantum U-shaped network. Finally, the present invention generates a fully acoustic quantum data set and inputs it into the trained quantum U-shaped network to generate a new quantum state data set.

[0073] Example 2: Generating Quantum Computing Simulation Data

[0074] In this embodiment, the present invention aims to generate data related to a specific quantum computing simulation. First, the present invention collects data from N quantum computing simulations and constructs corresponding quantum circuits. Then, the present invention builds a quantum U-shaped network and sets the network parameters according to the size of the quantum simulation data. Next, the present invention converts the original quantum data into noisy data by adding Gaussian noise, and uses this data to train the quantum U-shaped network. Finally, the present invention generates noisy quantum data and inputs it into the trained quantum U-shaped network to generate new quantum computing simulation data.

[0075] Example 3: Generating Quantum Communication Data

[0076] In this embodiment, the present invention aims to generate data related to quantum communication. First, the present invention collects data from N quantum communication events and constructs corresponding quantum circuits. Then, the present invention builds a quantum U-shaped network and sets the network parameters according to the size of the quantum communication data. Next, the present invention converts the original quantum data into noisy data by adding Gaussian noise, and uses this data to train the quantum U-shaped network. Finally, the present invention generates noisy quantum data and inputs it into the trained quantum U-shaped network to generate new quantum communication data.

[0077] Example 4: Generating Quantum Cryptographic Data

[0078] In this embodiment, the present invention aims to generate data related to quantum cryptography. First, the present invention collects N quantum cryptography data points and constructs corresponding quantum circuits. Then, the present invention builds a quantum U-shaped network and sets the network parameters according to the size of the quantum cryptography data. Next, the present invention converts the original quantum data into noisy data by adding Gaussian noise, and uses this data to train the quantum U-shaped network. Finally, the present invention generates noisy quantum data and inputs it into the trained quantum U-shaped network to generate new quantum cryptography data.

[0079] These four examples demonstrate how quantum data generation methods can be applied to generate new quantum data in different quantum domains. This approach helps to expand existing finite quantum data and provides more data resources for fields such as quantum machine learning, quantum data processing, and quantum complexity simulation.

[0080] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for generating quantum data, characterized in that, The quantum data generation method trains the designed quantum U-shaped network using existing limited quantum data, enabling the quantum U-shaped network to transform random noise quantum data into quantum data of the same type as the training data, i.e., generating a completely new quantum data of the same type. The quantum data generation method specifically includes the following steps: Step 1: Collect N types of quantum data that need to be generated, and construct various quantum circuits to collect them on the existing NISQ quantum computer; Step two: Construct a quantum U-shaped network. The quantum U-shaped network includes downsampling and upsampling parts. The downsampling part is implemented by n-fold quantum convolutional layers and pooling layers, while the upsampling part is implemented by n-fold quantum convolutional layers. Each quantum convolutional layer consists of a two-qubit quantum gate, and the pooling layer is implemented by a single-qubit quantum gate and a controlled NOT gate to compress the entangled information of the two qubits into a single qubit. n is a positive integer, determined according to the size of the generated quantum data. Step 3: Gradually add Gaussian noise to the N quantum data generated in Step 1 until they become completely noisy quantum data; apply a random rotation operation Ry(θ) to each bit of the quantum data, where θ follows a normal distribution; Step four: Using the full-noise quantum data and the corresponding noise-free data pairs generated in step three, train the diffusion model of the quantum U-shaped network built in step two to make it converge. Step 5: Generate a noisy quantum data set and repeatedly input it into the quantum U-shaped network trained in step 4. This will generate a new set of quantum data of the same type as the quantum data collected in step 1. Repeating this step multiple times will generate multiple new sets of quantum data.

2. The quantum data generation method as described in claim 1, characterized in that, The quantum data generation method employs a denoising diffusion generation model during the training process; the quantum U-shaped network consists of two bridge-connected multi-scale entangled renormalization networks; the quantum U-shaped network adopts a U-shaped network structure Q-Unet composed of downsampling-upsampling full quantum convolutional networks.

3. The quantum data generation method as described in claim 1, characterized in that, The 3-bit GHz type of the quantum data generation method is: ， ， The 8 bits of information are encoded onto a classical graph. An 8-bit quantum state is encoded onto a 16 × 16 graph. The pixel with graph coordinates (0, 0) corresponds to the quantum computing basis |00000000>, and the pixel with graph coordinates (16, 16) corresponds to the quantum computing basis |11111111>. Each pixel on the graph corresponds to the computing basis of this 8-bit quantum state, and the height of the bar on each pixel corresponds to the probability of the quantum state in that computing basis.

4. The quantum data generation method as described in claim 3, characterized in that, In the quantum data generation method, the success rate of quantum state generation is measured by the sum of the probabilities of the |000> and |111> components of the quantum state in the data generated by the quantum U-shaped network; whether the generated quantum state belongs to the same class as the GHZ-type quantum states of the training network is determined by detecting the noise level P of the generated data. noise Measure; if P noise If the noise level is less than ε, the generated data belongs to the GHZ class. ε is a set threshold that can be set according to the usage scenario; noise level P noise The definition of P is as follows: noise = 1 − P0 − P 255 Here (P0, P) 255 ) represent the probabilities of the generated quantum state being in the computation basis (|00000000>, |11111111>).

5. A computer device, characterized in that, The computer device includes a memory and a processor. The memory stores a computer program, which, when executed by the processor, causes the processor to perform the quantum data generation method according to any one of claims 1 to 4.

6. A computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the quantum data generation method according to any one of claims 1 to 4.

7. An information data processing terminal, characterized in that, The information data processing terminal is used to implement the quantum data generation method according to any one of claims 1 to 4.

8. A quantum data generation system based on the quantum data generation method according to any one of claims 1 to 4, characterized in that, The quantum data generation system includes: The data training module is used to train the designed quantum U-shaped network using existing limited quantum data; The data generation module enables the quantum U-shaped network to transform random noise quantum data into quantum data of the same type as the training data, that is, to generate a completely new quantum data of the same kind.

9. A quantum device, characterized in that, The quantum device includes the quantum data generation system of claim 8.

Images