Magnetic material surface temperature gradient generation method based on laser regulation and control

Abstract

The invention provides a magnetic material surface temperature gradient generation method based on laser regulation and control, and relates to the field of data processing. According to the method, optical absorption, electron energy relaxation, phonon thermal conductivity and magnetic anisotropy parameters of a magnetic material are obtained, modeling is carried out in a three-dimensional space according to a silicon-based substrate, a cobalt layer and a platinum layer, and a cross-scale photo-thermal-magnetic full-coupling model is constructed. The method comprises the following steps: generating a laser coding tensor through light intensity, wavelength and phase parameters of a laser space modulation array, driving a model to solve in a simulation environment, generating temperature gradient evolution data and an implicit field quantity, and constructing a temperature gradient control database. After target temperature gradient distribution is set, a depth prediction model is used for outputting a temperature gradient evolution sequence, sensitivity distribution is calculated through the model, the optimal laser coding tensor is determined, finally verification and application to a laser array are carried out, and temperature distribution consistent with the target temperature gradient is achieved. By implementing the technical scheme, the temperature gradient on the surface of the magnetic material is convenient to generate.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and specifically to a method for generating a surface temperature gradient of a magnetic material based on laser control. Background Technology

[0002] Against the backdrop of rapid development in magnetic materials and spintronics research, utilizing temperature gradient-driven thermal spin effects, anomalous Nernst effects, and magnetothermal coupling modulation has become an important technological direction for realizing novel magnetic functional devices, on-chip spin thermoelectric conversion components, and programmable magnetic structures. As magnetic material devices evolve towards micro-nano scales, high integration, and high response speeds, device performance increasingly depends on the precisely controllable temperature gradient distribution on the surface of magnetic materials.

[0003] However, temperature gradient, as a complex physical quantity jointly determined by light field distribution, material energy absorption behavior, ultrafast carrier relaxation dynamics, translayer thermal diffusion, and magnetic domain microstructure response, exhibits strong coupling and nonlinear characteristics on a spatial scale. This makes traditional methods of generating temperature gradients based on macroscopic heat source loading, single laser spot irradiation, or fixed light field mode control unable to meet the requirements of complex temperature gradient shape, controllable direction, adjustable amplitude, and designable uniformity, which is not conducive to the generation of temperature gradients on the surface of magnetic materials.

[0004] Therefore, there is an urgent need for a method to generate a surface temperature gradient of magnetic materials based on laser control. Summary of the Invention

[0005] This application provides a method for generating surface temperature gradients of magnetic materials based on laser control, which facilitates the generation of surface temperature gradients of magnetic materials.

[0006] The first aspect of this application provides a method for generating a surface temperature gradient of a magnetic material based on laser modulation. The method includes: acquiring optical absorption parameters, electron energy relaxation parameters, phonon thermal conductivity parameters, and magnetic anisotropy parameters of the magnetic material, and modeling them in three-dimensional space based on a silicon substrate, a cobalt layer, and a platinum layer to construct a cross-scale photothermal-magnetic fully coupled model; acquiring the light intensity parameters, wavelength parameters, and phase parameters of a laser spatial modulation array composed of multiple light spot units, and encoding them into a laser-coded tensor in spatial order; reading the laser-coded tensor in a simulation environment, and simultaneously driving the cross-scale photothermal-magnetic fully coupled model to perform batch solving to generate temperature gradient evolution data and implicit field quantities, and then processing the data based on the temperature gradient evolution data and the... An implicit field quantity constitutes a temperature gradient manipulation database; a target temperature gradient distribution on the surface of a magnetic material is set, and a temperature gradient evolution sequence is output through a depth prediction model based on the temperature gradient manipulation database; under the constraint of the target temperature gradient distribution, the sensitivity distribution is calculated using the cross-scale photothermal-magnetic fully coupled model to generate a reverse update direction, and a unique optimal laser coding tensor is determined based on the reverse update direction; the unique optimal laser coding tensor is imported into the cross-scale photothermal-magnetic fully coupled model for verification, and under the condition of satisfying a preset deviation tolerance, the unique optimal laser coding tensor is applied to the laser spatial modulation array to generate a temperature gradient distribution on the surface of the magnetic material that is consistent with the target temperature gradient distribution.

[0007] A second aspect of this application provides a laser-controlled magnetic material surface temperature gradient generation device. The device includes an acquisition module and a processing module. The acquisition module acquires optical absorption parameters, electron energy relaxation parameters, phonon thermal conductivity parameters, and magnetic anisotropy parameters of the magnetic material, and models them in three-dimensional space based on a silicon substrate, a cobalt layer, and a platinum layer to construct a cross-scale photothermal-magnetic fully coupled model. The acquisition module also acquires the light intensity parameters, wavelength parameters, and phase parameters of a laser spatial modulation array composed of multiple light spot units, and encodes them into a laser-coded tensor in spatial order. The processing module reads the laser-coded tensor in a simulation environment and simultaneously drives the cross-scale photothermal-magnetic fully coupled model to perform batch solutions to generate temperature gradient evolution data and implicit field quantities, and processes the data according to the temperature gradient evolution. The data and the implicit field quantity constitute a temperature gradient manipulation database; the processing module is further configured to set the target temperature gradient distribution on the surface of the magnetic material, and output a temperature gradient evolution sequence based on the temperature gradient manipulation database through a depth prediction model; the processing module is further configured to calculate the sensitivity distribution using the cross-scale photothermal-magnetic fully coupled model under the constraint of the target temperature gradient distribution, to generate a reverse update direction, and determine a unique optimal laser coding tensor based on the reverse update direction; the processing module is further configured to import the unique optimal laser coding tensor into the cross-scale photothermal-magnetic fully coupled model for verification, and apply the unique optimal laser coding tensor to the laser spatial modulation array under the condition of satisfying a preset deviation tolerance, so as to generate a temperature gradient distribution on the surface of the magnetic material that is consistent with the target temperature gradient distribution.

[0008] A third aspect of this application provides an electronic device including a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, and both the user interface and the network interface are used to communicate with other devices. The processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method described above.

[0009] A fourth aspect of this application provides a non-transitory computer-readable storage medium storing instructions that, when executed, perform the method described above.

[0010] In summary, one or more technical solutions provided in this application have at least the following technical effects or advantages: By constructing a complete control chain that coordinates a cross-scale photothermal-magnetic fully coupled model, a laser-encoded tensor system, a temperature gradient manipulation database, and a depth prediction model, the temperature gradient generation process on the surface of magnetic materials is transformed from empirical adjustment into a systematic process that is plannable, predictable, reversible, and precisely reproducible. Under multi-physics constraints, a precise mapping between laser spatial modulation and temperature gradient evolution is achieved. This allows the spatial shape, direction, amplitude, and uniformity of the temperature gradient to be actively designed and dynamically controlled at high resolution through laser-encoded tensors, rather than relying on traditional macroscopic adjustment methods. This significantly improves the spatial freedom and repeatability of temperature gradient manipulation. Furthermore, the combination of the temperature gradient manipulation database and the depth prediction model enables precise control of the temperature gradient from the input light field to the surface. The prediction process for the output temperature gradient has been transformed from high-cost simulation to low-cost, rapid inference, providing a data foundation and reasoning capability for subsequent optimization calculations. Reversible optimization of the laser-encoded tensor is achieved through a reverse update direction guided by sensitivity distribution, enabling the target temperature gradient distribution to uniquely determine the corresponding laser modulation method under physical constraints. This solves the problem of accurate back-calculation and global optimization of temperature gradients in traditional technologies. Finally, closed-loop verification through simulation and actual laser spatial modulation array applications ensures consistency and robustness of the temperature gradient modulation scheme across theoretical analysis, model prediction, and practical implementation. This provides a programmable, high-precision, and engineering-feasible means of generating temperature gradients for thermal field control, magnetocaloric effect enhancement, and spin thermoelectric measurement of magnetic material devices. Therefore, it facilitates the generation of temperature gradients on the surface of magnetic materials. Attached Figure Description

[0011] Figure 1 A schematic flowchart illustrating a method for generating a surface temperature gradient of a magnetic material based on laser control, provided for an embodiment of this application; Figure 2 A schematic diagram of a laser-controlled magnetic material surface temperature gradient generation device provided for an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0012] Explanation of reference numerals in the attached figures: 21. Acquisition module; 22. Processing module; 31. Processor; 32. Communication bus; 33. User interface; 34. Network interface; 35. Memory. Detailed Implementation

[0013] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0014] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0015] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0016] To address the aforementioned technical problems, this application provides a method for generating a surface temperature gradient of a magnetic material based on laser control, referring to... Figure 1 , Figure 1 This is a flowchart illustrating a method for generating a surface temperature gradient of a magnetic material based on laser control, provided in an embodiment of this application. The method is applied to a server and includes steps S110 to S160, as follows:

[0017] S110: Obtain the optical absorption parameters, electronic energy relaxation parameters, phonon thermal conductivity parameters, and magnetic anisotropy parameters of the magnetic material, and build a cross-scale photothermal-magnetic fully coupled model based on the silicon substrate, cobalt layer, and platinum layer in three-dimensional space.

[0018] Specifically, in this embodiment, the server is used to perform a large number of numerical calculations and data management. It is a computing platform for running simulation software, optimization programs, and deep prediction models. It can be a workstation cluster configured with high-performance CPUs and graphics processing units to complete the solution of cross-scale photothermal-magnetic fully coupled models and the construction of temperature gradient manipulation databases in a short time. For example, when performing batch simulations of multiple sets of laser-encoded tensors, the server can simultaneously schedule multiple computing tasks to calculate the temperature gradient evolution process of the magnetic material surface under different laser irradiation conditions, thereby obtaining a sufficient amount of sample data within a reasonable time.

[0019] Magnetic materials refer to materials that undergo significant magnetization under the influence of an external magnetic field and retain partial magnetization even after the external magnetic field is removed. In this embodiment, the magnetic material is a metal thin film structure containing a cobalt layer, such as a cobalt thin film or a cobalt-platinum multilayer film deposited on a silicon substrate. The cobalt layer provides a ferromagnetic response and is the core carrier for observing the magnetic response and studying the spin thermal effect after temperature gradient control. For example, in the measurement of the anomalous Nernst effect, the cobalt layer, as a magnetic material, determines the measured transverse voltage signal by its magnetization direction together with the temperature gradient direction.

[0020] Furthermore, a reference sample containing only a silicon substrate and a multilayer film sample containing a silicon substrate, a cobalt layer, and a platinum layer are first placed sequentially on the sample stage. An ellipsometer is controlled to scan within a preset wavelength range and incident angle range, measuring the amplitude ratio and phase difference parameters at each wavelength. Based on a multilayer thin-film optical model, the server performs nonlinear fitting on the measured amplitude ratio and phase difference parameters to deduce the real and imaginary parts of the complex refractive index of the cobalt and platinum layers at different wavelengths. The imaginary part is denoted as the extinction coefficient, and the optical absorption parameters are calculated from the extinction coefficient. The optical absorption parameters can be expressed as a function:

[0021] in, Indicates at wavelength The optical absorption parameter at a given point is a wavelength-dependent scalar function used to characterize the absorption intensity of magnetic materials for incident light energy at different wavelengths. Indicates at wavelength The extinction coefficient at a certain point is a function of the imaginary part of the complex refractive index obtained by fitting. It is used to characterize the attenuation of electromagnetic waves inside the material and takes a value of non-negative real number. The wavelength of the laser or probe light during the scanning process is represented by an independent variable within a preset wavelength scanning range. Through this functional relationship, a calculable connection is established between the optical absorption parameters and the ellipsometric spectral measurement results. The server ultimately obtains a table of optical absorption parameters stored discretely by wavelength, which is used to subsequently construct the optical field solution sub-model in the cross-scale photothermal-magnetic fully coupled model.

[0022] In obtaining the electron energy relaxation parameters, a curve of electron temperature decaying over time was established through a femtosecond pump-probe experiment. The server controlled a femtosecond laser system to apply pump pulses to the magnetic material sample and monitored the intensity of the reflected or transmitted signals using delayed probe pulses, recording the transient optical response under different time delays. The server mapped the transient optical response to an electron temperature time series using a known response model, and then performed an exponential decay fitting on the electron temperature time series, which can be described by a first-order relaxation model:

[0023] in, Indicates time The electron temperature at time is a time function describing the electron's energy state; The lattice equilibrium temperature is the temperature at which a material tends to stabilize after laser excitation has ceased or after a sufficiently long period of time has elapsed. It represents the initial electron temperature at the moment immediately after laser excitation ends, which is the peak temperature reached by the electronic system after absorbing the pump pulse; It represents the time starting from the end of the laser pump pulse, and is the time independent variable of the relaxation process; The electron energy relaxation time, represented by the fitted electron energy relaxation parameter, characterizes the time scale required for electrons to transfer energy to the lattice and return to near equilibrium. By fitting samples with different layer structures or different process parameters, the server can obtain the electron energy relaxation parameters of cobalt and platinum layers respectively, and store them as inputs to the electron energy evolution equation in the cross-scale photothermal-magnetic fully coupled model.

[0024] Phonon thermal conductivity parameters can be obtained through time-domain thermal reflection experiments or steady-state thermal reflection experiments. The server controls a modulated light source to apply periodic or pulsed heating to the sample surface, while simultaneously monitoring the change in the sample surface reflection signal over time or with the modulation frequency using a probe light. The surface temperature response curve is obtained by utilizing the mapping relationship between the reflection signal and the surface temperature. The server fits the temperature response curve to a one-dimensional or two-dimensional thermal diffusion model to deduce the effective thermal conductivity of the silicon substrate, cobalt layer, and platinum layer, and denotes this thermal conductivity as phonon thermal conductivity parameters. These parameters are used as coefficients in the thermal diffusion term of the thermal field solution sub-model, controlling the diffusion rate of heat within different layers and at interlayer interfaces. For example, when the phonon thermal conductivity parameter of the silicon substrate is large while that of the cobalt layer is small, the solution result of the cross-scale photothermal-magnetic fully coupled model will exhibit a steeper temperature gradient near the cobalt layer and a slower temperature change inside the substrate. This characteristic accurately reflects the heat dissipation behavior of the actual device.

[0025] Magnetic anisotropy parameters can be obtained through magneto-optical Kerr effect testing. The server controls the magneto-optical Kerr effect measurement system to scan under different applied magnetic field directions and intensities, measuring the Kerr rotation angle of the reflected light as a function of the applied magnetic field. This curve is then fitted using a pre-defined magnetic anisotropy energy model to obtain the anisotropy constant and the direction of the easy magnetization axis. The magnetic anisotropy parameters can be correlated with the magnetization intensity using formulas, such as an approximate linear relationship:

[0026] in, Indicates at wavelength The magneto-optical Kerr effect rotation angle measured at the point is an observation recorded during the scanning process of the applied magnetic field; The proportionality coefficient, which is related to the measurement optical path, material optical properties, and experimental geometry, is a parameter obtained during the fitting process and is used to connect the Kerr signal and the magnetization. The magnetization intensity of a magnetic material in the measurement direction is a physical quantity that determines the change in the Kerr signal. By fitting magnetic fields of different angles and intensities, the constraint form of magnetic anisotropy energy on the magnetization direction can be obtained. This information is then organized into a set of magnetic anisotropy parameters, providing magnetic free energy constraints for the magnetic field solution sub-model in the cross-scale photothermal-magnetic fully coupled model.

[0027] After obtaining the optical absorption parameters, electronic energy relaxation parameters, phonon thermal conductivity parameters, and magnetic anisotropy parameters, the server constructs a three-dimensional multilayer geometric model of a silicon substrate, a cobalt layer, and a platinum layer in the simulation software. Specifically, based on the lateral dimensions and thickness parameters of the actual device, the bulk region of the silicon substrate, the thin-layer region of the cobalt layer, and the surface layer region of the platinum layer are defined sequentially in three-dimensional space, ensuring that the interface positions of each layer are consistent with the sample fabrication process. Then, in the optical field solving sub-model, the wavelength-discrete optical absorption parameters are bound to the geometric regions of the cobalt and platinum layers, allowing the optical field solving sub-model to convert the laser intensity distribution into a volumetric energy deposition distribution within each grid cell using the optical absorption parameters. For example, this can be achieved using the following form:

[0028] in, Indicates spatial location At wavelength The volumetric energy deposition distribution is the source term output by the optical field solution sub-model to the thermal field solution sub-model; Indicates the spatial position of the laser. Location, wavelength is The light intensity distribution at that time was calculated from the laser coding tensor and optical propagation. This represents the coordinates extending inward from the material surface along the thickness direction, used to describe the energy attenuation path within the material. This relationship allows the optical field solution sub-model to determine the local energy deposition based on optical absorption parameters, laser spatial distribution, and thickness coordinates.

[0029] In the thermal field solution sub-model, the server binds phonon thermal conductivity parameters to the corresponding grid regions of the silicon substrate, cobalt layer, and platinum layer, and introduces electron energy relaxation parameters into the coupling equation between electron temperature and lattice temperature. Through time stepping, the temperature field distribution in each grid is updated at each time step, enabling the thermal field solution sub-model to derive the temperature field and temperature gradient evolution processes inside the material and on the surface of the magnetic material, starting from the volumetric energy deposition source term given by the optical field solution sub-model. In the magnetic field solution sub-model, the server assigns magnetic anisotropy parameters to the cobalt layer region, controls the evolution of magnetization direction through magnetic free energy expressions and effective field expressions, and uses the temperature field and temperature gradient field as inputs affecting magnetic anisotropy and exchange interactions, enabling the magnetic field solution sub-model to provide the magnetic domain dynamics behavior under photothermal excitation. In this way, optical absorption parameters, electronic energy relaxation parameters, phonon thermal conductivity parameters, and magnetic anisotropy parameters are uniformly bound to the optical field solution sub-model, thermal field solution sub-model, and magnetic field solution sub-model in a three-dimensional multilayer structure, forming a cross-scale optical-thermal-magnetic fully coupled model that is continuously coupled in time and space. This provides a physically accurate basic model for subsequent laser-coded tensor-driven simulation, temperature gradient manipulation database construction, and temperature gradient control optimization.

[0030] S120. Obtain the light intensity parameters, wavelength parameters, and phase parameters of the laser spatial modulation array composed of multiple light spot units, and encode them into a laser coded tensor in spatial order.

[0031] Specifically, the laser-coded tensor is a high-dimensional data structure that packages the intensity, wavelength, and phase parameters of all light spot elements in spatial order. It is used to uniformly transmit configuration information of the laser spatial modulation array between servers, simulation environments, and control systems. Mathematically, the laser-coded tensor is an array structure with multi-dimensional indexes, including the lateral and vertical positions of the light spot elements and the channel dimensions of different parameter types. Physically, it comprehensively describes the spatial light field distribution provided by the laser spatial modulation array to the surface of the magnetic material at a given moment.

[0032] Furthermore, the server sends control commands to the spatial light modulator, digital micromirror device, or liquid crystal light valve to discretize the continuous incident laser beam on a two-dimensional plane according to a preset grid, with each grid region corresponding to a light spot unit. To ensure that all subsequent light spot units have traceable positional markers on the magnetic material surface, the center position coordinates of each light spot unit are measured during the calibration stage using a scale sample or imaging screen. Then, according to the set spatial step size and Mapping physical coordinates to discrete space coordinate indices For example, a relationship can be used:

[0033] in, The discrete row index of the light spot unit in the horizontal direction is a non-negative integer used to identify the row position of the light spot unit in the array; The discrete column index of the spot unit in the vertical direction is a non-negative integer used to identify the column position of the spot unit in the array; and The actual physical coordinates of the light spot unit projected onto the surface of the magnetic material are continuous spatial positions measured by an imaging or displacement platform; and This indicates the selected reference origin coordinates, used to uniformly translate and align the positions of all light spot units; and This represents the spacing between adjacent light spot units in the horizontal and vertical directions, used to quantize continuous coordinates into grid indices. Through this mapping, each light spot unit is assigned a unique spatial coordinate index. This forms a set of light spot units with programmable modulation capability and spatial index, providing a spatial reference for subsequent parameter calibration and laser-coded tensor construction.

[0034] After obtaining the set of light spot units with spatial coordinate indices, optical power measurement, spectral measurement, and interferometry are performed on each light spot unit to obtain light intensity parameters, wavelength parameters, and phase parameters. During optical power measurement, a tiny photodetector is placed at an equivalent location on the magnetic material surface, or a high dynamic range camera is used to measure individual light spot units or individually activated light spot units. The server records the measured optical power as follows: Combined with the illumination area of ​​the light spot unit The light intensity parameter can be calculated as follows:

[0035] in, Indicates spatial coordinate index as The light intensity parameter of the light spot unit is used to characterize the light power density per unit area of ​​the light spot unit; This represents the average optical power of the light spot unit within the measurement time window, and is a quantity obtained from the calibration of the power meter or camera; This indicates the effective irradiation area of ​​the light spot unit on the surface of the magnetic material, which is an area parameter obtained through geometric optics calculations or calibration images.

[0036] The wavelength parameters are acquired by individually illuminating each laser spot unit. The server controls the laser spatial modulation array to ensure that only the target laser spot unit outputs laser light, and guides the output light of that spot unit into a spectrometer. The spectrometer then acquires the spectral distribution curve of that spot unit across the entire visible or near-infrared band. The server analyzes the spectral distribution curve, identifies the position of the main peak, and uses the wavelength corresponding to the main peak as the main emission wavelength of that spot unit, thus forming the wavelength parameter of that spot unit. This parameter characterizes the main emission component of the spot unit in its spectral properties. For example, for a spot unit using a tunable light source, the server can accurately record the change in its main emission wavelength after switching operating states, giving the spectral attributes of each spot unit a clear physical meaning.

[0037] The acquisition of phase parameters relies on interferometry. The server selects a reference spot cell with a known phase and illuminates it simultaneously with the spot cell to be calibrated, causing the two laser beams to form an interference pattern on the imaging screen. The interference fringes captured by the imaging screen contain a spatial intensity distribution of alternating bright and dark areas. The server analyzes the spatial period, fringe offset, and bright / dark distribution of the interference fringes, and solves for the phase offset of the spot cell to be calibrated by combining the known phase state of the reference spot cell. The phase offset is used to describe the relative starting position of the electromagnetic field oscillation of the spot cell with respect to the reference spot cell, thus reflecting the actual contribution of the spot cell in the interference superposition. For example, in scenarios where multiple spot cells are superimposed to form a complex light field distribution, the phase parameter enables the laser coding scheme to describe the wavefront shape and interference structure changes of the light field.

[0038] After establishing the correspondence between the light intensity parameters, wavelength parameters, and phase parameters and the spatial coordinate indices of the light spot units, the server organizes these parameters into a laser-coded tensor according to a preset spatial order. This tensor is used to uniformly characterize the optical field structure of the laser spatial modulation array in subsequent simulations and control. Through this organization, the laser-coded tensor completely records the optical state of all light spot units in the laser spatial modulation array at the data structure level, and corresponds one-to-one with their physical spatial positions. This allows the laser-coded tensor to be used to generate local energy deposition distributions in a cross-scale opto-thermal-magnetic fully coupled model, and to be directly converted into driving commands for spatial light modulators or digital micromirror devices in the actual control stage, thereby achieving consistency between the simulation environment and the hardware system regarding the optical field structure.

[0039] S130. Read the laser-encoded tensor in the simulation environment, and simultaneously drive the cross-scale photothermal-magnetic fully coupled model to perform batch solutions to generate temperature gradient evolution data and implicit field quantities, and construct a temperature gradient manipulation database based on the temperature gradient evolution data and implicit field quantities.

[0040] Specifically, temperature gradient evolution data is one of the key outputs of the cross-scale photothermal-magnetic fully coupled model during batch solving. It describes the entire process of temperature gradient changes over time and space within magnetic materials, especially on their surfaces. It is a collection of temperature difference information acquired at multiple time steps and spatial locations. The temperature gradient evolution data not only reflects the magnitude and direction of the temperature difference between different locations on the surface of the magnetic material at each moment, but also records the dynamic trajectory of the temperature gradient from laser activation to stable establishment and possible decay.

[0041] Implicit field quantities refer to intermediate physical quantities that are closely related to the formation of temperature gradients but are not directly used as the target output in solving multi-scale photothermal-magnetic fully coupled models. They are used to supplement the description of the internal state and driving mechanism under the interaction of multiple physics fields. These physical quantities may include interface energy deposition rate, local effective heat flux density, specific heat change, electron temperature distribution, lattice temperature distribution, magnetic domain spin fluctuation intensity, and magnetization vector components. Although implicit field quantities are not directly used as control indicators of the target temperature gradient, they can serve as auxiliary features when constructing temperature gradient manipulation databases and training deep prediction models, improving the model's understanding of physical processes and extrapolation capabilities.

[0042] The temperature gradient manipulation database is a dataset consisting of a large number of different laser-coded tensor samples, their corresponding temperature gradient evolution data, and hidden field entries. It is a structured data resource that systematically binds "laser spatial modulation input," "temperature gradient response output," and "internal physical state." Each record in the temperature gradient manipulation database contains a set of laser-coded tensors, the temperature gradient evolution data obtained by solving this set of laser-coded tensors in a multi-scale photothermal-magnetic fully coupled model, and the hidden field associated with the temperature gradient evolution process. This facilitates the subsequent deep prediction model to learn the nonlinear mapping relationship between the laser-coded tensor and the temperature gradient response during training, and to use the hidden field to physically constrain the solution space during optimization.

[0043] Furthermore, the server reads the laser-encoded tensor in the simulation environment and maps the intensity, wavelength, and phase parameters corresponding to each spot element in the laser-encoded tensor to energy deposition load elements, optical absorption elements, and phase superposition elements in the multi-scale photothermal-magnetic fully coupled model, respectively. For each spot element, a set of mesh nodes corresponding to it in the three-dimensional geometric model is determined based on its spatial coordinate index. The intensity parameters are combined with the aforementioned optical absorption parameters to generate a volumetric energy deposition distribution within this region. The irradiation area of ​​multiple spot elements is then discretized through meshing to cover the entire surface of the magnetic material. The energy deposition load can be denoted as a volumetric energy source term function:

[0044] in, Indicates position and time The volumetric energy source term at that location is the local energy injection rate applied within the energy deposition load cell; Indicates at wavelength The optical absorption parameters of the corresponding material layer are scalar functions that vary with wavelength, obtained by ellipsometric spectroscopy measurement and fitting. Indicates spatial coordinate index as The light intensity parameter of the light spot unit is used to characterize the light power density per unit area of ​​the light spot unit; This represents the spatial distribution function of the light spot element in the 3D model, used to describe the position of the light spot element. Whether a point contributes and its contribution weight, with values ​​ranging from zero to one; This represents a laser time waveform function, used to describe the switching mode and pulse envelope shape of the laser along the time axis. The optical absorption unit utilizes... and Information determines the attenuation behavior of energy in the depth direction. The phase superposition unit uses the phase parameters of each spot unit to superimpose the field intensity distribution in the overlapping area, thereby forming a fine spatial interference light field.

[0045] Within each time step, the cross-scale photothermal-magnetic fully coupled model solves for the electron temperature field, lattice temperature field, and magnetic domain response based on the aforementioned volume energy source term. It can be jointly evolved using a dual-temperature model and the magnetodynamic equations, where the electron temperature field and lattice temperature field can be updated via coupling using the following equation: in, Representing the electron temperature field, it is a temperature function of the electron system defined on a three-dimensional grid; Representing the lattice temperature field, it is a temperature function of the lattice system defined on a three-dimensional grid; This represents evolution time and is the time independent variable in the two-temperature model. The electron specific heat parameter is a material property parameter used to characterize the amount of energy required for an electron temperature change; The lattice specific heat parameter is a material property parameter used to characterize the amount of energy required for a change in lattice temperature; The electronic lattice coupling parameter is a parameter used to describe the intensity of energy exchange between an electronic system and a lattice system. The equivalent thermal conductivity parameter is used to characterize the ability of heat to diffuse within a material. After being assigned a value by layer, the differences between each layer are reflected in the solution. Represents the spatial gradient operator, This represents the thermal diffusion term of the lattice temperature field modulated by thermal conductivity.

[0046] The magnetic domain response can be solved by the magnetization vector time evolution through the magnetodynamic equations, for example, using the form: in, This represents the magnetization vector field, used to characterize the magnetization direction and intensity at different locations within the cobalt layer; The gyromagnetic ratio is a constant parameter that characterizes the precession rate of a magnetic moment under the influence of an effective magnetic field. The effective magnetic field is the equivalent magnetic field distribution obtained by superimposing the applied magnetic field, the magnetic anisotropic field, the exchange field, and temperature-related terms. The damping coefficient is a parameter that controls energy dissipation and convergence speed during the precession of the magnetization vector. Through the above coupled solution, the server can obtain the state distribution of the electronic temperature field, lattice temperature field, and magnetic domain response throughout the entire three-dimensional model within the same time step.

[0047] After obtaining the electronic temperature field and lattice temperature field at continuous time steps, the server extracts the temperature gradient information of the magnetic material surface and its adjacent region from the lattice temperature field, constructing temperature gradient evolution data. The vector representation of the temperature gradient in three-dimensional space can be written as:

[0048] in, Indicates the location and time The lattice temperature gradient vector at a given location is used to describe the trend and strength of temperature changes in three spatial directions. , and These represent the lattice temperatures along... , , The partial derivatives in the three spatial directions are calculated at the mesh nodes using finite difference or finite element shape functions. At each time step, the server extracts values ​​from the mesh nodes located on the surface of the magnetic material. Furthermore, when necessary, spatial averaging or projection along a specific path is performed on a target area, such as statistical analysis of components in a certain direction, thereby forming a time-varying temperature gradient vector sequence. Temperature gradient evolution data is obtained by sampling at multiple time steps and multiple spatial locations. The set can be organized into a multidimensional array or time series set in the server to describe the entire process of the temperature gradient on the surface of magnetic materials over time, from its establishment and amplification to its stabilization and even decay, under specific laser-encoded tensor conditions. It can also be paired with the magnetic domain response data of the corresponding time step for subsequent research on the correlation between temperature gradient and magnetic structure changes.

[0049] During the construction of the implicit fields, the server simultaneously extracts the interface energy deposition rate, effective heat flux density, and magnetic domain spin fluctuation intensity when performing simulations of each set of laser-coded tensors. These implicit fields, along with the temperature gradient evolution data, are then written into the temperature gradient manipulation database. The interface energy deposition rate can be obtained at the cobalt-platinum interface by integrating or averaging the volumetric energy source term over the interface neighborhood, thus constructing the interface energy deposition rate function:

[0050] in, Indicates time The corresponding interface energy deposition rate is the average energy injection rate per unit area of ​​the interface region. The area represents the interface region, which is the selected projection area of ​​the interface between the cobalt and platinum layers. This indicates area integration on the interface area. The aforementioned volumetric energy source term is obtained by integrating or projecting it onto the thickness near the interface to obtain the interface contribution.

[0051] Effective heat flux density can be calculated using the lattice temperature field and thermal conductivity parameters. For example, the heat flux density vector at a certain location can be written as: in, Indicates position and time The effective heat flux density vector at a given location is used to describe the direction and intensity of heat flow at that location. Indicates the location The thermal conductivity parameter of the corresponding material layer is a function that is a piecewise constant or smoothly distributed in space after being assigned values ​​to each layer.

[0052] The domain spin fluctuation intensity can be statistically obtained from the magnetization vector field in the domain response. For example, the degree to which the magnetization vector deviates from the average magnetization vector in a certain target region can be expressed as: in, Indicates time The magnetic domain spin fluctuation intensity is used to quantitatively describe the degree of magnetization vector fluctuation in this region; The number of grid nodes or sampling points within the statistical area is a positive integer. Indicates the first Each node at time... The magnetization vector at that location; This indicates that the statistical region is at time [time]. The average magnetization vector is the average magnetization vector over all The result obtained after performing vector averaging; The vector norm is used to measure the magnitude of the magnetization vector difference. For each set of laser-encoded tensors, the server packages the corresponding temperature gradient evolution data, interface energy deposition rate time series, effective heat flux density distribution, and magnetic domain spin fluctuation intensity time series into a unified package, creating a data entry that corresponds one-to-one with the laser-encoded tensor. This data is then written into the temperature gradient manipulation database according to the index of the laser-encoded tensor, thus forming a high-dimensional data resource that can simultaneously reflect external light field modulation, temperature gradient evolution, and internal implicit physical states. This provides complete data support for subsequent deep prediction model training and laser-encoded tensor optimization.

[0053] S140. Set the target temperature gradient distribution on the surface of the magnetic material, and output the temperature gradient evolution sequence through a deep prediction model based on the temperature gradient manipulation database.

[0054] Specifically, the target temperature gradient distribution is a design description of the temperature gradient state that the surface of a magnetic material should exhibit in space and time. It is a pre-planning of the direction, magnitude, and variation of the temperature gradient across different regions, based on the functional requirements of the device or experimental objectives. A temperature gradient refers to the rate of change of temperature in space, which can be understood as "how much temperature changes for how far one travels in a certain direction." The target temperature gradient distribution requires that this rate of change be distributed according to a pre-defined pattern at different locations on the surface of the magnetic material.

[0055] The deep prediction model is a prediction model built on a deep learning framework to approximate the complex mapping relationship between "laser-coded tensors and temperature gradient evolution sequences". It consists of multi-layer convolutional networks, attention structures, and time-series modeling units. The deep prediction model is trained on a temperature gradient manipulation database, treating the laser-coded tensor as an input feature and the temperature gradient evolution data corresponding to the laser-coded tensor as a supervision signal. At the same time, the hidden field quantities are used as auxiliary features or physical constraints. This allows the model to learn to quickly predict "what kind of temperature gradient evolution process a certain spatial optical field structure will produce on the surface of a magnetic material" without re-invoking the cross-scale photothermal-magnetic fully coupled model.

[0056] A temperature gradient evolution sequence refers to the complete trajectory of the temperature gradient on the surface of a magnetic material over time under a given laser-coded tensor. It records the spatial distribution of the temperature gradient across multiple time steps in the form of a discrete-time series, providing a dynamic characterization of "how the temperature gradient is established, how it reaches a steady state, and whether transient fluctuations occur." Unlike the temperature gradient distribution on a single time slice, the temperature gradient evolution sequence reflects the dynamic characteristics of the system. For example, certain combinations of laser-coded tensors may cause the temperature gradient to oscillate violently initially before gradually stabilizing, while other combinations can establish a stable temperature gradient more quickly. These differences directly affect the controllability of the magnetic response and the selection of the measurement window.

[0057] Furthermore, the process of delineating at least one target region on the surface of the magnetic material can be completed under server control based on a simulation mesh or experimental imaging coordinate system. The server first establishes a two-dimensional coordinate system on the surface of the magnetic material and discretizes the surface into several mesh nodes, using indexes... The location of each surface mesh node is identified. Based on this, one or more mesh sets coinciding with the functional regions of the device are selected as target regions. For each mesh node within the target region, temperature gradient direction constraints, temperature gradient magnitude constraints, and temperature gradient uniformity constraints are given to constitute a quantitative description of the target temperature gradient distribution. The temperature gradient direction constraint is achieved by setting a desired unit direction vector at each mesh node in the target region. Implementation, written as: in, Indicates that the index within the target area is The temperature gradient direction constraint at the surface mesh node is a two-dimensional or three-dimensional unit vector used to characterize the desired spatial direction of temperature change. This represents the temperature gradient vector preset for this grid node during the target planning phase, and is an unnormalized temperature gradient design quantity. This represents the norm of the temperature gradient vector, used to normalize it into directional constraints. Temperature gradient magnitude constraints are achieved by setting an allowable temperature gradient magnitude range for each mesh node in the target region. The implementation is as follows:

[0058] in, Indicates that the index within the target area is The target temperature gradient magnitude at a surface mesh node is a non-negative real number used to constrain the magnitude of the temperature gradient at that location. and These represent the minimum and maximum allowable temperature gradient magnitudes in the design, respectively, and are two threshold parameters determined by the application scenario. Temperature gradient uniformity constraints are achieved by applying variance or range constraints to the temperature gradient magnitudes of all grid nodes within the target region. For example, it could be required that all... The standard deviation does not exceed the preset threshold This is to ensure that the temperature gradient does not exhibit excessive local fluctuations in space. After setting the above-mentioned directional, amplitude, and uniformity constraints, the server will configure all mesh nodes within the target area. Organized by spatial index, the target temperature gradient distribution field is formed and stored as a reference benchmark for subsequent deep prediction model training and evaluation. This ensures that the target temperature gradient distribution contains not only spatial distribution information but also three types of constraint information: direction, magnitude, and uniformity. As a result, it always converges around the same physical objective during the subsequent optimization process.

[0059] In the process of mapping the target temperature gradient distribution to the temperature gradient evolution data in the temperature gradient manipulation database and outputting the temperature gradient evolution sequence through a deep prediction model, the server first retrieves a subset of samples similar to the target temperature gradient distribution from the temperature gradient manipulation database. The laser-coded tensor, temperature gradient evolution data, and hidden field quantities from each sample are then loaded together for training or fine-tuning the deep prediction model. The deep prediction model employs a network architecture with spatial attention structure and physical constraints, using the laser-coded tensor as the primary input feature and the hidden field quantities as auxiliary features or conditional inputs. The spatial attention structure assigns higher attention weights to the spatial locations corresponding to the target region within the network, making the model more attentive to the temperature gradient evolution behavior of the target region when learning the mapping relationship. During training, the server constructs a loss function with the temperature gradient evolution data as the supervision signal, which can be written as a superposition of data fitting terms and target constraint terms. in, The total loss function during the training of a deep prediction model is a scalar that needs to be minimized using gradient descent. This represents the discrete time step index, used to enumerate different time slices in the temperature gradient evolution sequence; Represents the spatial index of the surface mesh node, used to enumerate discrete locations on the surface of the magnetic material; This indicates that the deep prediction model, under the current network parameters, is effective for index 0. Time step is The predicted temperature gradient is part of the temperature gradient evolution sequence output by the model; This represents the real temperature gradient evolution data obtained from simulations using a cross-scale photothermal-magnetic fully coupled model in the temperature gradient manipulation database, and is used as a supervision label; This represents the set of indices of all surface mesh nodes within the target area, used to limit the second loss term to be accumulated only within the target area; Indicates an index in the target temperature gradient distribution. The preset target temperature gradient vector is used to converge the model output towards the target direction and magnitude; The target constraint weight coefficient is a non-negative real number used to balance the model's accuracy in fitting the simulation data with its degree of matching the target temperature gradient distribution. By minimizing this loss function, the deep prediction model continuously adjusts in the parameter space, enabling it to output a temperature gradient evolution sequence that is both close to the simulation data and biased towards the target temperature gradient distribution, given the laser-coded tensor and the implicit field.

[0060] After the model training is completed, the server feeds any candidate laser-coded tensor as an input feature into the depth prediction model during the actual solution phase. The depth prediction model no longer relies on the stepwise solution of the cross-scale photothermal-magnetic fully coupled model and can directly output the corresponding temperature gradient evolution sequence. The server then provides a rapid evaluation basis for subsequent sensitivity analysis and laser-coded tensor optimization based on the deviation between the output sequence and the target temperature gradient distribution. Thus, it achieves efficient prediction and control of the target temperature gradient distribution on the surface of magnetic materials while ensuring physical rationality.

[0061] S150. Under the constraint of the target temperature gradient distribution, the sensitivity distribution is calculated using a cross-scale photothermal-magnetic fully coupled model to generate the reverse update direction, and the unique optimal laser coding tensor is determined based on the reverse update direction.

[0062] Specifically, calculating the sensitivity distribution refers to systematically analyzing the "sensitivity" of the temperature gradient response to each degree of freedom of the laser-coded tensor within the framework of a multi-scale photothermal-magnetic fully coupled model. This involves analyzing the impact of minute changes in the intensity, wavelength, or phase parameters of a single laser spot cell on the temperature gradient distribution within the target region, and then spatially mapping and quantifying this impact. Sensitivity distribution can be performed on a single parameter, such as examining the effect of a change in the intensity of a single laser spot cell on the amplitude of the temperature gradient in a certain region, or it can be performed on a combination of parameters, such as simultaneously examining the effect of changes in the combined wavelengths of multiple laser spot cells on the overall temperature gradient direction field.

[0063] The reverse update direction is a vector direction derived from the sensitivity distribution in the parameter space, indicating a movement towards reducing the target deviation. It guides the adjustment of the laser-coded tensor to make the temperature gradient distribution output by the multi-scale photothermal-magnetic fully coupled model closer to the target temperature gradient distribution. Intuitively, the reverse update direction can be understood as follows: Since there is a deviation between the current temperature gradient distribution and the target temperature gradient distribution, the sensitivity distribution tells us which parameter changes have a greater impact on the deviation, and the reverse update direction integrates these influences to provide a path for overall adjustment. The unique optimal laser-coded tensor refers to the laser-coded tensor configuration that, under the premise that both the target temperature gradient distribution constraint and the physical constraints of the multi-scale photothermal-magnetic fully coupled model are satisfied, is ultimately converged through iterative optimization, and is optimal and can be considered a unique solution under preset evaluation indicators. The "optimal" here is reflected in multiple dimensions, such as the smallest deviation in the temperature gradient direction, the smallest error in the temperature gradient magnitude, and the best temperature gradient uniformity. At the same time, it may also take into account indicators such as energy utilization efficiency, optical field complexity, or device safety. "Uniqueness" does not mean a globally unique solution in a mathematical sense, but rather a stable convergent solution obtained by iterative search in the reverse update direction under a given model, given initial conditions, given optimization strategy, and given physical constraints. It has determinism and repeatability and will not jump to a completely different configuration due to small perturbations.

[0064] Furthermore, the server first loads a pre-defined target temperature gradient distribution onto all grid nodes on the surface of the magnetic material, and denotes the target temperature gradient vector at each surface grid node as... ,in This represents the spatial position index located on the surface of the magnetic material; then, under the current laser-encoded tensor configuration, a complete solution for the optical, thermal, and magnetic fields is obtained through a cross-scale fully coupled optical-thermal-magnetic model, yielding the lattice temperature gradient field under the same time-scale discrete conditions. The server then constructs an objective function to measure the deviation between the actual temperature gradient distribution and the target temperature gradient distribution. It aggregates time- and space-related deviations using a weighted sum of squares. For example, the objective function can be constructed as follows: in, This represents the current laser-coded tensor parameter vector. The objective function value is a non-negative real number used to comprehensively characterize the overall deviation between the actual temperature gradient evolution on the surface of the magnetic material and the target temperature gradient distribution. This represents a high-dimensional parameter vector formed by concatenating the intensity parameters, wavelength parameters, and phase parameters of all light spot units in a fixed order, with each dimension corresponding to a certain type of parameter of a certain light spot unit; The grid position index on the surface of the magnetic material is a two-dimensional or three-dimensional discrete coordinate used to identify the temperature gradient sampling location; This represents the set of surface meshes corresponding to the target region, where errors are accumulated only within that region. This represents the discrete time step index, used to enumerate each time step in the temperature gradient evolution sequence; Indicates the location and time step The weighting coefficients at each location are used to emphasize the contribution of certain locations or time periods to the overall deviation according to task requirements. For example, greater weight can be assigned within the steady-state time window. This indicates that, given a laser-coded tensor parameter vector Under these conditions, the lattice temperature gradient vector calculated by the cross-scale photothermal-magnetic fully coupled model is the actual response quantity evaluated in the objective function; This represents a pre-defined target temperature gradient vector that does not change over time within the target region and is used to align the actual temperature gradient direction and magnitude.

[0065] After the server completes the objective function construction, in order to obtain the sensitivity distribution, it makes a small perturbation to each type of parameter in the laser coding tensor, and for each parameter... Apply positive and negative perturbations The temperature gradient evolution process was re-solved in the multi-scale photothermal-magnetic fully coupled model, and the perturbed objective function values ​​were obtained respectively. and The partial derivative of this parameter with respect to the objective function is approximated using a symmetric difference method: in, Indicates the first The sensitivity value of each parameter is the objective function relative to the parameters. The numerical approximation of the trend is used to measure the magnitude and direction of the effect of a small change in this parameter on the overall temperature gradient deviation while keeping other parameters constant. Represents the laser-coded tensor parameter vector The first in Each component can be the light intensity parameter, wavelength parameter, or phase parameter of a certain light spot unit; This represents the magnitude of a small perturbation applied to the parameter, a small real number used to trigger a detectable change in the objective function without violating the linear approximation. After the server performs the above perturbation calculation on all parameters, it sets the sensitivity value... By reorganizing the light spot units and parameter types, we can obtain the sensitivity distribution that reflects the contribution of each light spot unit's intensity parameter, wavelength parameter, and phase parameter to the target temperature gradient deviation. Geometrically, this can be understood as an approximation of the gradient of the objective function in a high-dimensional parameter space.

[0066] After the sensitivity distribution is calculated, the server uses the sensitivity distribution to determine the inverse update direction that can reduce the target temperature gradient bias. The target temperature gradient bias has already been calculated in the objective function. Encoding was performed, and the sensitivity distribution... The partial derivative approximation with respect to each parameter provides information on whether the objective function increases or decreases if the parameter changes in a certain direction. To update the parameters along the direction that decreases the objective function, the server constructs a reverse update direction vector. This can be achieved using negative gradient normalization:

[0067] in, Indicates the reverse update direction vector In the The component in each parameter dimension is a dimensionless directional coefficient used to indicate whether the parameter should increase or decrease in the next update; This indicates the sensitivity value obtained in the previous section, and the sign determines the direction of the objective function's response to the increment of this parameter; This represents the absolute value of sensitivity, used for amplitude normalization of sensitivity; Represents a preset positive decimal to avoid [in...]. When the value approaches zero, it becomes unstable or the denominator becomes zero. Through this construction method, when the absolute value of the sensitivity of a certain parameter is large, the corresponding... Larger, after normalization The parameter still maintains a relatively large influence weight, indicating that the spot cell corresponding to this parameter contributes strongly to the target temperature gradient deviation and needs to be adjusted first. When the absolute value of the sensitivity of a certain parameter is very small, the normalized value... A value close to zero indicates that this parameter has a weak impact on the objective function, and its weight can be relatively small during updates. Because... The negative sign was introduced earlier; if the sensitivity of a certain parameter... A positive value indicates that increasing this parameter will increase the objective function. A negative sensitivity indicates that during updates, the system will tend to decrease the parameter, thus moving towards reducing the objective function. Conversely, if the sensitivity of a parameter is negative, it means that increasing the parameter is beneficial for reducing the objective function. If the value is positive, updates will move in the direction that increases this parameter. The server will... Combined in a predetermined order, they form a reverse update direction vector. This vector provides a joint adjustment direction in the parameter space that comprehensively considers the light intensity parameters, wavelength parameters and phase parameters of all light spot units, so that the target temperature gradient deviation can be reduced as a whole when stepping along this direction.

[0068] Obtain the reverse update direction vector Subsequently, the server iteratively updates the laser-coded tensor in the parameter space until convergence, thereby determining the unique optimal laser-coded tensor under the constraint of the target temperature gradient distribution. During the iterative update process, at the... The laser-coded tensor parameter vector at the next iteration time is denoted as... The server updates according to the current reverse direction. and step size coefficient The update yields the parameter vector for the next iteration:

[0069] in, Indicates the first The laser coding tensor parameter vector after the next iteration is the new parameter configuration obtained after the current iteration step is completed; Indicates the first The laser coding tensor parameter vector at the next iteration is the parameter configuration before the update; Indicates the first The step size coefficient in the next iteration is a positive real number used to control the magnitude of movement along the reverse update direction. It can adopt a fixed step size strategy or an adaptive step size strategy based on the change of the objective function. For example, the objective function value under multiple candidate step sizes can be calculated tentatively and the step size that makes the objective function decrease the fastest can be selected. Indicates the first The inverse update direction vector calculated based on the current sensitivity distribution in each iteration is the descent direction in the parameter space. After each update, the server re-invokes the cross-scale photothermal-magnetic fully coupled model, using the new laser coding tensor parameter vector. Solve the temperature gradient evolution process and calculate the new objective function value. The process continues to calculate the corresponding sensitivity distribution and reverse update direction. The iterative process terminates when a preset convergence condition is met, for example, when the change in the objective function value between two adjacent iterations satisfies the following:

[0070] Or when the objective function value itself has already fallen below the target threshold, that is: in, The preset convergence criterion threshold is a positive decimal representing the tolerance range of the objective function. The upper bound of the desired objective function is a non-negative real number determined by application requirements. When the objective function value is less than this value, the actual temperature gradient distribution can be considered sufficiently close to the target temperature gradient distribution. When any condition is met, the server will generate the laser-coded tensor parameter vector obtained in the current iteration. The unique optimal laser-coded tensor, labeled under the constraint of the target temperature gradient distribution, has a set of parameters that have definite convergence and repeatability under given initial conditions, physical models and optimization strategies. It can stably reproduce the temperature gradient evolution behavior that is highly consistent with the target temperature gradient distribution in a cross-scale photothermal-magnetic fully coupled model, and can be directly sent to a laser spatial modulation array for engineering applications.

[0071] S160. The unique optimal laser coding tensor is imported into the cross-scale photothermal-magnetic fully coupled model for verification. Under the condition of satisfying the preset deviation tolerance, the unique optimal laser coding tensor is applied to the laser spatial modulation array to generate a temperature gradient distribution on the surface of the magnetic material that is consistent with the target temperature gradient distribution.

[0072] Specifically, the preset deviation tolerance condition sets a quantitative tolerance range for "how much difference is allowed between the model output temperature gradient distribution and the target temperature gradient distribution," serving as a bridge between the theoretical optimal solution and the engineering feasible solution. Due to errors in actual material properties, deviations in processing and assembly, and noise in sensing and measurement, it is impossible to require the model output temperature gradient to be strictly equal to the target temperature gradient at every spatial location and time step. Instead, a deviation tolerance needs to be preset, such as the temperature gradient direction deviating by no more than a certain angle range within the target region, the relative error of the temperature gradient amplitude not exceeding a certain percentage, and the uniformity index not falling below a certain threshold. During the verification phase, the server judges the simulation results based on these tolerance conditions. If all indicators fall within the tolerance range, the uniquely optimal laser coding tensor is considered to have passed model verification and can be used to drive the actual laser spatial modulation array. For example, it can be required that "when the target temperature gradient is a certain value, the deviation of the actual simulation result across the entire region does not exceed five percent." Once this is met, it is considered that the preset deviation tolerance condition has been met.

[0073] Furthermore, after the server completes the solution of the unique optimal laser coding tensor, when performing physical consistency verification on this unique optimal laser coding tensor in the simulation environment, the cross-scale photothermal-magnetic fully coupled model maintains the same three-dimensional geometry, material property parameters, boundary conditions, and time stepping strategy as the optimization stage. Only the laser excitation input in the model is replaced with the light intensity parameters, wavelength parameters, and phase parameters corresponding to each spot unit in the unique optimal laser coding tensor. Under this configuration, the server restarts the cross-scale photothermal-magnetic fully coupled model solution process, acquiring the lattice temperature field and calculating the temperature gradient vector at each location on the magnetic material surface at each time step. This process can be represented as a mapping from the unique optimal laser coding tensor parameter vector to the target temperature gradient evolution data, which can be written as:

[0074] in, This represents the unique optimal laser coding tensor parameter vector. Under the influence of the magnetic material, the lattice temperature gradient evolution data obtained by the cross-scale photothermal-magnetic fully coupled model is defined at the position on the surface of the magnetic material. and discrete time step The temperature gradient vector function is used to characterize the evolution of the target temperature gradient data over time. This represents the multiphysics operator corresponding to the cross-scale photothermal-magnetic fully coupled model, used to integrate the coupled solution process of optical, thermal and magnetic fields; The spatial position index on the surface of a magnetic material is a two-dimensional or three-dimensional discrete coordinate defined on the surface grid. The time step index is determined according to the time stepping strategy. The discrete time points obtained by partitioning; The vector representation of the unique optimal laser coding tensor in parameter space is a high-dimensional vector obtained by concatenating the intensity parameters, wavelength parameters and phase parameters of all light spot units in a fixed order. This represents a set of three-dimensional geometric configurations used to describe the geometric dimensions and relative positions of the silicon substrate, cobalt layer, and platinum layer. It represents a set of material properties parameters, including optical absorption parameters, electronic energy relaxation parameters, phonon thermal conductivity parameters, and magnetic anisotropy parameters, etc. This represents the set of boundary and initial condition configurations used to describe thermal flow boundaries, convective boundaries, and initial temperature and initial magnetization states, etc. The time stepping strategy is used to specify the time discretization resolution during the simulation process. Using this method, the server generates a set of target temperature gradient evolution data consistent with the optimization stage, based on a uniquely optimal laser-coded tensor, without changing any physical constraints, thus providing a foundation for subsequent deviation determination.

[0075] After obtaining the target temperature gradient evolution data, the server compares this data point-by-point with a pre-defined target temperature gradient distribution to evaluate whether the actual response of the uniquely optimal laser-coded tensor in the physical model meets the preset deviation tolerance conditions. To this end, the server performs a point-by-point comparison on the target region of the magnetic material surface. Inside, for each surface mesh location and each selected time step Calculate the target temperature gradient distribution Temperature gradient evolution data obtained from simulation By analyzing the error vector between the two sides and using the norm of the error vector to quantitatively describe the deviation, we can construct the maximum deviation index and the root mean square deviation index.

[0076] in, It represents the maximum temperature gradient deviation over the entire target region and the entire time window, and is used to characterize the upper bound of the error at the most unfavorable spatial and temporal points. This represents the root mean square temperature gradient deviation over the target region and time window, reflecting the average deviation level in an overall sense. Represents the surface location in the target temperature gradient distribution. The pre-planned target temperature gradient vector is the ideal temperature gradient that is desired to be achieved during the optimization process; This represents the set of nodes on the surface mesh representing the target region, and the deviation integral is performed only on the mesh points within this region; This indicates the number of time steps involved in the statistics, which is the number of discrete time points within the selected time window. This represents the total number of surface mesh nodes within the target area. The server will... and It is compared with a threshold given by a preset deviation tolerance condition, such as the maximum deviation tolerance threshold. and root mean square deviation tolerance threshold When the following conditions are met:

[0077] If the deviations between the target temperature gradient evolution data generated by the unique optimal laser-coded tensor in the multi-scale photothermal-magnetic fully coupled model and the target temperature gradient distribution in terms of direction, amplitude and spatial uniformity are all within acceptable ranges, then the preset deviation tolerance condition is considered to have been met, and the unique optimal laser-coded tensor can be transferred from the simulation space to the actual hardware implementation stage.

[0078] After confirming that the preset deviation tolerance conditions are met, the server maps the intensity, wavelength, and phase parameters in the uniquely optimal laser-coded tensor to corresponding spot cell driving commands, and sends these driving commands to the laser spatial modulation array to reconstruct the optical field structure consistent with the simulation on the physical hardware, thereby generating a temperature gradient distribution on the magnetic material surface that is consistent with the target temperature gradient distribution. During the mapping process, the server first indexes the parameters inside the laser-coded tensor according to the spot cell spatial index. Organize the data and assign light intensity parameters to each spatial index. Wavelength parameters and phase parameters The digital-to-analog conversion voltage value, switching duty cycle, or phase delay command converted into control hardware can be abstracted as a control vector:

[0079] in, Indicates spatial index as The beam unit drive command vector is a set of hardware control quantities sent to the corresponding pixel or micromirror in the laser spatial modulation array. It can include amplitude channel commands, wavelength channel selection commands, and phase modulation commands, etc. The light intensity parameter of the spot cell in the unique optimal laser coding tensor is the target light power density setting value determined during the optimization process. The wavelength parameter of the light spot unit is the emission wavelength setting value selected in a multi-source or tunable light source system; The phase parameter of the light spot unit is the phase offset setting value optimized to form the desired interference structure; This represents the set of hardware calibration parameters used to describe hardware characteristics such as the gain of the digital-to-analog converter, the grayscale response curve of the spatial light modulator, and the voltage-phase mapping characteristics of the phase modulator. (Function) These calibration parameters are used to convert ideal physical quantities into practically quantifiable control codewords. The server will handle all of these. The control frame is composed of light spot units arranged in spatial index order and sent to the laser spatial modulation array controller via a high-speed communication interface. The controller configures the output state of each light spot unit according to the received drive commands, so that the actual light intensity distribution, wavelength combination, and phase superposition structure on the magnetic material surface are consistent with those in the simulation stage. Since the unique optimal laser coding tensor has been verified for deviation tolerance in the cross-scale photothermal-magnetic fully coupled model, it can be expected that under the action of this light field structure, the temperature gradient distribution naturally formed on the magnetic material surface through optical absorption and thermal diffusion will statistically match the target temperature gradient distribution very closely. Thus, accurate reconstruction of the temperature gradient from theoretical design, numerical verification to engineering implementation can be achieved without additional physical parameter tuning.

[0080] This application also provides a laser-controlled magnetic material surface temperature gradient generation device, referring to... Figure 2 , Figure 2 This application provides a schematic diagram of a laser-controlled magnetic material surface temperature gradient generation device. The device is a server, comprising an acquisition module 21 and a processing module 22. The acquisition module 21 acquires the optical absorption parameters, electron energy relaxation parameters, phonon thermal conductivity parameters, and magnetic anisotropy parameters of the magnetic material, and models them in three-dimensional space based on a silicon substrate, a cobalt layer, and a platinum layer to construct a cross-scale photothermal-magnetic fully coupled model. The acquisition module 21 also acquires the light intensity parameters, wavelength parameters, and phase parameters of a laser spatial modulation array composed of multiple light spot units, and encodes them into a laser-encoded tensor in spatial order. The processing module 22 reads the laser-encoded tensor in the simulation environment and simultaneously drives the cross-scale photothermal-magnetic fully coupled model to perform batch solutions to generate temperature gradient evolution data. The processing module 22 is used to set the target temperature gradient distribution on the surface of the magnetic material and output the temperature gradient evolution sequence through a depth prediction model based on the temperature gradient manipulation database. The processing module 22 is also used to calculate the sensitivity distribution using a cross-scale photothermal-magnetic fully coupled model under the constraint of the target temperature gradient distribution, to generate a reverse update direction, and to determine a unique optimal laser coding tensor based on the reverse update direction. The processing module 22 is also used to import the unique optimal laser coding tensor into the cross-scale photothermal-magnetic fully coupled model for verification, and, under the condition of satisfying a preset deviation tolerance, apply the unique optimal laser coding tensor to a laser spatial modulation array to generate a temperature gradient distribution on the surface of the magnetic material that is consistent with the target temperature gradient distribution.

[0081] It should be noted that the above embodiments of the apparatus are only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0082] This application also provides an electronic device, with reference to... Figure 3 , Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include: at least one processor 31, at least one network interface 34, a user interface 33, a memory 35, and at least one communication bus 32.

[0083] The communication bus 32 is used to enable communication between these components.

[0084] The user interface 33 may include a display screen and a camera. Optionally, the user interface 33 may also include a standard wired interface and a wireless interface.

[0085] The network interface 34 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0086] The processor 31 may include one or more processing cores. The processor 31 connects to various parts of the server via various interfaces and lines, executing instructions, programs, code sets, or instruction sets stored in the memory 35, and calling data stored in the memory 35 to perform various server functions and process data. Optionally, the processor 31 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 31 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content to be displayed on the screen; and the modem handles wireless communication. It is understood that the modem may also not be integrated into the processor 31 and may be implemented as a separate chip.

[0087] The memory 35 may include random access memory (RAM) or read-only memory. Optionally, the memory 35 may include a non-transitory computer-readable storage medium. The memory 35 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 35 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 35 may also be at least one storage device located remotely from the aforementioned processor 31. Figure 3 As shown, the memory 35, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program based on a laser-controlled method for generating surface temperature gradients of magnetic materials.

[0088] exist Figure 3 In the electronic device shown, the user interface 33 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 31 can be used to call an application program stored in the memory 35 that is a method for generating a surface temperature gradient of a magnetic material based on laser control. When executed by one or more processors, the electronic device performs one or more methods as described in the above embodiments.

[0089] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0090] This application also provides a non-transitory computer-readable storage medium storing instructions. When executed by one or more processors, these instructions cause an electronic device to perform one or more of the methods described in the above embodiments.

[0091] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0092] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual couplings or direct couplings or communication connections may be through some service interfaces; indirect couplings or communication connections between apparatuses or units may be electrical or other forms.

[0093] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0094] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0095] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0096] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of other embodiments of this disclosure upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for generating a surface temperature gradient of a magnetic material based on laser control, characterized in that, The method includes: The optical absorption parameters, electronic energy relaxation parameters, phonon thermal conductivity parameters, and magnetic anisotropy parameters of magnetic materials are obtained, and a three-dimensional model is constructed based on silicon substrate, cobalt layer, and platinum layer to build a cross-scale photothermal-magnetic fully coupled model. The intensity parameters, wavelength parameters, and phase parameters of a laser spatial modulation array composed of multiple light spot units are obtained and encoded into a laser coded tensor in spatial order. The laser-encoded tensor is read in the simulation environment, and the cross-scale photothermal-magnetic fully coupled model is driven to perform batch solving to generate temperature gradient evolution data and implicit field quantities. A temperature gradient manipulation database is constructed based on the temperature gradient evolution data and the implicit field quantities. A target temperature gradient distribution on the surface of a magnetic material is defined, and a temperature gradient evolution sequence is output through a deep prediction model based on the temperature gradient manipulation database. Under the constraint of the target temperature gradient distribution, the sensitivity distribution is calculated using the cross-scale photothermal-magnetic fully coupled model to generate the reverse update direction, and the unique optimal laser coding tensor is determined based on the reverse update direction. The unique optimal laser coding tensor is imported into the cross-scale photothermal-magnetic fully coupled model for verification. Under the condition of satisfying the preset deviation tolerance, the unique optimal laser coding tensor is applied to the laser spatial modulation array to generate a temperature gradient distribution on the surface of the magnetic material that is consistent with the target temperature gradient distribution.

2. The method for generating a surface temperature gradient of a magnetic material based on laser control according to claim 1, characterized in that, The process involves acquiring the optical absorption parameters, electron energy relaxation parameters, phonon thermal conductivity parameters, and magnetic anisotropy parameters of the magnetic material, and modeling it in three-dimensional space based on a silicon substrate, cobalt layer, and platinum layer to construct a cross-scale photothermal-magnetic fully coupled model, specifically including: The optical absorption parameters, electronic energy relaxation parameters, phonon thermal conductivity parameters, and magnetic anisotropy parameters of the magnetic material were obtained by ellipsometry measurement, femtosecond pump-probe experiment, thermal reflection experiment, and magneto-optical Kerr effect test, respectively. When modeling a multilayer structure of silicon substrate, cobalt layer and platinum layer in three-dimensional space, the optical absorption parameters, the electronic energy relaxation parameters, the phonon thermal conductivity parameters and the magnetic anisotropy parameters are respectively bound to the optical field solution sub-model, thermal field solution sub-model and magnetic field solution sub-model of the corresponding layer structure to construct the cross-scale photothermal-magnetic fully coupled model.

3. The method for generating a surface temperature gradient of a magnetic material based on laser control according to claim 1, characterized in that, The process of acquiring the intensity parameters, wavelength parameters, and phase parameters of a laser spatial modulation array composed of multiple light spot units, and encoding them into a laser coded tensor in spatial order, specifically includes: The incident laser beam is spatially divided by a laser spatial modulation array with programmable optical modulation capabilities to form a set of spot units with unique spatial coordinate indices. Optical power measurement, spectral measurement and interferometry are performed on each spot unit in the set of spot units to obtain the corresponding light intensity parameters, wavelength parameters and phase parameters; After binding the light intensity parameter, the wavelength parameter, and the phase parameter with the corresponding light spot unit spatial coordinate index, they are organized into a laser coding tensor according to a preset spatial order. The laser coding tensor is used to characterize the light field structure of the laser spatial modulation array.

4. The method for generating a surface temperature gradient of a magnetic material based on laser control according to claim 1, characterized in that, The process of reading the laser-encoded tensor in the simulation environment and simultaneously driving the multi-scale photothermal-magnetic fully coupled model to perform batch solutions to generate temperature gradient evolution data and implicit field quantities, and constructing a temperature gradient manipulation database based on the temperature gradient evolution data and the implicit field quantities, specifically includes: The light intensity parameters, wavelength parameters, and phase parameters in the laser coding tensor are mapped to the energy deposition load unit, optical absorption unit, and phase superposition unit of the cross-scale photothermal-magnetic fully coupled model, respectively. The illumination positions of multiple light spot units are matched with the model grid nodes by a grid discretization method, and the electronic temperature field, lattice temperature field, and magnetic domain response are solved in the same time step. Continuous temperature gradient evolution data are generated based on the electronic temperature field, lattice temperature field, and magnetic domain response. The interface energy deposition rate, effective heat flux density, and magnetic domain spin fluctuation intensity are extracted from the laser-coded tensor as implicit field quantities, and the temperature gradient evolution data and the implicit field quantities are combined to form the temperature gradient manipulation database.

5. The method for generating a surface temperature gradient of a magnetic material based on laser control according to claim 1, characterized in that, The process of setting the target temperature gradient distribution on the surface of the magnetic material and outputting the temperature gradient evolution sequence through a deep prediction model based on the temperature gradient manipulation database specifically includes: At least one target region is defined on the surface of a magnetic material, and temperature gradient direction constraints, temperature gradient magnitude constraints, and temperature gradient uniformity constraints are set for the target region to obtain the target temperature gradient distribution. The target temperature gradient distribution is mapped to the temperature gradient evolution data in the temperature gradient manipulation database, and the laser-coded tensor is used as an input feature to introduce a depth prediction model with spatial attention structure and physical constraints. The depth prediction model outputs the temperature gradient evolution sequence corresponding to the laser-coded tensor.

6. The method for generating a surface temperature gradient of a magnetic material based on laser control according to claim 1, characterized in that, Under the constraint of the target temperature gradient distribution, the sensitivity distribution is calculated using the cross-scale photothermal-magnetic fully coupled model to generate a reverse update direction, and a unique optimal laser coding tensor is determined based on the reverse update direction. Specifically, this includes: The target temperature gradient distribution is used as a constraint condition and input into the cross-scale photothermal-magnetic fully coupled model. The temperature gradient changes caused by small perturbations of the light intensity parameters, wavelength parameters and phase parameters corresponding to each spot unit of the laser coding tensor are solved one by one to form a sensitivity distribution that reflects the degree of contribution of each spot unit to the temperature gradient. Based on the sensitivity distribution, a reverse update direction that can reduce the target temperature gradient deviation is determined; The laser-coded tensor is iteratively updated in the reverse update direction until convergence, so as to determine the unique optimal laser-coded tensor under the constraint of the target temperature gradient distribution.

7. The method for generating a surface temperature gradient of a magnetic material based on laser control according to claim 1, characterized in that, The process of importing the unique optimal laser coding tensor into the cross-scale photothermal-magnetic fully coupled model for verification, and applying the unique optimal laser coding tensor to the laser spatial modulation array under the condition of satisfying the preset deviation tolerance, to generate a temperature gradient distribution on the magnetic material surface that is consistent with the target temperature gradient distribution, specifically includes: While maintaining consistency in three-dimensional geometry, material properties, boundary conditions, and time stepping strategy, the light intensity parameters, wavelength parameters, and phase parameters corresponding to the unique optimal laser coding tensor are replaced with the laser excitation input in the cross-scale photothermal-magnetic fully coupled model to generate target temperature gradient evolution data of the magnetic material surface. The target temperature gradient evolution data is compared point by point with the target temperature gradient distribution to determine whether the preset deviation tolerance condition is met. When the preset deviation tolerance condition is met, the unique optimal laser coding tensor is mapped to the corresponding spot unit driving instruction and sent to the laser spatial modulation array to control the laser spatial modulation array to generate a temperature gradient distribution on the surface of the magnetic material that is consistent with the target temperature gradient distribution.

8. A device for generating a surface temperature gradient of a magnetic material based on laser control, characterized in that, The apparatus is used to perform the method for generating a surface temperature gradient of a magnetic material as described in any one of claims 1 to 7, the apparatus comprising an acquisition module and a processing module, wherein... The acquisition module is used to acquire the optical absorption parameters, electronic energy relaxation parameters, phonon thermal conductivity parameters and magnetic anisotropy parameters of the magnetic material, and to build a cross-scale photothermal-magnetic fully coupled model based on the silicon substrate, cobalt layer and platinum layer in three-dimensional space. The acquisition module is also used to acquire the light intensity parameters, wavelength parameters and phase parameters of the laser spatial modulation array composed of multiple light spot units, and encode them into a laser coded tensor in spatial order; The processing module is used to read the laser-encoded tensor in the simulation environment, and simultaneously drive the cross-scale photothermal-magnetic fully coupled model to perform batch solving to generate temperature gradient evolution data and implicit field quantities, and to construct a temperature gradient manipulation database based on the temperature gradient evolution data and the implicit field quantities. The processing module is also used to set the target temperature gradient distribution on the surface of the magnetic material, and output the temperature gradient evolution sequence through a deep prediction model based on the temperature gradient manipulation database. The processing module is further configured to calculate the sensitivity distribution using the cross-scale photothermal-magnetic fully coupled model under the constraint of the target temperature gradient distribution, so as to generate the reverse update direction, and determine the unique optimal laser coding tensor based on the reverse update direction. The processing module is further configured to import the unique optimal laser coding tensor into the cross-scale photothermal-magnetic fully coupled model for verification, and, under the condition of satisfying the preset deviation tolerance, apply the unique optimal laser coding tensor to the laser spatial modulation array to generate a temperature gradient distribution on the surface of the magnetic material that is consistent with the target temperature gradient distribution.

9. An electronic device, characterized in that, The electronic device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions. The user interface and the network interface are both used to communicate with other devices. The processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium stores instructions that, when executed, perform the method as described in any one of claims 1 to 7.

Images