Physical system optimization method and device, storage medium and electronic equipment

Abstract

The invention discloses a physical system optimization method and device, a storage medium and electronic equipment, and the method comprises the steps: obtaining a design parameter space of a to-be-optimized physical system and a plurality of mutually conflicting performance indexes; mapping the design parameter space into a Riemannian manifold; the inherent symmetry of the physical system in the ideal state is identified, and the trade-off relation among the multiple performance indexes is represented as a symmetry breaking mode according to the inherent symmetry; executing geometric optimization on a Riemannian manifold on the basis of a symmetry breaking mode so as to search a design parameter set meeting target performance requirements while keeping physical constraints; and updating the physical system according to the design parameter set. The optimization efficiency of the physical system can be improved.

Description

Technical Field

[0001] This application relates to the field of physical system optimization technology, specifically to a physical system optimization method, apparatus, storage medium, and electronic device. Background Technology

[0002] The core challenge in designing and optimizing physical systems, especially complex systems involving multiple coupled physical domains (such as electromagnetics, thermal, mechanical, and fluid dynamics), lies in coordinating different and often conflicting performance metrics. This process is known as performance trade-offs or multi-objective optimization. For example, in integrated circuit design, increasing computing speed often leads to increased power consumption and heat dissipation difficulties; in aerospace structural design, reducing weight may weaken structural strength and stiffness. These trade-offs essentially stem from the interactions and constraints between underlying physical laws and are prevalent in various physical systems, from microscopic chips to macroscopic engineering projects.

[0003] Currently, the design and optimization of physical systems mainly rely on general numerical algorithms or domain-specific experience, resulting in low efficiency, no guarantee of convergence, and difficulty in cross-domain applicability in the optimization process. Summary of the Invention

[0004] This application provides a physical system optimization method, apparatus, storage medium, and electronic device, which can improve the optimization efficiency of physical systems.

[0005] In a first aspect, embodiments of this application provide a physical system optimization method, including: Obtain the design parameter space and multiple conflicting performance metrics of the physical system to be optimized; The design parameter space is mapped to a Riemannian manifold; Identify the inherent symmetry of the physical system under ideal conditions, and characterize the trade-offs between multiple performance indices as symmetry breaking modes based on the inherent symmetry. Based on the symmetry breaking mode, geometric optimization is performed on the Riemannian manifold to search for a set of design parameters that meet the target performance requirements while preserving physical constraints. The physical system is updated based on the set of design parameters.

[0006] In the physical system optimization method provided in this application embodiment, identifying the inherent symmetry of the physical system in an ideal state includes: Analyze the physical laws or idealized models followed by the physical system, and determine the invariant transformation group of the physical system that remains unchanged under transformations; Based on the invariant transformation group, the inherent symmetry of the physical system is obtained.

[0007] In the physical system optimization method provided in this application embodiment, the step of characterizing the trade-off relationship between multiple performance indicators as a symmetry breaking mode based on the inherent symmetry includes: Determine the Lie algebra corresponding to the inherent symmetry; Identify, from the Lie algebras, the broken generators that cause conflicts among the multiple performance metrics; Based on the broken generator, the trade-off relationship between the performance indicators is mapped to a symmetry breaking mode.

[0008] In the physical system optimization method provided in this application embodiment, the step of performing geometric optimization on the Riemannian manifold based on the symmetry breaking mode to search for a set of design parameters that satisfy the target performance requirements while maintaining physical constraints includes: Based on the symmetry breaking mode, a modified vector field is constructed to guide the optimization. Based on the modified vector field, a geometric integrator is used to perform iterative optimization on the Riemannian manifold to update the position of the design parameters on the Riemannian manifold while maintaining the physical constraints of the physical system. When the iterative optimization converges, the final design parameter points on the Riemannian manifold are mapped back to the design parameter space to obtain the design parameter set.

[0009] In the physical system optimization method provided in this application embodiment, the step of constructing a correction vector field for guiding optimization based on the symmetry breaking mode includes: The symmetry breaking mode is decomposed into multiple independent breaking components; For each broken component, a corresponding vector field component is constructed, the direction of which points to the partial recovery direction of the symmetry corresponding to the broken component; The modified vector field is obtained by weighted combination of the vector field components.

[0010] In the physical system optimization method provided in this application embodiment, mapping the design parameter space to a Riemannian manifold includes: Based on the performance metrics and the design parameter space, a metric tensor is generated. The design parameter space is mapped to a Riemannian manifold using the metric tensor.

[0011] In the physical system optimization method provided in the embodiments of this application, the inherent symmetry includes at least one of time translation symmetry, spatial translation symmetry, spatial rotational symmetry, gauge symmetry, and scale symmetry.

[0012] Secondly, embodiments of this application provide a physical system optimization apparatus, comprising: The acquisition unit is used to acquire the design parameter space and multiple conflicting performance indicators of the physical system to be optimized. Mapping unit, used to map the design parameter space to a Riemannian manifold; The identification unit is used to identify the inherent symmetry of the physical system under ideal conditions, and to characterize the trade-off relationship between multiple performance indicators as a symmetry breaking mode based on the inherent symmetry. An iterative unit is used to perform geometric optimization on the Riemannian manifold based on the symmetry breaking mode, so as to search for a set of design parameters that meet the target performance requirements while maintaining physical constraints; An update unit is used to update the physical system according to the set of design parameters.

[0013] Thirdly, this application provides a storage medium storing a plurality of instructions adapted for loading by a processor to execute any of the physical system optimization methods described above.

[0014] Fourthly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the physical system optimization method described in any of the preceding claims.

[0015] In summary, the physical system optimization method provided in this application includes obtaining the design parameter space and multiple conflicting performance indicators of the physical system to be optimized; mapping the design parameter space to a Riemannian manifold; identifying the inherent symmetry of the physical system in an ideal state, and characterizing the trade-offs between the multiple performance indicators as symmetry-breaking modes based on the inherent symmetry; performing geometric optimization on the Riemannian manifold based on the symmetry-breaking modes to search for a set of design parameters that satisfy the target performance requirements while maintaining physical constraints; and updating the physical system according to the set of design parameters. This application embodiment can improve the optimization efficiency of physical systems. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram illustrating an application scenario of the physical system optimization method provided in the embodiments of this application.

[0018] Figure 2This is a flowchart illustrating the physical system optimization method provided in the embodiments of this application.

[0019] Figure 3 This is a schematic diagram of the physical system optimization device provided in the embodiments of this application.

[0020] Figure 4 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0021] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0022] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, components, features, and elements with the same names in different embodiments of this application may have the same meaning or different meanings, the specific meaning of which must be determined by its interpretation in that specific embodiment or further in conjunction with the context of that specific embodiment.

[0023] It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0024] In the following description, the use of suffixes such as "module," "part," or "unit" to denote elements is solely for the purpose of illustrative purposes and has no specific meaning in itself. Therefore, "module," "part," or "unit" may be used interchangeably.

[0025] In the description of this application, it should be noted that the terms "upper," "lower," "left," "right," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application. In addition, terms such as "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0026] Currently, the design and optimization of physical systems mainly rely on general numerical algorithms or domain-specific experience, resulting in low efficiency, no guarantee of convergence, and difficulty in cross-domain applicability in the optimization process.

[0027] Based on this, embodiments of this application provide a physical system optimization method, apparatus, storage medium, and electronic device. Specifically, the physical system optimization apparatus can be integrated into an electronic device, which can be a server or a terminal, etc. The terminal can include mobile phones, wearable smart devices, tablets, laptops, and personal computers (PCs), etc., as well as other computers and auxiliary devices. The server can be a single server or a server cluster composed of multiple servers, and can be a physical server or a virtual server.

[0028] For example, such as Figure 1 As shown, the electronic device can acquire the design parameter space and multiple conflicting performance indicators of the physical system to be optimized; map the design parameter space to a Riemannian manifold; identify the inherent symmetry of the physical system in an ideal state, and characterize the trade-offs between multiple performance indicators as symmetry breaking modes based on the inherent symmetry; perform geometric optimization on the Riemannian manifold based on the symmetry breaking modes to search for a set of design parameters that meet the target performance requirements while maintaining physical constraints; and update the physical system based on the set of design parameters.

[0029] The technical solutions shown in this application will be described in detail below through specific embodiments. It should be noted that the order of description of the following embodiments is not intended to limit the priority of the embodiments.

[0030] Please see Figure 2 , Figure 2 This is a flowchart illustrating the physical system optimization method provided in this application embodiment. The specific flow of the physical system optimization method can be as follows: 101. Obtain the design parameter space and multiple conflicting performance indicators of the physical system to be optimized.

[0031] First, physical modeling can be performed on the physical system to be optimized to clarify its composition, working principle and the physical fields involved (such as electromagnetism, mechanics, thermodynamics, quantum systems, etc.).

[0032] Then, all design parameters that can be adjusted independently or jointly to affect the performance of the physical system to be optimized can be identified and listed. These design parameters collectively constitute the design parameter space of the physical system to be optimized. For example, in the field of integrated circuit design, design parameters may include transistor size, bias voltage, inductance and capacitance values, wiring topology, etc. In the field of structural mechanics, design parameters may include beam cross-sectional dimensions, material parameters, support locations, etc. In the field of optics, design parameters may include lens curvature, media refractive index, optical path length, etc.

[0033] In this embodiment, based on the function and design goals of the system to be optimized, a set (usually two or more) of performance indicators that need to be optimized can be defined. These performance indicators are often in conflict, meaning that improving one performance indicator often leads to a decrease in another. This conflict can be confirmed and described through theoretical analysis, simulation data, or empirical knowledge. For example, in the design of RF power amplifiers, gain and efficiency often constitute a pair of conflicting performance indicators.

[0034] In the specific implementation process, interfaces can be established with simulation software (such as circuit simulators and finite element analysis software) or experimental testing platforms so that the corresponding performance indicators can be obtained through simulation calculations or experimental measurements for any given design parameter space.

[0035] 102. Map the design parameter space to a Riemannian manifold.

[0036] In this embodiment, the transformation from the ordinary design parameter space to the Riemannian manifold with a specific geometric structure will be completed. The purpose is to establish a geometric framework that can naturally reflect the sensitivity of performance index changes and the intrinsic relationship between design parameters for the subsequent optimization process.

[0037] First, a metric tensor can be generated based on performance metrics and the design parameter space. In some embodiments, the importance of different directions in the design parameter space can be differentiated based on the gradient or sensitivity information of each performance metric relative to the design parameters, thereby generating the tensor. Specifically, in directions where performance metrics fluctuate drastically with changes in design parameters, the metric tensor is constructed to be assigned a larger "intrinsic distance" or "cost"; while in directions where performance metrics change gradually, a smaller distance is assigned.

[0038] Then, the design parameter space can be mapped to a Riemannian manifold using the metric tensor. Mathematically, this process means that the design parameter space is no longer regarded as an isotropic, flat Euclidean space, but rather as a curved space (i.e., a Riemannian manifold) endowed with the aforementioned specific metric tensor.

[0039] On this Riemannian manifold, the shortest path (geodesic) between two points is determined by the metric tensor, which intuitively reflects the "minimum cost of change" experienced when adjusting from one set of design parameters to another, while taking into account both performance sensitivity and physical constraints. Through this mapping, the design parameter space is elevated to a mathematical object with rich geometric information, which not only encodes the range of values ​​for the design parameters, but more importantly, it encodes the combined effects of performance indicators and physical laws on parameter changes.

[0040] 103. Identify the inherent symmetry of a physical system under ideal conditions, and characterize the trade-offs between multiple performance indicators as symmetry breaking modes based on the inherent symmetry.

[0041] This embodiment aims to transform the conflicting relationships of intuitive performance indicators into a precise mathematical description based on symmetry theory. In its specific implementation, step 103 can be divided into the following two stages: Phase 1: Identifying the inherent symmetries of physical systems under ideal conditions.

[0042] In some embodiments, the physical laws or idealized models followed by the physical system can be analyzed to determine the invariant transformation group of the physical system under transformation; based on the invariant transformation group, the inherent symmetry of the physical system can be obtained.

[0043] This inherent symmetry includes, but is not limited to, time translation symmetry, spatial translation symmetry, spatial rotation symmetry, gauge symmetry, and scale symmetry.

[0044] Specifically, firstly, we can analyze the fundamental physical laws governing the physical system to be optimized, or its idealized model when non-ideal factors and performance indices are ignored. Based on the analysis results, we can determine under what transformation operations the core dynamic equations, constitutive relations, or key physical quantities of the physical system to be optimized remain unchanged. The set of all these transformation operations can constitute a mathematically invariant transformation group (usually a Lie group). For example, for a conservative physical system without explicit time variables, it has invariance under time translation transformations, and the corresponding invariant transformation group is the time translation group; for a physical system that is homogeneous and isotropic in space, it may have invariance under spatial translation or rotation groups; in electromagnetic physical systems, it may exhibit invariance under specific gauge transformation groups.

[0045] Then, based on the identified invariant transformation group, the inherent symmetry of the physical system can be formally defined. This means that all the continuous transformation properties that the physical system to be optimized enjoys in its ideal state, which keep its physical essence unchanged, can be reduced to a series of explicit symmetry operations. These operations collectively characterize the perfection and self-consistency of the physical system to be optimized when it is not constrained by competition from specific performance indicators.

[0046] The second stage: Based on the inherent symmetry, the trade-off relationship between multiple performance indicators is characterized as a symmetry breaking mode.

[0047] In some embodiments, a Lie algebra corresponding to an inherent symmetry can be determined; broken generators that cause conflicts among multiple performance metrics can be identified from the Lie algebra; and based on the broken generators, the trade-offs between performance metrics can be mapped to symmetry breaking patterns.

[0048] Specifically, firstly, the Lie algebra corresponding to the inherent symmetry can be determined. This Lie algebra consists of a series of "infinitesimal generators," each representing a fundamental, minute direction of symmetry transformation. This Lie algebra provides a linearized description of the symmetry locally (i.e., in the parameter space or the tangent space of the state space), serving as a tool for quantitative analysis.

[0049] Next, broken generators are identified from this Lie algebra. Specifically, this involves analyzing which original symmetric transformation operations no longer maintain the invariance of the overall performance of the physical system when the physical system to be optimized is no longer in an ideal state, but rather the design parameters are adjusted to optimize a specific performance metric (such as pursuing maximum gain). More specifically, it involves identifying which generators correspond to transformation directions that improve one performance metric while significantly deteriorating another conflicting performance metric. These identified generators mark the root of the performance metric conflict at the symmetry level, i.e., broken generators.

[0050] Finally, based on broken generators, the trade-offs between performance metrics, which were originally qualitative, can be mapped to quantitative, categorizable symmetry breaking patterns. In this embodiment, each specific scenario's trade-off (e.g., the trade-off between gain and efficiency near a certain operating point) corresponds to a subspace composed of specific broken generators. Through this embodiment, complex multi-objective optimization problems can be systematically deconstructed and classified into different symmetry breaking types, thus providing clear mathematical guidance for the next step of designing targeted optimization strategies (such as attempting to partially restore specific symmetries to balance performance metrics).

[0051] 104. Based on the symmetry breaking mode, perform geometric optimization on the Riemannian manifold to search for a set of design parameters that meet the target performance requirements while maintaining physical constraints.

[0052] It is understood that this embodiment aims to search for optimal design parameters under the theoretical guidance of the geometric constraints and symmetry breaking of Riemannian manifolds. In specific implementation, step 104 may include the following steps: 1041. Based on the symmetry breaking mode, construct a modified vector field to guide the optimization.

[0053] In some embodiments, the symmetry breaking mode can be decomposed into multiple independent breaking components; for each breaking component, a corresponding vector field component is constructed, the direction of which points to the partial recovery direction of the symmetry corresponding to the breaking component; the vector field components are weighted and combined to obtain the modified vector field.

[0054] Specifically, firstly, the symmetry breaking mode can be decomposed into several independent breaking components, each of which corresponds to the degree and direction of a specific type of symmetry being broken (e.g., breaking of time translation symmetry, or breaking of a certain type of gauge symmetry).

[0055] Subsequently, for each independent broken component, a corresponding vector field component can be constructed. The direction of this vector field component is designed to point towards the partial recovery direction of the symmetry corresponding to the broken component. That is, adjusting the physical system in the design parameter space along this direction can alleviate the performance index conflict caused by the breaking of this specific symmetry.

[0056] Finally, based on the design objectives and the emphasis (i.e., preference) of each performance metric, appropriate weights can be assigned to each vector field component, and these weighted combinations can be generated to produce a global modified vector field. This modified vector field can be superimposed on the traditional performance gradient field, providing additional guidance from symmetry theory for the optimization search.

[0057] 1042. Based on the modified vector field, a geometric integrator is used to perform iterative optimization on the Riemannian manifold to update the position of the design parameters on the Riemannian manifold while maintaining the physical constraints of the physical system.

[0058] In this embodiment, the optimization direction of each iteration is not determined solely by the gradient of the traditional objective function.

[0059] Specifically, we can first calculate the gradient vectors corresponding to each performance metric at the current design parameter point (located on the Riemannian manifold). These gradient vectors can be expressed using the metric tensor on the Riemannian manifold, reflecting the steepest direction of performance metric improvement in the local geometry.

[0060] Simultaneously, the value of the corrected vector field at that point can be obtained. This corrected vector field embodies the optimization suggestions based on symmetry breaking analysis, and its direction points to an evolutionary path that can partially restore the broken symmetry, thereby potentially alleviating performance conflicts.

[0061] The final optimization direction is a weighted sum of the directions of the gradient vector and the correction vector field. In some embodiments, the two strategies of "directly pursuing performance metrics" and "improving the trade-off structure through symmetry adjustments" can be balanced by adjusting the weight parameters. This optimization direction is a tangent vector located in the tangent space of the current point of the Riemannian manifold.

[0062] Once the optimization direction is obtained, the position of the design parameters can be updated along that direction on the Riemannian manifold. It is understood that when updating the position of the design parameters, it is essential to ensure that the new design parameter points remain strictly on the surface of the Riemannian manifold, thereby automatically satisfying all physical constraints embedded in the Riemannian manifold structure.

[0063] In this embodiment, a geometric integrator can be used to perform iterative optimization on a Riemannian manifold, thereby updating the position of the design parameters on the Riemannian manifold while maintaining the physical constraints of the physical system.

[0064] This geometric integrator (e.g., a numerical integrator based on exponential mapping, Lie group integration, or projection methods) receives the current design parameter point and tangent vector as input. Instead of performing a simple linear superposition (which typically causes the new point to deviate from the Riemannian manifold), it computes a specific curve (such as a geodesic or geodesic-like curve) on the Riemannian manifold, starting from the current design parameter point, with its initial tangent direction aligned with the initial optimization direction. Subsequently, it moves along this specific curve by a "geometric distance" controlled by the optimization step size, thus accurately calculating the next iteration point. This mathematically ensures that the parameter update path is always constrained within a feasible design subspace (i.e., the Riemannian manifold) defined by physical laws.

[0065] In the embodiments of this application, after each iteration of optimization, the physical system can be re-evaluated (through simulation or model calculation) under new design parameters, performance indicators are refreshed, and symmetry breaking modes can also be re-evaluated or fine-tuned according to the new design parameter points, thereby generating a new round of optimization directions. Hyperparameters such as optimization step size and weight coefficients can be dynamically adjusted using adaptive strategies to achieve a balance between extensive exploration on the Riemannian manifold in the early stage of optimization and fine convergence in the later stage of optimization.

[0066] In this embodiment, by constructing the design parameter space as a Riemannian manifold and using a geometric integrator to perform iterative optimization, satisfying physical constraints no longer depends on adding external penalty terms or complex constraint handling logic to the optimization objective function. Instead, these physical constraints (such as energy conservation laws, equations of motion, constitutive relations, boundary conditions, etc.) are deeply encoded in the definition and geometry of the Riemannian manifold. As long as the optimization iteration process is always carried out on this Riemannian manifold, all intermediate and final solutions generated by the geometric integrator naturally satisfy these physical constraints. This fundamentally guarantees the physical realizability of the optimization scheme and significantly improves the numerical robustness and convergence reliability of the entire optimization process.

[0067] 1043. When the iterative optimization converges, the final design parameter points on the Riemann manifold are mapped back to the design parameter space to obtain the design parameter set.

[0068] Specifically, iterative optimization can be judged as converged based on preset convergence rules (e.g., the relative change of the target performance value is less than a threshold or the maximum number of iterations is reached).

[0069] When the iterative optimization converges, the final design parameter points on the Riemannian manifold can be obtained. Then, an inverse mapping from the Riemannian manifold coordinates to the design parameter space coordinates is performed to obtain the design parameter set.

[0070] It is understandable that this set of design parameters is the optimal design solution obtained from the optimization process, and the design parameters can be directly used to guide the manufacturing, configuration, or writing of control programs for the physical system.

[0071] 105. Update the physical system based on the set of design parameters.

[0072] This embodiment applies a set of design parameters to the corresponding design stage of a physical system. The specific implementation varies depending on the type of physical system and the design stage.

[0073] For example, if the physical system is in the design phase, the design parameters in the design parameter set are updated to the corresponding fields in the computer-aided design (CAD), electronic design automation (EDA), or simulation model. Examples include updating transistor dimensions in the integrated circuit layout, adjusting the geometric parameters of mechanical structures, and modifying lens curvature data in an optical system.

[0074] For example, if the physical system is a dynamically adjustable controller or programmable device (such as an FPGA), the design parameters (such as controller gain, filter coefficients, and operating point voltage) in the design parameter set are written into the corresponding configuration file or register to complete the online / offline update of the software or firmware.

[0075] In summary, the physical system optimization method provided in this application includes obtaining the design parameter space and multiple conflicting performance indices of the physical system to be optimized; mapping the design parameter space to a Riemannian manifold; identifying the inherent symmetry of the physical system in an ideal state, and characterizing the trade-offs between multiple performance indices as symmetry breaking modes based on the inherent symmetry; performing geometric optimization on the Riemannian manifold based on the symmetry breaking modes to search for a set of design parameters that satisfy the target performance requirements while maintaining physical constraints; and updating the physical system according to the set of design parameters. In this embodiment, the physical system optimization method maps performance index conflicts to symmetry breaking modes, enabling the optimization process to not only pursue local performance improvements but also to start from the deep physical structure of the physical system and guide the design parameters to evolve towards overall equilibrium, significantly improving optimization efficiency and convergence stability. Secondly, by performing iterative updates using a geometric integrator on a Riemannian manifold, all intermediate and final solutions automatically satisfy the embedded physical constraints, fundamentally ensuring the physical consistency and feasibility of the optimization results. Furthermore, the mathematical framework provided by this physical system optimization method is domain-independent and has strong theoretical guarantees, and can be widely applied to multiple physical fields such as integrated circuits, mechanical systems, electromagnetic design, and thermal management, providing a unified and reliable mathematical tool and engineering practice path for achieving cross-scale, multi-objective system optimization.

[0076] To facilitate better implementation of the physical system optimization method provided in this application, this application also provides a physical system optimization apparatus. The meanings of the terms used are the same as in the physical system optimization method described above, and specific implementation details can be found in the descriptions within the method embodiments.

[0077] Please see Figure 3 , Figure 3 This is a schematic diagram of the physical system optimization device provided in an embodiment of this application. The physical system optimization device may include an acquisition unit 201, a mapping unit 202, an identification unit 203, an iteration unit 204, and an update unit 205. The acquisition unit 201 is used to acquire the design parameter space and multiple conflicting performance indicators of the physical system to be optimized. Mapping unit 202 is used to map the design parameter space to a Riemannian manifold; The identification unit 203 is used to identify the inherent symmetry of the physical system under ideal conditions, and to characterize the trade-off relationship between multiple performance indicators as a symmetry breaking mode based on the inherent symmetry. Iteration unit 204 is used to perform geometric optimization on the Riemannian manifold based on symmetry breaking modes, in order to search for a set of design parameters that meet the target performance requirements while preserving physical constraints; Update unit 205 is used to update the physical system based on the set of design parameters.

[0078] For specific implementation methods of each of the above units, please refer to the embodiments of the above physical system optimization method, which will not be repeated here.

[0079] In summary, the physical system optimization apparatus provided in this application can acquire the design parameter space and multiple conflicting performance indicators of the physical system to be optimized by the acquisition unit 201; map the design parameter space to a Riemannian manifold by the mapping unit 202; identify the inherent symmetry of the physical system in an ideal state by the identification unit 203, and characterize the trade-off relationship between multiple performance indicators as a symmetry breaking mode based on the inherent symmetry; perform geometric optimization on the Riemannian manifold based on the symmetry breaking mode by the iteration unit 204, so as to search for a set of design parameters that meet the target performance requirements while maintaining physical constraints; and update the physical system according to the set of design parameters by the update unit 205. This application embodiment can improve the optimization efficiency of the physical system.

[0080] This application also provides an electronic device that may integrate the physical system optimization device of this application, such as... Figure 4 As shown, it illustrates a structural schematic diagram of the electronic device involved in the embodiments of this application, specifically: The electronic device may include components such as a processor 301 with one or more processing cores and a memory 302 with one or more computer-readable storage media. Those skilled in the art will understand that... Figure 4 The electronic device structure shown does not constitute a limitation on the electronic device and may include more or fewer components than shown, or combine certain components, or have different component arrangements. Wherein: The processor 301 is the control center of the electronic device. It connects various parts of the electronic device via various interfaces and lines. By running or executing software programs stored in the memory 302 and / or this application, and by calling data stored in the memory 302, it performs various functions and processes data, thereby providing overall monitoring of the electronic device. Optionally, the processor 301 may include one or more processing cores; preferably, the processor 301 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operation of the storage medium, user interface, and application programs, while the modem processor mainly handles wireless communication. It is understood that the modem processor may not be integrated into the processor 301.

[0081] The memory 302 can be used to store software programs and this application. The processor 301 executes various functional applications and data processing by running the software programs and this application stored in the memory 302. The memory 302 may mainly include a program storage area and a data storage area. The program storage area may store applications required for operating the storage medium and at least one function; the data storage area may store data created based on the use of the electronic device. In addition, the memory 302 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 volatile solid-state storage device. Accordingly, the memory 302 may also include a memory controller to provide the processor 301 with access to the memory 302.

[0082] Although not shown, the electronic device may also include a display unit, an input unit, and a power supply, etc., which will not be described in detail here. Specifically, in this embodiment, the processor 301 in the electronic device loads the executable files corresponding to the processes of one or more application programs into the memory 302 according to the following instructions, and the processor 301 runs the application programs stored in the memory 302 to realize various functions, as follows: Obtain the design parameter space and multiple conflicting performance metrics of the physical system to be optimized; Map the design parameter space to a Riemannian manifold; Identify the inherent symmetry of a physical system under ideal conditions, and characterize the trade-offs between multiple performance indices as symmetry breaking modes based on the inherent symmetry. Based on the symmetry breaking mode, geometric optimization is performed on the Riemannian manifold to search for a set of design parameters that meet the target performance requirements while preserving physical constraints. The physical system is updated based on the set of design parameters.

[0083] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0084] Therefore, embodiments of this application provide a storage medium storing a plurality of instructions that can be loaded by a processor to execute steps in any of the methods provided in embodiments of this application. For example, the instructions can execute the following steps: Obtain the design parameter space and multiple conflicting performance metrics of the physical system to be optimized; Map the design parameter space to a Riemannian manifold; Identify the inherent symmetry of a physical system under ideal conditions, and characterize the trade-offs between multiple performance indices as symmetry breaking modes based on the inherent symmetry. Based on the symmetry breaking mode, geometric optimization is performed on the Riemannian manifold to search for a set of design parameters that meet the target performance requirements while preserving physical constraints. The physical system is updated based on the set of design parameters.

[0085] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0086] The storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0087] Since the instructions stored in the storage medium can execute the steps of any method provided in the embodiments of this application, the beneficial effects that any method provided in the embodiments of this application can achieve can be realized. For details, please refer to the previous embodiments, which will not be repeated here.

[0088] The physical system optimization method, apparatus, storage medium, and electronic device provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method of physical system optimization, characterized by, The method comprises: obtaining a design parameter space of a physical system to be optimized and a plurality of mutually conflicting performance indicators; mapping the design parameter space into a Riemannian manifold; identifying inherent symmetries of the physical system in an ideal state, and representing trade-off relationships between the plurality of performance indicators as symmetry breaking patterns according to the inherent symmetries; performing geometric optimization on the Riemannian manifold based on the symmetry breaking patterns to search for a design parameter set satisfying a target performance requirement while maintaining physical constraints; and updating the physical system according to the design parameter set.

2. The physical system optimization method of claim 1, wherein, The identifying the inherent symmetries of the physical system in the ideal state comprises: analyzing physical laws or idealized models followed by the physical system to determine an invariable transformation group of the physical system that remains unchanged under transformation; obtaining the inherent symmetries of the physical system based on the invariable transformation group.

3. The physical system optimization method of claim 1, wherein, The representing the trade-off relationships between the plurality of performance indicators as symmetry breaking patterns according to the inherent symmetries comprises: determining Lie algebras corresponding to the inherent symmetries; identifying breaking generators causing conflicts among the plurality of performance indicators from the Lie algebras; and mapping the trade-off relationships between the performance indicators into the symmetry breaking patterns based on the breaking generators.

4. The physical system optimization method of claim 1, wherein, The performing geometric optimization on the Riemannian manifold based on the symmetry breaking patterns to search for a design parameter set satisfying a target performance requirement while maintaining physical constraints comprises: constructing a modified vector field for guiding optimization according to the symmetry breaking patterns; adopting a geometric integrator to perform iterative optimization on the Riemannian manifold based on the modified vector field to update a position of a design parameter on the Riemannian manifold and maintain physical constraints of the physical system; and when the iterative optimization converges, mapping a final design parameter point on the Riemannian manifold back to the design parameter space to obtain a design parameter set.

5. The physical system optimization method of claim 4, wherein, The constructing a modified vector field for guiding optimization according to the symmetry breaking patterns comprises: decomposing the symmetry breaking patterns into a plurality of independent breaking components; for each of the breaking components, constructing a corresponding vector field component, a direction of the vector field component pointing to a partial recovery direction of a symmetry corresponding to the corresponding breaking component; and combining the vector field components to obtain the modified vector field.

6. The physical system optimization method of claim 1, wherein, The mapping the design parameter space into a Riemannian manifold comprises: generating a metric tensor based on the performance indicators and the design parameter space; and mapping the design parameter space into a Riemannian manifold by using the metric tensor.

7. The physical system optimization method of any one of claims 1-6, wherein, The inherent symmetries comprise at least one of time translation symmetry, space translation symmetry, space rotation symmetry, gauge symmetry, and scale symmetry.

8. A physical system optimization apparatus, characterized by, The method comprises: obtaining a design parameter space of a physical system to be optimized and a plurality of mutually conflicting performance indicators; mapping the design parameter space into a Riemannian manifold; identifying inherent symmetries of the physical system in an ideal state, and representing trade-off relationships between the plurality of performance indicators as symmetry breaking patterns according to the inherent symmetries; performing geometric optimization on the Riemannian manifold based on the symmetry breaking patterns to search for a design parameter set satisfying a target performance requirement while maintaining physical constraints; and updating the physical system according to the design parameter set. an iteration unit configured to perform a geometric optimization on the Riemannian manifold based on the symmetry breaking pattern to search for a set of design parameters satisfying a target performance requirement while maintaining physical constraints; an updating unit configured to update the physical system according to the set of design parameters.

9. A storage medium, characterized by The storage medium stores a plurality of instructions adapted to be loaded by a processor to execute the physical system optimization method according to any one of claims 1-7.

10. An electronic device, comprising: A computer program product comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the physical system optimization method according to any one of claims 1-7 when executing the computer program.

Images