Generation method and generation device of quantum circuit auto-encoder

Abstract

The invention provides a quantum line auto-encoder generation method and device, and the method comprises the steps: generating a quantum line encoder and a first quantum channel according to an input quantum channel, and generating a second quantum channel according to an initial quantum line encoder and the first quantum channel; generating a target quantum state according to the second quantum channel and the initial quantum state; iteratively adjusting the parameter value of the initial quantum line encoder, so as to determine the single-bit noise and the target parameter value of the parameter-containing quantum line according to the target loss value of the target quantum state corresponding to the initial quantum line encoder; a third quantum channel is generated according to a target quantum line encoder and the first quantum channel, and the target quantum line encoder is constructed according to the single-bit noise and the target parameter value of the parameter-containing quantum line; generating a potential channel and a noise channel according to the third quantum channel; a reconstructed quantum channel is generated from the potential channel, the noise channel, and the target quantum line decoder.

Description

Technical Field

[0001] The present application relates to the field of quantum computing technology, and more specifically, to a method for generating a quantum circuit autoencoder, a generating device, an electronic device, a computer-readable storage medium, and a computer program product. Background Art

[0002] As a popular neural network tool, the network structure of the autoencoder maps the input raw data from high-dimensional to low-dimensional data, and then maps the low-dimensional data back to the reconstructed data of the original dimension, and then trains the neural network parameters with the purpose of reducing the reconstruction error of the reconstructed data relative to the original data.

[0003] In the process of implementing the concept of the present application, it was found that the fidelity of the quantum channel reconstructed by the quantum circuit autoencoder of the related art was poor. Summary of the invention

[0004] In view of this, the present application provides a method for generating a quantum circuit autoencoder, a generating device, an electronic device, a computer-readable storage medium, and a computer program product.

[0005] One aspect of the present application provides a method for generating a quantum circuit autoencoder, comprising:

[0006] In response to the compression instruction, an initial quantum circuit encoder and a first quantum channel are generated according to the input quantum channel, wherein each layer of the initial quantum circuit encoder is generated according to a plurality of single-bit noises and parameter-containing quantum circuits;

[0007] Generate a second quantum channel according to the initial quantum circuit encoder and the first quantum channel;

[0008] Generate a target quantum state according to the second quantum channel and the initial quantum state, wherein the initial quantum state is obtained according to the maximum mixed state and the maximum entangled state;

[0009] Iteratively adjusting the parameter value of the initial quantum circuit encoder to determine the target parameter value of the single-bit noise and the parameter-containing quantum circuit according to the target loss value of the target quantum state corresponding to the initial quantum circuit encoder;

[0010] Generate a third quantum channel according to a target quantum circuit encoder and the first quantum channel, wherein the target quantum circuit encoder is constructed according to the single-bit noise and the target parameter value of the parameter-containing quantum circuit;

[0011] According to the third quantum channel, a potential channel and a noise channel are generated;

[0012] In response to the reconstruction instruction, the reconstructed quantum channel is generated according to the potential channel, the noise channel and the target quantum circuit decoder, wherein the target quantum circuit decoder is the conjugate transpose of the target quantum circuit encoder, and the quantum circuit autoencoder includes the target quantum circuit encoder and the target quantum circuit decoder.

[0013] According to an embodiment of the present application, generating a potential channel and a noise channel according to the third quantum channel includes:

[0014] When the third quantum channel acts on a target quantum space, a deflection operation is performed on the third quantum channel to obtain the potential channel, wherein the original quantum space includes a potential space and a junk space, and the target quantum space includes the original quantum space and an isomorphic space corresponding to the junk space;

[0015] Extracting the system state of the quantum system corresponding to the garbage space in the target quantum space, and measuring the system state under a standard basis to obtain state information;

[0016] The noise channel is generated according to the state information.

[0017] According to an embodiment of the present application, the target quantum circuit decoder includes a first decoder and a second decoder;

[0018] The reconstructed quantum channel is generated according to the potential channel, the noise channel and the target quantum circuit decoder, including:

[0019] Generate a target channel according to the potential channel and the noise channel;

[0020] The reconstructed quantum channel is generated according to the target channel, the first decoder and the second decoder.

[0021] According to an embodiment of the present application, generating a second quantum channel according to an initial quantum circuit encoder and a first quantum channel includes:

[0022] Acquire a quantum circuit set, wherein the quantum circuit set includes a plurality of first quantum circuits;

[0023] Processing the plurality of the first quantum circuits based on a preprocessing formula to obtain the first quantum channel;

[0024] The second quantum channel is generated according to the initial quantum circuit encoder and the first quantum channel.

[0025] According to an embodiment of the present application, the above-mentioned original quantum space includes potential space and garbage space;

[0026] According to an embodiment of the present application, generating a target quantum state according to the second quantum channel and the initial quantum state includes:

[0027] The maximum mixing state acting on the latent space;

[0028] The maximum entangled state acting on the garbage space and its reconstruction space;

[0029] According to the maximum mixed state and the maximum entangled state, the initial quantum state is generated;

[0030] The second quantum channel is applied to the initial quantum state to obtain the target quantum state.

[0031] According to an embodiment of the present application, iteratively adjusting the parameter value of the initial quantum circuit encoder to determine the target parameter value of the single-bit noise and the parameter-containing quantum circuit according to the target loss value of the target quantum state corresponding to the initial quantum circuit encoder includes:

[0032] Determine the single-bit noise and the initial parameter values ​​of the parameter-containing quantum circuit, so as to input the target quantum state corresponding to the initial parameter value into the target loss function to obtain an initial loss value;

[0033] The initial parameter value is iteratively adjusted according to the initial loss value to obtain an adjusted parameter value, so that the adjusted parameter value is determined as the target parameter value when an iteration condition is met.

[0034] According to an embodiment of the present application, iteratively adjusting the initial parameter value according to the initial loss value to obtain an adjusted parameter value, so as to determine the adjusted parameter value as the target parameter value when an iteration condition is met, includes:

[0035] During the i-th iteration, the i-th gradient information of the i-th initial loss value corresponding to the i-th iteration is calculated using the finite difference algorithm;

[0036] Generate an i+1th initial parameter value for an i+1th iteration according to the i-th initial parameter value in the i-th iteration and the i-th gradient information, so as to generate an i+1th gradient information according to the i+1th initial parameter value in the i+1th iteration;

[0037] When the difference between the i-th gradient information and the (i+1)-th gradient information is less than a preset gradient threshold, the i-th initial parameter value corresponding to the i-th gradient information is determined as the target parameter value.

[0038] According to an embodiment of the present application, iteratively adjusting the initial parameter value according to the initial loss value to obtain an adjusted parameter value, so as to determine the adjusted parameter value as the target parameter value when an iteration condition is met, includes:

[0039] During the i-th iteration, the i-th gradient information of the i-th initial loss value corresponding to the i-th iteration is calculated using the finite difference algorithm;

[0040] Generate an i+1th initial parameter value for an i+1th iteration according to the i-th initial parameter value in the i-th iteration and the i-th gradient information;

[0041] When the number of iterations i satisfies a preset iteration threshold, the i+1th initial parameter value is determined as the target parameter value.

[0042] According to an embodiment of the present application, the target quantum circuit encoder includes a first encoder and a second encoder;

[0043] The step of generating a third quantum channel according to the target quantum circuit encoder and the first quantum channel comprises:

[0044] According to the target parameter value and the initial quantum circuit encoder, respectively generate the first encoder and the second encoder;

[0045] The third quantum channel is generated according to the first encoder, the second encoder and the first quantum channel.

[0046] Another aspect of the present application provides a device for generating a quantum circuit autoencoder, comprising:

[0047] A first generation module is used to generate an initial quantum circuit encoder and a first quantum channel according to an input quantum channel in response to a compression instruction, wherein each layer of the initial quantum circuit encoder is generated according to a plurality of single-bit noises and parameter-containing quantum circuits;

[0048] A second generating module is used to generate a second quantum channel according to the initial quantum circuit encoder and the first quantum channel in response to the reconstruction instruction;

[0049] A third generating module is used to generate a target quantum state according to the second quantum channel and the initial quantum state, wherein the initial quantum state is obtained according to the maximum mixed state and the maximum entangled state;

[0050] A determination module, configured to iteratively adjust the parameter value of the initial quantum circuit encoder to determine the target parameter value of the single-bit noise and the parameter-containing quantum circuit according to the target loss value of the target quantum state corresponding to the initial quantum circuit encoder;

[0051] A fourth generation module is used to generate a third quantum channel according to a target quantum circuit encoder and the first quantum channel, wherein the target quantum circuit encoder is constructed according to the single-bit noise and the target parameter value of the parameter-containing quantum circuit;

[0052] A fifth generating module, used for generating a potential channel and a noise channel according to the third quantum channel;

[0053] The sixth generation module is used to generate the reconstructed quantum channel according to the potential channel, the noise channel and the target quantum circuit decoder, wherein the target quantum circuit decoder is the conjugate transpose of the target quantum circuit encoder, and the quantum circuit autoencoder includes the target quantum circuit encoder and the target quantum circuit decoder.

[0054] Another aspect of the present application provides an electronic device, comprising:

[0055] one or more processors;

[0056] a memory for storing one or more programs,

[0057] When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described above.

[0058] Another aspect of the present application provides a computer-readable storage medium storing computer-executable instructions, which are used to implement the method described above when executed.

[0059] Another aspect of the present application provides a computer program product, which includes computer executable instructions, and the instructions are used to implement the method as described above when executed.

[0060] According to an embodiment of the present application, a second quantum channel is generated according to an initial quantum circuit encoder and a first quantum channel, a target quantum state is generated in combination with an initial quantum state, a single-bit noise for a target quantum circuit encoder and a target parameter value of a parameter-containing quantum circuit are determined according to a target loss value of the target quantum state corresponding to the initial quantum circuit encoder, a third quantum channel is generated in combination with the first quantum channel, a potential channel and a noise channel are generated according to the third quantum channel, and a reconstructed quantum channel is generated according to the potential channel, the noise channel and the target quantum circuit decoder. By constructing a parameter-containing quantum channel with the assistance of a single-bit noise of a controllable parameter and a target parameter value of a parameter-containing quantum circuit, and using it as the basic structure of an encoder and a decoder, it is more suitable for processing information compression tasks in a quantum channel composed of a group of quantum circuits, and at the same time, in the reconstruction stage, the original quantum channel is assisted in reconstruction in the form of noise, so that the information contained in the quantum circuit can complete the compression-recovery process with higher reconstruction fidelity. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] The above and other objects, features and advantages of the present application will become more apparent through the following description of the embodiments of the present application with reference to the accompanying drawings, in which:

[0062] Figure 1 A flow chart of a method for generating a quantum circuit autoencoder according to an embodiment of the present application is shown;

[0063] Figure 2 A schematic diagram of the structure of an initial quantum circuit encoder according to an embodiment of the present application is shown;

[0064] Figure 3 A schematic diagram of training target parameter values ​​according to an embodiment of the present application is shown;

[0065] Figure 4 A schematic diagram of a model of a quantum circuit autoencoder according to an embodiment of the present application is shown;

[0066] Figure 5 A block diagram of a generation device of a quantum circuit self-encoder according to an embodiment of the present application is shown;

[0067] Figure 6 A block diagram of an electronic device suitable for implementing the method described above according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0068] Below, embodiments of the present application will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present application. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present application. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present application.

[0069] The terms used herein are only for describing specific embodiments and are not intended to limit the present application. The terms "include", "comprising", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0070] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0071] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0072] The quantum autoencoder is used to process the information contained in the quantum state, mapping the input high-dimensional quantum state to a low-dimensional quantum state through a parameter-containing quantum circuit, and then mapping the low-dimensional quantum state to a reconstructed quantum state of the original dimension. The two mapping functions are controlled by two parameter-containing quantum circuit structures controlled by a set of parameters. The quantum autoencoder also trains the control parameters of the parameter-containing quantum circuit with the purpose of reducing the reconstruction error of the reconstructed data relative to the original data.

[0073] However, in practice, many tasks store information in quantum circuits. For example, many quantum machine learning tasks map classical data to quantum data through quantum circuits. At this time, the information is contained in both the quantum circuit and the quantum state obtained after the execution of the quantum circuit. The autoencoder scheme for quantum states does not seem to be directly applicable to quantum circuit information processing.

[0074] The technical solutions for quantum gate compression proposed in the related art are all prototypes that process the information in the quantum circuit in a similar way to an autoencoder. However, they only consider the compression and recovery scheme of the overall quantum gate composed of multiple single-qubit gates of the same form in an independent and identically distributed manner, but their reconstruction fidelity is limited.

[0075] In view of this, an embodiment of the present application provides a method and a device for generating a quantum circuit autoencoder, the method comprising: generating a quantum circuit encoder and a first quantum channel according to an input quantum channel in response to a compression instruction; generating a second quantum channel according to an initial quantum circuit encoder and the first quantum channel in response to a reconstruction instruction, wherein each layer in the initial quantum circuit encoder is generated according to a plurality of single-bit noises and parameter-containing quantum circuits; generating a target quantum state according to the second quantum channel and the initial quantum state; iteratively adjusting the parameter value of the single-bit noise to determine the target parameter value of the single-bit noise and the parameter-containing quantum circuit according to the target loss value of the target quantum state corresponding to the single-bit noise; generating a third quantum channel according to the target quantum circuit encoder and the first quantum channel, wherein the target quantum circuit encoder is constructed according to the target parameter value of the single-bit noise and the parameter-containing quantum circuit; generating a potential channel and a noise channel according to the third quantum channel; generating a reconstructed quantum channel according to the potential channel, the noise channel and the target quantum circuit decoder.

[0076] Figure 1 A flow chart of a method for generating a quantum circuit autoencoder according to an embodiment of the present application is shown. Figure 2 A schematic structural diagram of an initial quantum circuit encoder according to an embodiment of the present application is shown.

[0077] like Figure 1 As shown, the method for generating a quantum circuit self-encoder includes operations S101 to S107.

[0078] In operation S101, in response to a compression instruction, an initial quantum circuit encoder and a first quantum channel are generated according to an input quantum channel, wherein each layer in the initial quantum circuit encoder is generated according to a plurality of single-bit noises and parameter-containing quantum circuits;

[0079] In operation S102, a second quantum channel is generated according to the initial quantum circuit encoder and the first quantum channel;

[0080] In operation S103, a target quantum state is generated according to the second quantum channel and the initial quantum state, wherein the initial quantum state is obtained according to the maximum mixed state and the maximum entangled state;

[0081] In operation S104, iteratively adjusting the parameter value of the encoder to determine the target parameter value of the single-bit noise and the parameter-containing quantum circuit according to the target loss value of the target quantum state;

[0082] In operation S105, a third quantum channel is generated according to a target quantum circuit encoder and the first quantum channel, wherein the target quantum circuit encoder is constructed according to single-bit noise and a target parameter value of the parameter-containing quantum circuit;

[0083] In operation S106, a potential channel and a noise channel are generated according to the third quantum channel;

[0084] In operation S107, in response to the reconstruction instruction, a reconstructed quantum channel is generated according to the potential channel, the noise channel and the target quantum circuit decoder, wherein the target quantum circuit decoder is the conjugate transpose of the target quantum circuit encoder, and the quantum circuit autoencoder includes the target quantum circuit encoder and the target quantum circuit decoder.

[0085] According to an embodiment of the present application, when the method for generating a quantum circuit autoencoder is executed, a generation instruction required to execute the quantum circuit autoencoder is first obtained according to a generation request of the quantum circuit autoencoder, and the generation instruction includes a compression instruction and a reconstruction instruction.

[0086] According to an embodiment of the present application, the initial quantum state is the state of the original quantum space, which is specifically generated by the maximum mixed state and the maximum entangled state.

[0087] According to an embodiment of the present application, the initial quantum circuit encoder is a parameter-containing quantum channel assisted by controlled noise, and the specific structure is as follows: Figure 2 As shown, it consists of two parts. The first part is a plurality of single-bit noises controlled by parameters, namely Figure 2 In The second part is the traditional parameter-containing quantum circuit, that is, The structure on the right. Figure 2 What is shown is a layer of parameterized quantum channel. In specific tasks, multiple layers of this structure can be used according to the problem, i.e. Figure 2 Only one layer is shown, which can be found in Figure 2 Add multiple layers of the same two-part structure to the right of , where q0, q1, q2, and q3 are all quantum bits, and RX is a controlled Pauli X-rotation gate.

[0088] Specifically, the structures of the initial quantum circuit encoder and the target quantum circuit decoder can be expressed by Formula 1:

[0089] (1)

[0090] in, To control the parameters, is the parameter control noise of the jth layer, i.e., single-bit noise, is the parameter-containing quantum circuit of the jth layer. Parameter-controlled noise includes common noises, such as depolarization noise, bit flip noise, phase damping noise, etc. The specific form of the parameter-containing quantum circuit is shown in formula (2):

[0091] (2)

[0092] in, and They are Hermitian and unitary operators respectively.

[0093] According to an embodiment of the present application, the first quantum channel acts on the same Hilbert space A quantum channel is obtained by preprocessing the data of multiple quantum circuits on the same quantum state. A second quantum channel is generated based on the initial quantum circuit encoder and the first quantum channel. According to the second quantum channel and the initial quantum state as shown in formula (3), , generating the target quantum state .

[0094] (3)

[0095] By partially measuring the target quantum state, the target loss value corresponding to this group of parameters can be obtained. The target loss value is calculated as follows:

[0096]

[0097] According to an embodiment of the present application, the parameter value of the single-bit noise is iteratively adjusted , to determine the target parameter value of the single-bit noise according to the target loss value of the target quantum state corresponding to the single-bit noise . According to the target parameter values ​​of single-bit noise and parameterized subcircuits The target quantum circuit encoder and the first quantum channel are constructed to generate the third quantum channel. Based on the third quantum channel and the partial measurement results of the target quantum state, the potential channel is generated. and noisy channel . Generate a reconstructed quantum channel based on the potential channel, the noise channel and the target quantum circuit decoder .

[0098] According to an embodiment of the present application, a second quantum channel is generated according to an initial quantum circuit encoder and a first quantum channel, a target quantum state is generated in combination with an initial quantum state, a single-bit noise for a target quantum circuit encoder and a target parameter value of a parameter-containing quantum circuit are determined according to a target loss value of the target quantum state corresponding to the target quantum circuit encoder, a third quantum channel is generated in combination with the first quantum channel, a potential channel and a noise channel are generated according to the third quantum channel, and a reconstructed quantum channel is generated according to the potential channel, the noise channel and the target quantum circuit decoder. By constructing a parameter-containing quantum channel with the assistance of a single-bit noise of controllable parameters and a target parameter value of a parameter-containing quantum circuit, and using it as the basic structure of an encoder and a decoder, it is more suitable for processing information compression tasks in a quantum channel composed of a group of quantum circuits, and at the same time, in the reconstruction stage, the original quantum channel is assisted in reconstruction in the form of noise, so that the information contained in the quantum circuit can complete the compression-recovery process with higher reconstruction fidelity.

[0099] According to an embodiment of the present application, generating a potential channel and a noise channel according to a third quantum channel includes:

[0100] Performing a deflection operation on the third quantum channel to obtain a potential channel, wherein the original quantum space includes the potential space and the junk space, and the target quantum space includes the original quantum space and an isomorphic space corresponding to the junk space;

[0101] Extract the system state of the quantum system corresponding to the garbage space in the target quantum space, and measure the subspace of the system state under the standard basis to obtain the state information;

[0102] A noise channel is generated based on the state information.

[0103] According to an embodiment of the present application, the channel acts on the spatial target quantum space On, among them, is a potential channel, For garbage channels, for The isomorphic space of , performs the deviation operation on the channel, and obtains the potential channel, as shown in formula (4):

[0104] (4)

[0105] According to an embodiment of the present application, the system state of the quantum system corresponding to the garbage space and its reconstruction space in the target quantum space is extracted, and the state of this system is recorded as , perform measurement operations on it under the standard basis and obtain status information , . The status information is shown in formula (5):

[0106] (5)

[0107] According to an embodiment of the present application, a noise channel as shown in formula (6) is constructed according to the state information .

[0108] (6)

[0109] According to an embodiment of the present application, the target quantum circuit decoder includes a first decoder and the second decoder .

[0110] Wherein, according to the potential channel, the noise channel and the target quantum circuit decoder, a reconstructed quantum channel is generated, including:

[0111] Generate a target channel according to the potential channel and the noise channel;

[0112] A reconstructed quantum channel is generated according to the target channel, the first decoder and the second decoder.

[0113] According to an embodiment of the present application, the potential channel and noisy channel A new channel, namely the target channel, is synthesized and constructed together with the first decoder and the second decoder as shown in formula (7): :

[0114] (7)

[0115] According to an embodiment of the present application, generating a second quantum channel according to an initial quantum circuit encoder and a first quantum channel includes:

[0116] Acquire a quantum circuit set, wherein the quantum circuit set includes a plurality of first quantum circuits acting on the same Hilbert space;

[0117] Processing the plurality of first quantum circuits based on a preprocessing formula to obtain a first quantum channel;

[0118] A second quantum channel is generated according to the initial quantum circuit encoder and the first quantum channel.

[0119] According to an embodiment of the present application, a quantum circuit set is obtained , the quantum circuit set consists of multiple actions in the same Hilbert space The first quantum circuit . Process multiple first quantum circuits to obtain the first quantum channel , satisfying formula (8):

[0120] (8)

[0121] in, For any quantum state, is the conjugate transpose.

[0122] According to an embodiment of the present application, according to the initial quantum circuit encoder , and the first quantum channel , generating the second quantum channel as shown in formula (3).

[0123] According to an embodiment of the present application, the original quantum space includes the potential space and Junkspace and the isomorphic space of junkspace .

[0124] According to an embodiment of the present application, generating a target quantum state according to a second quantum channel and an initial quantum state includes:

[0125] The maximum mixing state acting on the latent space;

[0126] The maximum entangled state acting on the garbage space and its reconstruction space;

[0127] Generate an initial quantum state according to the maximum mixed state and the maximum entangled state;

[0128] The second quantum channel is applied to the initial quantum state to obtain the target quantum state.

[0129] According to an embodiment of the present application, the original channel (i.e., the first quantum channel ) is compressed to obtain the space where the channel is located, which is the mixed state above In the space where the entangled state is a maximum entangled state acting on the junk space (i.e., the space that is redundant with the potential space) and its isomorphic space. In which, assuming that the original space is , then the mixed state and entangled states As shown in formula (9) and formula (10) respectively:

[0130] (9)

[0131] (10)

[0132] in, and The potential space and Junkspace The dimensions of , i and j are both standard bases.

[0133] According to the embodiment of the present application, according to the mixed state and entangled states , generating the initial quantum state ; The second quantum channel and the initial quantum state shown in formula (3) Compound into target quantum state .

[0134] Figure 3 A schematic diagram of training target parameter values ​​according to an embodiment of the present application is shown.

[0135] According to an embodiment of the present application, iteratively adjusting the parameter value of the initial quantum circuit encoder to determine the single-bit noise and the target parameter value of the parameter-containing quantum circuit according to the target loss value of the target quantum state corresponding to the initial quantum circuit encoder includes:

[0136] Determine the initial parameter values ​​of the single-bit noise and the parameter-containing quantum circuit (generally by random initialization), so as to input the target quantum state corresponding to the initial parameter value into the target loss function to obtain the initial loss value;

[0137] The initial parameter value is iteratively adjusted according to the initial loss value to obtain an adjusted parameter value, so as to determine the adjusted parameter value as the target parameter value when an iteration condition is met.

[0138] According to an embodiment of the present application, iteratively adjusting the initial parameter value according to the initial loss value to obtain the adjusted parameter value, and determining the adjusted parameter value as the target parameter value when the iteration condition is met, includes:

[0139] During the i-th iteration, the i-th gradient information of the i-th initial loss value corresponding to the i-th iteration is calculated using the finite difference algorithm;

[0140] Generate an i+1th initial parameter value for an i+1th iteration according to the i-th initial parameter value and the i-th gradient information in the i-th iteration, so as to generate an i+1th gradient information according to the i+1th initial parameter value for the i+1th iteration;

[0141] When the difference between the i-th gradient information and the (i+1)-th gradient information is less than a preset gradient threshold, the i-th initial parameter value corresponding to the i-th gradient information is determined as the target parameter value.

[0142] According to an embodiment of the present application, iteratively adjusting the initial parameter value according to the initial loss value to obtain the adjusted parameter value, and determining the adjusted parameter value as the target parameter value when the iteration condition is met, includes:

[0143] During the i-th iteration, the i-th gradient information of the i-th initial loss value corresponding to the i-th iteration is calculated using the finite difference algorithm;

[0144] Generate the i+1th initial parameter value of the i+1th iteration according to the i-th initial parameter value and the i-th gradient information in the i-th iteration process;

[0145] When the number of iterations i meets a preset iteration threshold, the i+1th initial parameter value is determined as the target parameter value.

[0146] According to an embodiment of the present application, a classical optimizer is used to train the target quantum state:

[0147] Reference Figure 3 , first randomly initialize the initial parameter value ;

[0148] Then repeat the following steps wheel:

[0149] (1) The initial parameter value of each round is , for the initial parameter value The corresponding target quantum state is input into the target loss function as shown in formula (11) to obtain the initial loss value;

[0150] (11)

[0151] The observation operator is , and They are and The corresponding number of systematic bits. is the maximum entangled state of the two bits, generated according to formula (12):

[0152] (12)

[0153] Among them, I, Z, and X are all Pauli operators.

[0154] (2) Use the finite difference method to estimate the gradient information of the target loss function at this point, as shown in formula (12):

[0155] (12)

[0156] in, and and The difference is that parameters, specifically expressed as , .

[0157] Update the initial parameter value according to the gradient information to obtain the initial parameter value for the next iteration , as shown in formula (13)

[0158] (13)

[0159] in is the learning rate set according to actual needs, for example 0.01, is the gradient information of the current point.

[0160] Determine whether the current gradient information has converged or whether the maximum training round has been reached. If any of the conditions are met, the loop is terminated, otherwise the above steps are repeated. In this way, the approximate optimal parameters can be output.

[0161] In one embodiment, when the difference between the i-th gradient information and the (i+1)-th gradient information is less than a preset gradient threshold, the i-th initial parameter value corresponding to the i-th gradient information is determined as the target parameter value.

[0162] According to the embodiment of the present application, the preset gradient threshold can be specifically set according to actual needs, for example, 10 -5 , when the difference is less than 10 -5 , it can be considered converged, so the i-th initial parameter value corresponding to the i-th gradient information is determined as the target parameter value

[0163] In another embodiment, when the number of iterations i satisfies a preset iteration threshold, the i+1th initial parameter value is determined as the target parameter value. It can be set according to actual needs, for example, T can be 50, 100, etc.

[0164] It should be noted that Figure 3 middle For the "potential" system, For the "junk" system, isomorphic to the “junk” system, is the input system of the first quantum channel, is the input system of the first quantum channel, For the "potential" system, For the "junk" system, Output system for the "junk" system.

[0165] Figure 4 A schematic diagram of a model of a quantum circuit autoencoder according to an embodiment of the present application is shown.

[0166] According to an embodiment of the present application, the target quantum circuit encoder includes a first encoder and the second encoder

[0167] The step of generating a third quantum channel according to the target quantum circuit encoder and the first quantum channel includes:

[0168] Generate a first encoder and a second encoder respectively according to the target parameter value and the initial quantum circuit encoder;

[0169] A second quantum channel is generated according to the first encoder, the second encoder and the first quantum channel.

[0170] According to an embodiment of the present application, according to the target parameter value and the initial quantum circuit encoder, the following are generated respectively: Figure 4 The first encoder shown and the second encoder .

[0171] According to an embodiment of the present application, according to the first encoder and the second encoder and the first quantum channel , generate the second quantum channel, as shown in formula (14):

[0172] (14).

[0173] According to an embodiment of the present application, the quantum channel is reconstructed by calculation With the original channel The fidelity of can verify the effect of training. The calculation method of fidelity is shown in formula (15):

[0174]

[0175] (15)

[0176] in, Indicates the original channel The corresponding Choi state is, and is a set of isomorphic quantum systems.

[0177] According to an embodiment of the present application, if the fidelity does not meet the preset fidelity requirement, the above-mentioned iterative update can be performed again, wherein the preset fidelity can be specifically set according to actual needs.

[0178] Figure 5 A block diagram of a generation device of a quantum circuit autoencoder according to an embodiment of the present application is shown.

[0179] like Figure 5As shown, the generation device 600 of the quantum circuit self-encoder includes a first generation module 510, a second generation module 520, a third generation module 530, a determination module 540, a fourth generation module 550, a fifth generation module 560, and a sixth generation module 570.

[0180] A first generating module 510 is used to generate an initial quantum circuit encoder and a first quantum channel according to an input quantum channel in response to a compression instruction, wherein each layer in the initial quantum circuit encoder is generated according to a plurality of single-bit noises and parameter-containing quantum circuits;

[0181] A second generating module 520 is used to generate a second quantum channel according to the initial quantum circuit encoder and the first quantum channel in response to the reconstruction instruction;

[0182] A third generating module 530 is used to generate a target quantum state according to the second quantum channel and an initial quantum state, wherein the initial quantum state is obtained by the quantum bit acting on the original quantum space;

[0183] A determination module 540, configured to iteratively adjust the parameter value of the initial quantum circuit encoder to determine the target parameter value of the single-bit noise and the parameter-containing quantum circuit according to the target loss value of the target quantum state corresponding to the initial quantum circuit encoder;

[0184] A fourth generating module 550 is used to generate a third quantum channel according to a target quantum circuit encoder and the first quantum channel, wherein the target quantum circuit encoder is constructed according to the single-bit noise and a target parameter value of the parameter-containing quantum circuit;

[0185] A fifth generating module 560, configured to generate a potential channel and a noise channel according to the third quantum channel;

[0186] The sixth generation module 570 is used to generate the reconstructed quantum channel according to the potential channel, the noise channel and the target quantum circuit decoder, wherein the target quantum circuit decoder is the conjugate transpose of the target quantum circuit encoder, and the quantum circuit autoencoder includes the target quantum circuit encoder and the target quantum circuit decoder.

[0187] According to an embodiment of the present application, a second quantum channel is generated according to an initial quantum circuit encoder and a first quantum channel, a target quantum state is generated in combination with an initial quantum state, a single-bit noise for a target quantum circuit encoder and a target parameter value of a parameter-containing quantum circuit are determined according to a target loss value obtained by measuring the target quantum state, a third quantum channel is generated in combination with the first quantum channel, a potential channel and a noise channel are generated according to the third quantum channel, and a reconstructed quantum channel is generated according to the potential channel, the noise channel and the target quantum circuit decoder. By constructing a parameter-containing quantum channel with the assistance of the target parameter value of the single-bit noise of a controllable parameter, and using it as the basic structure of an encoder and a decoder, it is more suitable for processing information compression tasks in a quantum channel composed of a group of quantum circuits, and at the same time, in the reconstruction stage, the original quantum channel is assisted in the reconstruction in the form of noise, so that the information contained in the quantum circuit can complete the compression-recovery process with higher reconstruction fidelity.

[0188] According to any one or more of the modules of the embodiments of the present application, at least part of the functions of any one of them can be implemented in one module. According to any one or more of the modules of the embodiments of the present application, they can be split into multiple modules for implementation. According to any one or more of the modules of the embodiments of the present application, they can be at least partially implemented as hardware circuits, such as field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems on chips, systems on substrates, systems on packages, application-specific integrated circuits (ASICs), or can be implemented by hardware or firmware in any other reasonable way of integrating or packaging the circuits, or implemented in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, according to one or more of the modules of the embodiments of the present application, they can be at least partially implemented as computer program modules, and when the computer program modules are run, the corresponding functions can be executed.

[0189] For example, any multiple of the first generation module 510, the second generation module 520, the third generation module 530, the determination module 540, the fourth generation module 550, the fifth generation module 560, and the sixth generation module 570 can be combined in one module / unit / sub-unit for implementation, or any one of the modules / units / sub-units can be split into multiple modules / units / sub-units. Alternatively, at least part of the functions of one or more of the modules / units / sub-units can be combined with at least part of the functions of other modules / units / sub-units and implemented in one module / unit / sub-unit.

[0190] According to an embodiment of the present application, at least one of the first generation module 510, the second generation module 520, the third generation module 530, the determination module 540, the fourth generation module 550, the fifth generation module 560, and the sixth generation module 570 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware, and firmware or in an appropriate combination of any of them. Alternatively, at least one of the first generation module 510, the second generation module 520, the third generation module 530, the determination module 540, the fourth generation module 550, the fifth generation module 560, and the sixth generation module 570 can be at least partially implemented as a computer program module, and when the computer program module is run, the corresponding function can be executed.

[0191] It should be noted that the generating device part of the quantum circuit self-encoder in the embodiment of the present application corresponds to the generating method part of the quantum circuit self-encoder in the embodiment of the present application. The description of the generating device part of the quantum circuit self-encoder specifically refers to the generating method part of the quantum circuit self-encoder, which will not be repeated here.

[0192] Figure 6 A block diagram of an electronic device suitable for implementing the method described above according to an embodiment of the present application is shown. Figure 6 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0193] like Figure 6 As shown, the electronic device 600 according to an embodiment of the present application includes a processor 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage part 608 to a random access memory (RAM) 603. The processor 601 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a dedicated microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 601 may also include an onboard memory for caching purposes. The processor 601 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present application.

[0194] In RAM 603, various programs and data required for the operation of electronic device 600 are stored. Processor 601, ROM 602 and RAM 603 are connected to each other via bus 604. Processor 601 performs various operations of the method flow according to the embodiment of the present application by executing the program in ROM 602 and / or RAM 603. It should be noted that the program can also be stored in one or more memories other than ROM 602 and RAM 603. Processor 601 can also perform various operations of the method flow according to the embodiment of the present application by executing the program stored in the one or more memories.

[0195] According to an embodiment of the present application, the electronic device 600 may further include an input / output (I / O) interface 605, which is also connected to the bus 604. The electronic device 600 may further include one or more of the following components connected to the input / output (I / O) interface 605: an input portion 606 including a keyboard, a mouse, etc.; an output portion 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 608 including a hard disk, etc.; and a communication portion 609 including a network interface card such as a LAN card, a modem, etc. The communication portion 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the input / output (I / O) interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed into the storage portion 608 as needed.

[0196] According to an embodiment of the present application, the method flow according to the embodiment of the present application can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable storage medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or installed from the removable medium 611. When the computer program is executed by the processor 601, the above-mentioned functions defined in the system of the embodiment of the present application are executed. According to an embodiment of the present application, the system, equipment, device, module, unit, etc. described above can be implemented by a computer program module.

[0197] The present application also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present application is implemented.

[0198] According to an embodiment of the present application, the computer-readable storage medium may be a non-volatile computer-readable storage medium. For example, it may include, but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, apparatus, or device.

[0199] For example, according to an embodiment of the present application, the computer-readable storage medium may include the ROM 602 and / or the RAM 603 described above and / or one or more memories other than the ROM 602 and the RAM 603 .

[0200] An embodiment of the present application also includes a computer program product, which includes a computer program, which contains program code for executing the method provided by the embodiment of the present application. When the computer program product runs on an electronic device, the program code is used to enable the electronic device to implement the method provided by the embodiment of the present application.

[0201] When the computer program is executed by the processor 601, the above functions defined in the system / device of the embodiment of the present application are executed. According to the embodiment of the present application, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0202] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 609, and / or installed from a removable medium 611. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0203] According to an embodiment of the present application, the program code for executing the computer program provided by the embodiment of the present application can be written in any combination of one or more programming languages, and specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, such as Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or completely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).

[0204] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they may sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of the boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions. It will be appreciated by those skilled in the art that the features recorded in the various embodiments of the present application can be combined and / or combined in a variety of ways, even if such a combination or combination is not clearly recorded in the present application. In particular, without departing from the spirit and teachings of the present application, the features described in the various embodiments of the present application may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present application.

[0205] The embodiments of the present application are described above. However, these embodiments are only for the purpose of illustration, and are not intended to limit the scope of the present application. Although each embodiment is described above, this does not mean that the measures in each embodiment cannot be used in combination advantageously. The present application does not depart from the scope of the present application, and those skilled in the art may make various substitutions and modifications, which should all fall within the scope of the present application.

Claims

1. A method for generating a quantum circuit autoencoder, characterized in that: include: In response to the compression instruction, an initial quantum circuit encoder and a first quantum channel are generated according to the input quantum channel, wherein each layer in the initial quantum circuit encoder is generated according to a plurality of single-bit noises and parameter-containing quantum circuits; Generate a second quantum channel according to the initial quantum circuit encoder and the first quantum channel; Generate a target quantum state according to the second quantum channel and an initial quantum state, wherein the initial quantum state is a state of an original quantum space; Iteratively adjusting the parameter value of the initial quantum circuit encoder to determine the single-bit noise and the target parameter value of the parameter-containing quantum circuit according to the target loss value of the target quantum state corresponding to the initial quantum circuit encoder; generating a third quantum channel according to a target quantum circuit encoder and the first quantum channel, wherein the target quantum circuit encoder is constructed according to the single-bit noise and a target parameter value of the parameter-containing quantum circuit; generating a potential channel and a noise channel according to the third quantum channel; In response to a reconstruction instruction, a reconstructed quantum channel is generated according to the potential channel, the noisy channel and a target quantum circuit decoder, wherein the target quantum circuit decoder is the conjugate transpose of the target quantum circuit encoder, and the quantum circuit autoencoder includes the target quantum circuit encoder and the target quantum circuit decoder.

2. The method according to claim 1, characterized in that According to the third quantum channel, generating a potential channel and a noise channel, including: When the third quantum channel acts on a target quantum space, a deflection operation is performed on the third quantum channel to obtain the potential channel, wherein the original quantum space includes a potential space and a junk space, and the target quantum space includes the original quantum space and an isomorphic space corresponding to the junk space; Extracting a system state of a quantum system corresponding to the garbage space in the target quantum space, and measuring the system state under a standard basis to obtain state information; The noise channel is generated according to the state information.

3. The method according to claim 1 or 2, characterized in that: The target quantum circuit decoder includes a first decoder and a second decoder; The step of generating the reconstructed quantum channel according to the potential channel, the noise channel and the target quantum circuit decoder comprises: Generate a target channel according to the potential channel and the noise channel; The reconstructed quantum channel is generated according to the target channel, the first decoder and the second decoder.

4. The method according to claim 1, characterized in that: Generating a second quantum channel according to the initial quantum circuit encoder and the first quantum channel comprises: Acquire a quantum circuit set, wherein the quantum circuit set includes a plurality of first quantum circuits; Processing the plurality of first quantum circuits based on a preprocessing formula to obtain the first quantum channel; The second quantum channel is generated according to the initial quantum circuit encoder and the first quantum channel.

5. The method according to claim 1 or 4, characterized in that: The original quantum space includes potential space and junk space; The step of generating a target quantum state according to the second quantum channel and the initial quantum state comprises: The maximum mixing state acting on the latent space; The maximum entangled state acting on the garbage space and its reconstruction space; generating the initial quantum state according to the maximum mixed state and the maximum entangled state; The second quantum channel is applied to the initial quantum state to obtain the target quantum state.

6. The method according to claim 1, characterized in that Iteratively adjusting the parameter value of the initial quantum circuit encoder to determine the single-bit noise and the target parameter value of the parameter-containing quantum circuit according to the target loss value of the target quantum state corresponding to the initial quantum circuit encoder, comprising: Determining initial parameter values ​​of the single-bit noise and the parameter-containing quantum circuit, so as to input a target quantum state corresponding to the initial parameter value into a target loss function to obtain an initial loss value; The initial parameter value is iteratively adjusted according to the initial loss value to obtain an adjusted parameter value, so as to determine the adjusted parameter value as the target parameter value when an iteration condition is met.

7. The method according to claim 6, characterized in that Iteratively adjusting the initial parameter value according to the initial loss value to obtain an adjusted parameter value, and determining the adjusted parameter value as the target parameter value when an iteration condition is satisfied, comprising: During the i-th iteration, using a finite difference algorithm to calculate the i-th gradient information of the i-th initial loss value corresponding to the i-th iteration; Generate an i+1th initial parameter value for an i+1th iteration according to the i-th initial parameter value and the i-th gradient information in the i-th iteration, so as to generate an i+1th gradient information according to the i+1th initial parameter value for the i+1th iteration; When the difference between the i-th gradient information and the (i+1)-th gradient information is less than a preset gradient threshold, an i-th initial parameter value corresponding to the i-th gradient information is determined as the target parameter value.

8. The method according to claim 6, characterized in that Iteratively adjusting the initial parameter value according to the initial loss value to obtain an adjusted parameter value, and determining the adjusted parameter value as the target parameter value when an iteration condition is satisfied, comprising: During the i-th iteration, using a finite difference algorithm to calculate the i-th gradient information of the i-th initial loss value corresponding to the i-th iteration; Generate an i+1th initial parameter value for an i+1th iteration according to the i-th initial parameter value in the i-th iteration and the i-th gradient information; When the number of iterations i satisfies a preset iteration threshold, the (i+1)th initial parameter value is determined as the target parameter value.

9. The method according to claim 1, characterized in that: The target quantum circuit encoder includes a first encoder and a second encoder; The step of generating a third quantum channel according to the target quantum circuit encoder and the first quantum channel includes: Generate the first encoder and the second encoder respectively according to the target parameter value and the initial quantum circuit encoder; The third quantum channel is generated according to the first encoder, the second encoder and the first quantum channel.

10. A device for generating a quantum circuit autoencoder, characterized in that: include: A first generation module is used to generate an initial quantum circuit encoder and a first quantum channel according to an input quantum channel in response to a compression instruction, wherein each layer in the initial quantum circuit encoder is generated according to a plurality of single-bit noises and parameter-containing quantum circuits; A second generating module is used to generate a second quantum channel according to the initial quantum circuit encoder and the first quantum channel in response to the reconstruction instruction; A third generating module is used to generate a target quantum state according to the second quantum channel and an initial quantum state, wherein the initial quantum state is obtained according to a maximum mixed state and a maximum entangled state; A determination module, configured to iteratively adjust the parameter value of the initial quantum circuit encoder to determine the single-bit noise and the target parameter value of the parameter-containing quantum circuit according to the target loss value of the target quantum state corresponding to the initial quantum circuit encoder; a fourth generating module, configured to generate a third quantum channel according to a target quantum circuit encoder and the first quantum channel, wherein the target quantum circuit encoder is constructed according to the single-bit noise and a target parameter value of the parameter-containing quantum circuit; A fifth generating module, used for generating a potential channel and a noise channel according to the third quantum channel; A sixth generation module is used to generate the reconstructed quantum channel according to the potential channel, the noise channel and a target quantum circuit decoder, wherein the target quantum circuit decoder is the conjugate transpose of the target quantum circuit encoder, and the quantum circuit autoencoder includes the target quantum circuit encoder and the target quantum circuit decoder.

Images