Investment portfolio optimization method and device adaptive to neutral atom quantum computer

Abstract

The invention relates to the technical field of quantum computing, and provides an investment portfolio optimization method and device adaptive to a neutral atom quantum computer, and the method comprises the steps: carrying out the quantum coding of an investment portfolio optimization problem, and constructing a secondary unconstrained binary optimization model; mapping the secondary unconstrained binary optimization model to a two-dimensional atom array of a neutral atom quantum computer, and constructing a native optimization Hamiltonian; based on the native optimized Hamiltonian, constructing a variable component sub-circuit adaptive to a neutral atom quantum computer system structure, and executing the variable component sub-circuit on a neutral atom quantum computer to obtain a quantum measurement result; and executing a quantum-classical hybrid optimization process based on the quantum measurement result to obtain optimization parameters for investment portfolio configuration. According to the method, a large amount of controlled NOT gate and switching operation is avoided, and the circuit complexity is reduced, so that the calculation efficiency, the adaptability and the engineering practicability of quantum calculation in an investment portfolio optimization problem are improved.

Description

Technical Field

[0001] This invention relates to the field of quantum computing technology, and in particular to a portfolio optimization method and apparatus adapted for neutral atom quantum computers. Background Technology

[0002] Portfolio optimization (PO) is a core problem in modern financial engineering, aiming to achieve optimal asset allocation decisions given expected returns and risk constraints. In this field, risk parity models and Conditional Value at Risk (CVaR) models have been widely researched and applied. However, with the increasing complexity of asset classes and markets, these classic computational methods face significant bottlenecks.

[0003] The search space for portfolio optimization problems grows exponentially with the number of assets. As asset size increases, classical algorithms often struggle to provide high-quality approximate solutions within the practically required timeframes (e.g., seconds or minutes), limiting their application in real-time trading or large-scale asset allocation scenarios. Real-world investment processes often involve various complex constraints, such as risk budgets, leverage limits, and transaction costs. These constraints result in a complex, non-convex structure in the solution space of the optimization problem, further increasing the difficulty and computational cost of classical numerical optimization methods.

[0004] In recent years, quantum computing has been introduced to attempt to solve combinatorial optimization problems such as portfolio optimization. Although related research has shown some potential at the theoretical level, existing quantum computing schemes still face significant shortcomings in terms of computational efficiency and system adaptability in practical joint investment applications.

[0005] Specifically, as the number of assets or constraints increases, the number of quantum operations required in quantum computing grows rapidly, leading to more complex computational processes and a significant increase in execution time and resource consumption. Under current quantum computing equipment conditions, this type of method often struggles to maintain stable computational performance in practical operation, failing to meet the computational efficiency and reliability requirements of joint investment scenarios. This can easily result in a significant increase in the number of quantum operations required in quantum computing, deeper quantum circuits, and lower overall computational efficiency, making it difficult to meet the speed and stability requirements of actual investment decisions.

[0006] Therefore, overcoming the shortcomings of the existing technology is an urgent problem to be solved in this technical field. Summary of the Invention

[0007] The technical problem this invention aims to solve is that when using quantum computers for joint investment calculations, as the asset size or constraints increase, there are still problems of low computational efficiency and insufficient adaptability, making it difficult to meet the actual requirements of portfolio optimization scenarios for computational efficiency and stability.

[0008] To achieve the above objectives, in a first aspect, the present invention provides a portfolio optimization method adapted for neutral atom quantum computers, the method comprising the following steps: Obtain the portfolio optimization problem input by the user, encode the portfolio optimization problem using quantum encoding, and construct a quadratic unconstrained binary optimization model; The quadratic unconstrained binary optimization model is mapped to a two-dimensional atom array of a neutral atom quantum computer, and a native optimized Hamiltonian is constructed based on the atomic layout obtained by mapping and the native interactions of the neutral atom quantum computer. Based on the native optimized Hamiltonian, a variable quantum circuit adapted to the neutral atom quantum computer architecture is constructed, and the variable quantum circuit is executed on the neutral atom quantum computer to obtain quantum measurement results; Based on the quantum measurement results, a quantum-classical hybrid optimization process is performed to obtain optimized parameters for portfolio allocation, thereby providing users with recommended optimal investment solutions.

[0009] Furthermore, the process of quantum-encoding the portfolio optimization problem to construct a quadratic unconstrained binary optimization model includes: An objective function for portfolio optimization is established based on the expected returns, covariance matrix, risk budget constraints, leverage constraints, and / or transaction cost constraints of each asset in the portfolio; and the objective function for portfolio optimization is expressed as a quadratic unconstrained binary optimization model. Each asset is represented as a binary variable, the risk coupling relationship between assets is represented as a two-body term in a quadratic unconstrained binary optimization model, and the constraint term reflecting the portfolio constraint relationship is represented as a penalty term in a quadratic unconstrained binary optimization model. The two-body terms in the aforementioned quadratic unconstrained binary optimization model are configured to achieve coupling through the native interaction of the Rydberg blocking phase in a neutral atom quantum computer, thus obtaining a quadratic unconstrained binary optimization model suitable for implementation in a neutral atom quantum computer.

[0010] Further, the step of mapping the quadratic unconstrained binary optimization model to a two-dimensional atom array of a neutral atom quantum computer, and constructing a native optimized Hamiltonian based on the mapped atom layout and the native interactions of the neutral atom quantum computer, includes: Based on the coefficients of the two-body terms in the aforementioned quadratic unconstrained binary optimization model, an asset correlation diagram reflecting the coupling relationship between assets is constructed. Based on the asset correlation diagram and the Rydberg blocking radius of the neutral atom quantum computer, the spatial position of the physical atoms corresponding to each asset in the two-dimensional atom array is determined, and a physical layout diagram is generated, such that the physical atoms corresponding to the assets that are related in the asset correlation diagram are adjacent to each other. Based on the spatial positional relationships of physical atoms defined in the physical layout diagram, and combined with the Rydberg blocking native interaction in the neutral atom quantum computer, a native optimized Hamiltonian corresponding to the two-body term in the quadratic unconstrained binary optimization model is constructed.

[0011] Furthermore, the construction of the native optimized Hamiltonian corresponding to the two-body term in the quadratic unconstrained binary optimization model, based on the spatial positional relationships of the physical atoms defined by the physical layout diagram and combined with the Rydberg blocking native interaction in the neutral atom quantum computer, includes: Based on the physical arrangement diagram, pairs of physical atoms capable of interacting through the Rydberg blockade effect are identified; A native optimized Hamiltonian is constructed based on the Ising interaction and / or XY interaction of the neutral atom quantum computer. The strength of the Ising interaction and / or XY interaction between the paired physical atoms is set by adjusting at least one physical control parameter, including laser intensity, laser detuning, and interatomic spacing.

[0012] Furthermore, the portfolio optimization method further includes: The two-dimensional atom array is rearranged at the bit level using the atomic rearrangement capability of the neutral atom quantum computer; the spatial positions of the physical atoms are then rearranged globally or in groups based on the physical layout diagram. This is to ensure that the physical atoms corresponding to qubits with coupling coefficients greater than a set threshold in the second-order unconstrained binary optimization model are geometrically close to each other.

[0013] Furthermore, the step of constructing a variable quantum circuit adapted to the neutral atom quantum computer architecture based on the native optimized Hamiltonian, and executing the variable quantum circuit on the neutral atom quantum computer to obtain quantum measurement results includes: A variable quantum circuit is constructed, comprising alternating single-qubit gate layers and two-qubit interaction layers; wherein, the operation of the single-qubit gate layer includes Rx gate and Ry gate operations, and the two-qubit interaction layer is composed of the native XY interaction and local Z rotation of the neutral atom quantum computer; The variable quantum circuit is driven by a parallel laser pulse sequence to run on the neutral atom quantum computer and perform quantum state measurements, thereby obtaining the probability distribution of the quantum state as the quantum measurement result.

[0014] Furthermore, the parallel laser pulse sequence is generated by a pulse-level compiler; wherein the pulse-level compiler maps the adjustable parameters in the variable quantum circuit to at least one pulse control parameter, the pulse control parameter including at least one of pulse width, pulse intensity, laser detuning and phase.

[0015] Furthermore, in the process of performing a quantum-classical hybrid optimization process based on the quantum measurement results to obtain the optimization parameters for portfolio configuration, the quantum measurement results are first subjected to quantum statistical filtering before the quantum-classical hybrid optimization is performed; the quantum-classical hybrid optimization process includes at least one of a stochastic approximate gradient optimization algorithm, a Bayesian optimization algorithm, and an optimization algorithm based on second-order information.

[0016] In a second aspect, the present invention also provides a portfolio optimization apparatus adapted to a neutral atom quantum computer, for implementing the portfolio optimization method adapted to a neutral atom quantum computer described in the first aspect, the apparatus comprising: At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor for performing the portfolio optimization method for adapting a neutral atom quantum computer as described in the first aspect.

[0017] Thirdly, the present invention also provides a non-volatile computer storage medium storing computer-executable instructions that are executed by one or more processors to perform the method described in the first aspect and / or the method described in the second aspect.

[0018] Unlike existing technologies, this invention can construct a native optimization Hamiltonian by encoding the portfolio optimization problem into a quadratic unconstrained binary optimization model and directly mapping it to a two-dimensional array of neutral atoms. By combining the Deburg blocking native interaction with a quantum-classical hybrid optimization mechanism, it avoids a large number of controlled NOT gates and commutation operations, reduces circuit complexity, and thus improves the computational efficiency, adaptability, and engineering practicality of quantum computing in portfolio optimization problems. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments of the present invention will be briefly described below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0020] Figure 1 This is a flowchart of the portfolio optimization method provided in Embodiment 1; Figure 2 This is a flowchart of the construction of a quadratic unconstrained binary optimization model for the portfolio optimization method provided in Embodiment 1; Figure 3 This is a flowchart of the construction of the native optimized Hamiltonian in the portfolio optimization method provided in this embodiment; Figure 4 This is a flowchart of the portfolio optimization method provided in Embodiment 1, which constructs the native optimized Hamiltonian based on the physical layout diagram; Figure 5 This is a flowchart illustrating the process of obtaining quantum measurement results using the portfolio optimization method provided in Embodiment 1. Figure 6 This is a physical arrangement diagram of the two-dimensional atomic array in the portfolio optimization method provided in this embodiment; Figure 7 This is an architectural diagram of the apparatus for the portfolio optimization method provided in Embodiment 2. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0022] In this invention, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0023] Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0024] Example 1: In this embodiment, key terms are defined as follows: A qubit refers to the basic logical unit of quantum information processing; a physical atom specifically refers to the physical entity that carries the function of a qubit in the two-dimensional optical tweezers array of the neutral atom quantum computer adapted to this invention; unless otherwise specified, the qubit mentioned in this embodiment refers to the physical carrier realized by the aforementioned physical atoms, and the two form a one-to-one correspondence. This definition aims to clarify the unity between logical concepts and their physical implementations, and the following descriptions are based on this understanding. Portfolio optimization problems typically involve selecting or allocating among multiple assets and are affected by returns, risks, and constraints. Since such problems can be abstracted into combinatorial optimization in a discrete decision variable space, they are suitable for solving using quantum computing.

[0025] Reference Figure 1 As shown, this invention proposes a portfolio optimization method adapted for neutral atom quantum computers, the method comprising: Step 10: Obtain the portfolio optimization problem input by the user, perform quantum encoding on the portfolio optimization problem, and construct a quadratic unconstrained binary optimization model.

[0026] The Quadratic Unconstrained Binary Optimization (QUBO) model is a mathematical framework for modeling combinatorial optimization problems. Its variables are binary (0 or 1), the objective function is a quadratic function, and there are no constraints (constraints can be incorporated into the objective function through penalty terms). Quantum encoding the portfolio optimization problem into a Quadratic Unconstrained Binary Optimization model allows complex constraints to be uniformly incorporated into the objective function, thereby reducing the problem's structural complexity and providing a standardized input form for subsequent quantum computing.

[0027] Step 20: Map the quadratic unconstrained binary optimization model to a two-dimensional atom array of a neutral atom quantum computer, and construct a native optimized Hamiltonian based on the atom layout obtained by the mapping and the native interactions of the neutral atom quantum computer.

[0028] The neutral atom quantum computer described above is a quantum computing system that uses neutral atoms trapped in an optical trap as qubits and leverages Rydberg interactions to achieve coupling between qubits. It possesses a naturally reconfigurable two-dimensional or multi-dimensional array structure and high scalability. The two-dimensional atomic array refers to the qubit topology formed by arranging multiple neutral atoms according to a two-dimensional planar regular or irregular geometric structure in the neutral atom quantum computer. The logical variables and their interactions in the quadratic unconstrained binary optimization model are mapped to the spatial arrangement and interactions of atoms in the two-dimensional atomic array. The native interactions of the neutral atom quantum computer refer to the interaction forms naturally generated by physical mechanisms, typically including effective two-body or multi-body interactions based on the Rydberg blocking effect. Mapping the quadratic unconstrained binary optimization model to the two-dimensional atomic array of the neutral atom quantum computer, and combining it with the native interactions of the neutral atom quantum computer to construct a native optimization Hamiltonian, ensures a high degree of matching between the mathematical structure of the optimization problem and the physical characteristics of the hardware. This improves the efficiency of quantum resource utilization, thereby enhancing the computational efficiency and architectural adaptability of quantum computing in portfolio optimization problems, laying the foundation for the efficient execution of subsequent quantum circuits.

[0029] Step 30: Based on the native optimized Hamiltonian, construct a variable quantum circuit adapted to the neutral atom quantum computer architecture, and execute the variable quantum circuit on the neutral atom quantum computer to obtain quantum measurement results.

[0030] Variable quantum circuits are the core carriers for executing quantum approximation optimization algorithms. Their structure consists of multiple layers of parameterized quantum gates with alternating interactions. By adjusting the circuit parameters, the measurement results of a neutral atom quantum computer are gradually made closer to the lowest energy state of the Hamiltonian, improving the feasibility of solving large-scale combinatorial optimization problems.

[0031] Step 40: Based on the quantum measurement results, perform a quantum-classical hybrid optimization process to obtain optimized parameters for portfolio configuration, so as to provide users with recommended optimal investment solutions.

[0032] The quantum-classical hybrid optimization described above is a computational approach feasible on current medium-scale quantum devices with noise. It uses quantum and classical computers in tandem: the quantum part prepares trial states and measures the objective function (such as energy or cost) through parameterized quantum circuits, while the classical part uses optimization algorithms (such as gradient descent) to iteratively update parameters based on the measurement results, thereby approximating the optimal solution. In this embodiment, the optimization parameters are used to characterize the allocation results of each portfolio problem. These optimization parameters are determined by the quantum measurement results obtained during the quantum-classical hybrid optimization process and represent the allocation ratio of each portfolio. Based on these optimization parameters, recommended results for portfolio allocation can be generated, thereby providing users with corresponding portfolio solutions.

[0033] In this scheme, a quantum-classical hybrid optimization process is performed using quantum measurement results to achieve iterative updates of variational parameters, thereby obtaining optimization parameters for portfolio allocation. This approach combines the parallel advantages of quantum computing with the stability of classical algorithms, thus improving the stability and engineering practicality of the portfolio optimization problem-solving process under the condition of limited quantum computing resources.

[0034] This invention represents the portfolio optimization problem as a quadratic unconstrained binary optimization model, further mapping it to a two-dimensional atom array of a neutral atom quantum computer. By combining the native interactions of the neutral atom platform of the neutral atom quantum computer to construct an optimized Hamiltonian, the optimization problem can be implemented on a neutral atom quantum computer. By constructing a variable quantum circuit adapted to the architecture of the neutral atom quantum computer based on the native optimized Hamiltonian, the dependence on complex non-native quantum gate operations is avoided, making the variable quantum circuit more suitable for execution under the actual operating conditions of the neutral atom quantum computer. This invention, through a combination of quantum encoding, physical mapping, variational circuit execution, and hybrid optimization, forms a portfolio optimization method that can be implemented on a neutral atom quantum computer, thereby improving the solution efficiency and system adaptability of quantum computing in portfolio optimization problems, and providing a feasible technical path for the application of neutral atom quantum computers in the field of financial optimization.

[0035] In another embodiment, refer to Figure 2 As shown, step 10 includes: Step 101: Establish an objective function for portfolio optimization based on the expected returns, covariance matrix, risk budget constraints, leverage constraints, and / or transaction cost constraints of each asset in the portfolio; and express the objective function for portfolio optimization as a quadratic unconstrained binary optimization model.

[0036] In this embodiment, the portfolio optimization objective function is established based on user-defined investment parameters and constraints. These user-defined parameters may include the expected returns of each asset, risk budget constraints, leverage constraints, and / or transaction cost constraints. Based on these parameters and constraints, a portfolio optimization objective function is established and expressed as a quadratic unconstrained binary optimization model. Through this setup, the complex optimization problem, which originally involved multiple constraints, is transformed into a standardized unconstrained optimization form, reducing the complexity of problem modeling and providing a unified mathematical framework for subsequent quantum computing solutions.

[0037] In this embodiment, the classic Markowitz mean-variance model is employed, and an L1 regularization term is introduced to control portfolio sparsity (i.e., selecting fewer assets). The Markowitz mean-variance model is the foundation of modern portfolio theory, determining optimal asset allocation by minimizing portfolio risk (measured by return variance) given expected return, or maximizing expected return given risk. L1 regularization adds the sum of the absolute values ​​of weights as a penalty term to the optimization model to induce sparsity, making some weights exactly zero, thereby achieving automatic selection of features or assets and improving the model's simplicity. With the above settings, the portfolio optimization objective function can be expressed as: ; Where x is a binary decision vector, and each element xi∈{0,1} represents whether to invest in the i-th asset (1 for investment, 0 for no investment); Σx is the covariance matrix of the assets, used to quantify risk; x Σx represents a two-body term, corresponding to the overall risk of the investment portfolio. μ x is a linear term, corresponding to the expected return. λ is the L1 regularization term, μ is the expected return vector of the assets, and λ is the L1 regularization coefficient, which is used to control the sparsity of the portfolio. The larger the value of λ, the more inclined it is to select fewer assets.

[0038] In this embodiment, each asset is mapped to a qubit, meaning the selection state of asset i is represented by the state of qubit i (1 for investment, 0 for no investment). The two-body term x in the objective function... Σx, representing the overall risk of the portfolio, will be encoded as a two-body term between qubits; a linear term. μ x corresponds to the expected return, which will be encoded as a local field acting on a single qubit; As a constraint, it can also be incorporated into the objective function in the form of a penalty term. Through the above setup, the constrained portfolio optimization problem is transformed into a quadratic unconstrained binary optimization model with binary variables as independent variables.

[0039] Step 102: Represent each asset as a binary variable, the risk coupling relationship between assets as a two-body term in a quadratic unconstrained binary optimization model, and the constraint term reflecting portfolio constraints as a penalty term in the same model. By representing each asset as a binary variable and the risk coupling relationship between assets as a two-body term, while introducing the constraint term reflecting portfolio constraints as a penalty term, asset selection decisions, asset correlations, and constraints are uniformly characterized within the same optimization model. This avoids additional processing of complex constraints during quantum computing and improves the adaptability of the quadratic unconstrained binary optimization model to quantum optimization algorithms.

[0040] Step 103: The two-body terms in the quadratic unconstrained binary optimization model are configured to achieve coupling through the native interaction of the Rydberg blocking phase in a neutral atom quantum computer, thereby obtaining a quadratic unconstrained binary optimization model suitable for implementation in a neutral atom quantum computer.

[0041] By strategically configuring the two-body terms in the quadratic unconstrained binary optimization model, it is possible to achieve asset coupling through the Rydberg blocking effect—a native interaction in neutral atom quantum computers. This ensures that the constructed quadratic unconstrained binary optimization model physically matches the native interaction mechanism of neutral atom quantum computers, reducing the additional quantum gate operations required in the conversion from logical model to physical implementation, and lowering circuit depth and implementation complexity. Through this configuration, not only is the portfolio optimization problem effectively encoded in quantum computing, but the Rydberg blocking native interaction characteristic of neutral atom quantum computers is also fully utilized, establishing a one-to-one correspondence between the optimization model's mathematical structure and physical implementation, thereby improving the algorithm's execution efficiency on neutral atom quantum computers.

[0042] In a preferred embodiment, refer to Figure 3 and Figure 6 As shown, step 20 includes: Step 201: Based on the coefficients of the two-body terms in the quadratic unconstrained binary optimization model, construct an asset correlation diagram that reflects the coupling relationship between assets.

[0043] In this graph, nodes correspond to assets in the investment portfolio, and edges and their weights characterize the strength of risk coupling between assets. The asset correlation graph provides an intuitive representation of the risk coupling relationships between assets in a graph structure, thus offering a clear structural basis for subsequent physical mapping and avoiding the implementation complexity associated with arranging complex coupling relationships between assets using quantum gates one by one.

[0044] Step 202: Based on the asset correlation map and the Rydberg blocking radius of the neutral atom quantum computer, determine the spatial position of the physical atoms corresponding to each asset in the two-dimensional atom array, and generate a physical layout map such that the physical atoms corresponding to the assets that are related in the asset correlation map are adjacent to each other.

[0045] Rydberg blocking is a quantum effect: when an atom is excited to a highly excited Rydberg state, its large electric dipole moment strongly alters the energy level structure of surrounding atoms, thus preventing other nearby atoms from being simultaneously excited to the Rydberg state. This effect is widely used in neutral atom quantum computing to realize high-fidelity multi-qubit entanglement gates and quantum simulations. In this embodiment, the physical atoms corresponding to assets with strong risk coupling relationships in the asset correlation graph are arranged within a neighborhood that satisfies the Rydberg blocking condition, thereby mapping the logical asset coupling relationship to an effective interaction relationship between atoms at the physical level.

[0046] Step 203: Based on the spatial positional relationship of physical atoms defined in the physical layout diagram, and combined with the Rydberg blocking native interaction, construct the native optimized Hamiltonian corresponding to the two-body term in the quadratic unconstrained binary optimization model.

[0047] The Rydberg blocking native interaction refers to the phenomenon where, when an atom is excited to a Rydberg state with a high principal quantum number, its electron orbital radius increases significantly, leading to extremely strong dipole-dipole or van der Waals interactions between atoms. This mechanism requires no additional coupling structure and is the intrinsic physical basis for achieving controlled quantum logic operations and many-body entanglement in neutral atom systems. It can be directly used to construct quantum gates or simulate strongly correlated many-body systems. The portfolio optimization objective function can be directly achieved through the native interactions of a neutral atom quantum computer, reducing dependence on non-native quantum gates or multi-layer quantum gate decomposition, thereby reducing quantum circuit depth and improving overall operational fidelity.

[0048] An asset correlation graph is constructed using a quadratic unconstrained binary optimization model. Based on this graph, the spatial layout of the physical atoms corresponding to the assets in a reconfigurable two-dimensional atom array is planned, generating a physical layout diagram. This ensures that the physical atoms corresponding to assets with relatively large coupling coefficients are arranged within the Rydberg blocking radius, thus reflecting the risk coupling relationships between assets in the quadratic unconstrained binary optimization model at the physical level. This helps to correlate the abstract optimization model with the native interaction mechanism of a neutral atom quantum computer, reducing the complexity caused by the mismatch between model structure and physical implementation during problem mapping. Simultaneously, by selectively adjusting the spatial layout of the atom array, the main coupling relationships are reflected within the neighborhood satisfying the Rydberg blocking condition, which helps to reduce ineffective interactions between non-critical assets and improves the targeting and controllability of subsequent Hamiltonian construction and quantum computing processes.

[0049] In a preferred embodiment, refer to Figure 4 As shown, step 203 further includes: Step 2031: Based on the physical layout diagram, determine the pairs of physical atoms that can generate interactions through Rydberg blocking primary interactions.

[0050] Step 2032: Construct the native optimized Hamiltonian based on the Ising interaction and / or XY interaction of the neutral atom quantum computer.

[0051] The Ising interaction is a form of interaction describing state-dependent energy coupling between two qubits. For paired qubits, the system energy changes with the coupling strength when they are in the same or different quantum states, thus forming an energy preference or penalty for a specific combination of states. In neutral atom quantum computers, the Ising interaction can be realized through the blocking effect or effective interaction between Rydberg excited states. Its interaction strength can be adjusted by physical control parameters such as laser intensity, laser detuning, and atomic spacing, making it suitable for expressing two-body coupling terms in quadratic unconstrained binary optimization models. The XY interaction is a form of interaction describing state exchange behavior between paired qubits. Under this interaction, excited states can transfer between different qubits, thereby achieving quantum state mixing and diffusion. In neutral atom quantum computers, the XY interaction can be realized through resonant dipole-dipole interactions or equivalent primordial interaction mechanisms between Rydberg excited states. Its interaction strength can also be adjusted by physical control parameters such as laser intensity, laser detuning, and atomic spacing, making it suitable as a mixing interaction in variable quantum algorithms. In this embodiment, the two-body term in the quadratic unconstrained binary optimization model directly corresponds to the interaction forms (Ising interaction and / or XY interaction) that can be natively realized under the neutral atom quantum computer architecture, thereby avoiding mapping the optimization problem to a non-native Hamiltonian that requires a large number of composite quantum gate decompositions and reducing the structural complexity of the native optimization Hamiltonian at the physical implementation level.

[0052] Step 2033: The strength of the Ising interaction and / or XY interaction between the paired physical atoms is set by adjusting at least one physical control parameter, including laser intensity, laser detuning amount and interatomic spacing.

[0053] In this embodiment, without introducing additional logical quantum gates or complex gate sequences, the two-body coupling relationship in the quadratic unconstrained binary optimization model is realized, adjustable, and precisely expressed within the framework of the native interaction of a neutral atom quantum computer. This helps to reduce the overall implementation complexity of the quantum computing process and improve the stability and reliability of the optimization results.

[0054] In a preferred embodiment, the portfolio optimization method further includes: The two-dimensional atom array is rearranged at the bit level using the atomic rearrangement capability of the neutral atom quantum computer; the spatial positions of the physical atoms are rearranged globally or grouped according to the physical layout diagram.

[0055] This is to ensure that the physical atoms corresponding to qubits with coupling coefficients greater than a set threshold in the second-order unconstrained binary optimization model are geometrically close to each other.

[0056] The above settings can reduce the invalid interactions or redundant scheduling requirements introduced by the mismatch between the spatial positions of logical qubits and physical atoms, thereby reducing the dependence on additional quantum operations in the subsequent Hamiltonian construction and variable quantum circuit execution, and improving the overall execution efficiency.

[0057] In a preferred embodiment, refer to Figure 5 As shown, step 30 includes: Step 301: Construct a variable quantum circuit, which includes alternating single-qubit gate layers and two-qubit interaction layers; wherein the operation of the single-qubit gate layer includes Rx gate and Ry gate operations, and the two-qubit interaction layer is composed of the native XY interaction and local Z rotation of the neutral atom quantum computer.

[0058] Variable quantum circuits refer to quantum circuit structures with adjustable parameters. They utilize parameterized quantum operations performed on a quantum computer, combined with classical optimization algorithms to iteratively update these parameters, gradually approximating the optimal solution to the target optimization problem through quantum measurement. An Rx gate is a single-qubit rotation gate used to rotate the quantum state of a qubit around the X-axis by a predetermined angle, where the rotation angle is an adjustable parameter. A Ry gate is a single-qubit rotation gate used to rotate the quantum state of a qubit around the Y-axis by a predetermined angle, where the rotation angle is also an adjustable parameter. Ry gates can modulate the quantum state in different rotation directions than Rx gates, allowing for more complete parameterization of the qubit's quantum state.

[0059] By simultaneously introducing Rx and Ry gate operations into a single-qubit gate layer, the quantum state can be tuned across multiple degrees of freedom, thereby improving the expressive power and flexibility of variable quantum circuits in the portfolio optimization solution space. Employing a single-qubit gate layer containing both Rx and Ry gates in the variable quantum circuit allows for the parameterized modulation of the quantum state to coordinate with the subsequent two-qubit interaction layer, which is beneficial for constructing variable quantum circuits within the architecture of neutral atom quantum computers. This avoids achieving two-qubit interactions through multi-level quantum logic gate combinations, thus reducing the overall circuit depth of the variable quantum circuit and minimizing accumulated errors.

[0060] In this embodiment, a Rydberg-QAOA circuit is used as the implementation of a variable quantum circuit. This circuit consists of alternating single-qubit gate layers and two-qubit interaction layers. The single-qubit gate layers operate on Rx and Ry gates, with mathematical forms corresponding to evolutions based on the driving Hamiltonian. The two-qubit interaction layers consist of the native XY interaction and local Z rotation of the neutral atom quantum computer, with mathematical forms corresponding to evolutions based on the native optimized Hamiltonian.

[0061] A Rydberg-QAOA circuit is used as a specific embodiment of the variable quantum circuit. This circuit consists of L alternating evolution layers, each layer comprising: By adjustable parameter γ k The controlled local Z-rotation operation has the following mathematical form: ; By adjustable parameter β k The controlled operations based on the native XY interaction of the neutral atom quantum computer are mathematically expressed as follows: .

[0062] On the neutral atom quantum computer, the circuit is executed by a sequence of laser pulses applied in parallel, thereby enabling the evolution and manipulation of quantum states.

[0063] Step 302: The variable quantum circuit is driven to run on the neutral atom quantum computer using a parallel laser pulse sequence, and quantum state measurements are performed to obtain the probability distribution of the quantum state as the quantum measurement result.

[0064] By driving the variable quantum circuit with a laser pulse sequence to run on a neutral atom quantum computer, the single-qubit gate layer and the two-qubit interaction layer can be implemented in parallel on multiple physical atoms, thereby improving the parallelism and execution efficiency of quantum operations and making better use of the parallel control capability of the neutral atom quantum computer.

[0065] In another embodiment, the parallel laser pulse sequence is generated by a pulse-level compiler; wherein the pulse-level compiler maps the adjustable parameters in the variable quantum circuit to at least one pulse control parameter, the pulse control parameter including at least one of pulse width, pulse intensity, laser detuning, and phase.

[0066] The pulse-level compiler refers to a compilation module used to convert abstract operational parameters at the quantum circuit level into underlying physical control signals for a quantum computer. Its output is a sequence of laser pulses that can directly drive the execution of a neutral atom quantum computer. By employing a pulse-level compiler, the generated laser pulse sequences can be uniformly scheduled and optimized while meeting the physical constraints of the neutral atom quantum computer. These sequences correspond to parameterized rotation operations in the single-qubit gate layer and native interaction control in the two-qubit interaction layer, respectively, thus avoiding execution deviations caused by abstracting and compiling quantum circuits only at the logic quantum gate level. This configuration ensures that the variable quantum circuits conform to the native operational characteristics and physical constraints of the neutral atom quantum computer during actual operation, improving the consistency and repeatability of quantum operations and enhancing the effective physical response capability of parameter updates during quantum-classical hybrid optimization.

[0067] In another embodiment, the quantum-classical hybrid optimization process based on the quantum measurement results to obtain optimization parameters for portfolio configuration first performs quantum statistical filtering on the quantum measurement results before performing the quantum-classical hybrid optimization; the quantum-classical hybrid optimization process includes at least one of stochastic approximate gradient optimization algorithm, Bayesian optimization algorithm, and second-order information-based optimization algorithm.

[0068] The stochastic approximate gradient optimization algorithm is an optimization algorithm that approximates the parameter update direction through a stochastic method when precise gradient information cannot be directly obtained. It is suitable for iterative optimization of variational parameters under conditions where statistical noise exists in quantum measurement results. The Bayesian optimization algorithm is a parameter search method based on a probabilistic model. It uses existing quantum measurement results to statistically model the objective function and selects new parameter combinations for quantum circuit execution accordingly. It is suitable for improving parameter search efficiency under conditions where quantum computing resources are limited. The optimization algorithm based on second-order information refers to an optimization algorithm that introduces curvature information of the objective function with respect to parameter changes during the parameter optimization process to guide the parameter update direction and step size. It is suitable for accelerating parameter convergence and improving optimization stability when approaching the optimal solution region.

[0069] By performing quantum statistical filtering on the quantum measurement results before the quantum-classical hybrid optimization process begins, the random fluctuations introduced by the finite number of samplings and hardware noise during quantum measurement can be effectively reduced, providing a more stable statistical input for subsequent parameter optimization. Based on this, by employing at least one of the following algorithms—including stochastic approximate gradient optimization, Bayesian optimization, and second-order information-based optimization—the quantum-classical hybrid optimization process can flexibly select parameter update strategies according to different optimization stages and quantum measurement characteristics. Specifically, stochastic approximate gradient optimization is suitable for robust parameter search under noisy conditions, Bayesian optimization is beneficial for improving parameter search efficiency under finite quantum measurement budgets, and second-order information-based optimization helps accelerate parameter convergence when approaching the optimal solution.

[0070] In this embodiment, the above settings improve the stability and robustness of the quantum-classical hybrid optimization process under the premise of limited quantum computing resources, thereby facilitating the acquisition of optimization parameters for portfolio allocation.

[0071] Example 2: like Figure 7 The diagram shown is an architectural schematic of a portfolio optimization device adapted to a neutral atom quantum computer according to an embodiment of the present invention. This portfolio optimization device, adapted to a neutral atom quantum computer, includes one or more processors 21 and a memory 22. Figure 7 Take a processor 21 as an example.

[0072] Processor 21 and memory 22 can be connected via a bus or other means. Figure 7 Taking the example of a connection between China and Israel via a bus.

[0073] Memory 22, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs and non-volatile computer-executable programs, such as the portfolio optimization method for adapting a neutral atom quantum computer in Embodiment 1. Processor 21 executes the method by running the non-volatile software programs and instructions stored in memory 22.

[0074] Memory 22 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, memory 22 may optionally include memory remotely located relative to processor 21, which can be connected to processor 21 via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0075] The program instructions / modules are stored in the memory 22. When executed by one or more processors 21, they perform the method described in Embodiment 1 above, for example, the method described above. Figure 1-5 The steps shown.

[0076] Those skilled in the art will readily understand that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A portfolio optimization method adapted for neutral atom quantum computers, characterized in that, The portfolio optimization method includes: Obtain the portfolio optimization problem input by the user, encode the portfolio optimization problem using quantum encoding, and construct a quadratic unconstrained binary optimization model; The quadratic unconstrained binary optimization model is mapped to a two-dimensional atom array of a neutral atom quantum computer, and a native optimized Hamiltonian is constructed based on the atomic layout obtained by mapping and the native interactions of the neutral atom quantum computer. Based on the native optimized Hamiltonian, a variable quantum circuit adapted to the neutral atom quantum computer architecture is constructed, and the variable quantum circuit is executed on the neutral atom quantum computer to obtain quantum measurement results; Based on the quantum measurement results, a quantum-classical hybrid optimization process is performed to obtain optimized parameters for portfolio allocation, thereby providing users with recommended optimal investment solutions.

2. The portfolio optimization method according to claim 1, characterized in that, The process of quantum encoding the portfolio optimization problem and constructing a quadratic unconstrained binary optimization model includes: An objective function for portfolio optimization is established based on the expected returns, covariance matrix, risk budget constraints, leverage constraints, and / or transaction cost constraints of each asset in the portfolio; and the objective function for portfolio optimization is expressed as a quadratic unconstrained binary optimization model. Each asset is represented as a binary variable, the risk coupling relationship between assets is represented as a two-body term in a quadratic unconstrained binary optimization model, and the constraint term reflecting the portfolio constraint relationship is represented as a penalty term in a quadratic unconstrained binary optimization model. The two-body terms in the aforementioned quadratic unconstrained binary optimization model are configured to achieve coupling through the native interaction of the Rydberg blocking phase in a neutral atom quantum computer, thus obtaining a quadratic unconstrained binary optimization model suitable for implementation in a neutral atom quantum computer.

3. The portfolio optimization method according to claim 2, characterized in that, The step of mapping the quadratic unconstrained binary optimization model to a two-dimensional atom array of a neutral atom quantum computer, and constructing a native optimized Hamiltonian based on the mapped atom layout and the native interactions of the neutral atom quantum computer, includes: Based on the coefficients of the two-body terms in the aforementioned quadratic unconstrained binary optimization model, an asset correlation diagram reflecting the coupling relationship between assets is constructed. Based on the asset correlation diagram and the Rydberg blocking radius of the neutral atom quantum computer, the spatial position of the physical atoms corresponding to each asset in the two-dimensional atom array is determined, and a physical layout diagram is generated, such that the physical atoms corresponding to the assets that are related in the asset correlation diagram are adjacent to each other. Based on the spatial positional relationships of physical atoms defined in the physical layout diagram, and combined with the Rydberg blocking native interaction in the neutral atom quantum computer, a native optimized Hamiltonian corresponding to the two-body term in the quadratic unconstrained binary optimization model is constructed.

4. The portfolio optimization method according to claim 3, characterized in that, The spatial relationships of physical atoms defined based on the physical layout diagram, combined with the Rydberg blocking primary interaction in the neutral atom quantum computer, construct the primary optimized Hamiltonian corresponding to the two-body term in the quadratic unconstrained binary optimization model, including: Based on the physical arrangement diagram, pairs of physical atoms capable of interacting through the Rydberg blockade effect are identified; A native optimized Hamiltonian is constructed based on the Ising interaction and / or XY interaction of the neutral atom quantum computer. The strength of the Ising interaction and / or XY interaction between the paired physical atoms is set by adjusting at least one physical control parameter, including laser intensity, laser detuning, and interatomic spacing.

5. The portfolio optimization method according to claim 3, characterized in that, The portfolio optimization method further includes: The two-dimensional atom array is rearranged at the bit level using the atomic rearrangement capability of the neutral atom quantum computer; the spatial positions of the physical atoms are then rearranged globally or grouped according to the physical layout diagram. This is to ensure that the physical atoms corresponding to qubits with coupling coefficients greater than a set threshold in the second-order unconstrained binary optimization model are geometrically close to each other.

6. The portfolio optimization method according to claim 1, characterized in that, The process of constructing a variable quantum circuit adapted to the neutral atom quantum computer architecture based on the native optimized Hamiltonian, and executing the variable quantum circuit on the neutral atom quantum computer to obtain quantum measurement results includes: A variable quantum circuit is constructed, comprising alternating single-qubit gate layers and two-qubit interaction layers; wherein, the operation of the single-qubit gate layer includes Rx gate and Ry gate operations, and the two-qubit interaction layer is composed of the native XY interaction and local Z rotation of the neutral atom quantum computer; The variable quantum circuit is driven by a parallel laser pulse sequence to run on the neutral atom quantum computer and perform quantum state measurements, thereby obtaining the probability distribution of the quantum state as the quantum measurement result.

7. The portfolio optimization method according to claim 6, characterized in that, The parallel laser pulse sequence is generated by a pulse-level compiler; wherein the pulse-level compiler maps the adjustable parameters in the variable quantum circuit to at least one pulse control parameter, the pulse control parameter including at least one of pulse width, pulse intensity, laser detuning and phase.

8. The portfolio optimization method according to claim 1, characterized in that, The quantum-classical hybrid optimization process based on the quantum measurement results, to obtain the optimization parameters for portfolio configuration, first performs quantum statistical filtering on the quantum measurement results, and then performs the quantum-classical hybrid optimization; the quantum-classical hybrid optimization process includes at least one of the following: stochastic approximate gradient optimization algorithm, Bayesian optimization algorithm, and optimization algorithm based on second-order information.

9. A portfolio optimization device adapted for neutral atom quantum computers, characterized in that, include: At least one processor; And a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the processor for performing the portfolio optimization method for an adapted neutral atom quantum computer as described in any one of claims 1-8.

10. A non-volatile computer storage medium, characterized in that, The computer storage medium stores computer-executable instructions, which are executed by one or more processors to perform the portfolio optimization method for an adapted neutral atom quantum computer as described in any one of claims 1-8.

Images