A magnonic random number generator, magnonic probabilistic computing integrated device and a probability generation method

Abstract

The application belongs to the field of magnetic devices, and relates to a magnonic random number generator, a magnonic probability calculation integrated device and a probability generation method. The magnonic random number generator comprises a pulse input module, a random number generation module and an output module, and the input end of the random number generation module and the output end of the random number generation module are connected with the pulse input module and the output module through an adhesive layer. The random number generation module is used for generating random pulses with the bistable characteristic of strong nonlinear effect under the condition of receiving pulse signals from the pulse input module and receiving the random pulses by the output module. The random number generation module is a magnetic waveguide or a thin film with a high nonlinear frequency shift coefficient and low damping deposited on an insulating substrate. The random number generated by the magnonic random number generator can reduce energy consumption, improve the generation rate and thermal stability of the random number, and can be further used for probability calculation of the random number.

Description

Technical Field

[0001] This invention belongs to the technical field of magnetic devices, and more specifically, relates to a magnetic random number generator, a magnetic probability calculation integrated device, and a probability generation method. Background Technology

[0002] Probabilistic computation takes into account the inherent uncertainties present in many real-world scenarios and is widely used in problems such as Bayesian learning, protein folding, optimization, and cryptography, as well as in artificial intelligence based on neural network computation. The core of probabilistic computation is the random number generator, which plays a crucial role in the training and optimization of neural networks.

[0003] Most transistor-based random number generators are pseudo-random number generators. Pseudo-random number generators use deterministic algorithms to produce seemingly random sequences of numbers, but in reality, their output sequences are entirely determined by the initial seed and the algorithm, making them repeatable and detrimental to data confidentiality. In contrast, true random number generators (TRNGs) rely on the unpredictability of physical processes, thus providing more reliable and non-replicable randomness.

[0004] To improve data security, various true random number generators based on different physical random processes have been developed. For example, memristor-based random number generators primarily utilize the probabilistic nature of particle migration in multilayer thin films to achieve randomness in the output voltage; they are simple in structure and have high storage density. Photonic quantum effect random number generators typically utilize the uncertainty principle of quantum mechanics, which is used to ensure that the generated random numbers are truly random. Magnetic tunnel junction-based random number generators work by reducing the energy difference between parallel and antiparallel magnetic moments in the magnetic tunnel junction, allowing thermal perturbation to overcome the flipping energy difference and thus achieve random flipping of the magnetic moments. By controlling temperature, applied magnetic fields, and energy differences, magnetic tunnel junction random number generators can produce GHz-frequency random numbers. Some studies have demonstrated their advantages over traditional computers in areas such as large prime factorization and combinatorial optimization, and they can even rival quantum computers.

[0005] However, the random number generators based on memristors mentioned above are limited by the particle migration rate, typically at the kHz level. Photonic quantum random number generators can generate random numbers at rates up to GHz or even THz, but they are generally large in size and have complex optical paths, making them difficult to integrate with electronic devices. Random number generators based on magnetic tunnel junctions suffer from poor thermal stability because their random flipping mechanism relies solely on thermal perturbations. Summary of the Invention

[0006] In view of the above-mentioned defects or improvement needs of the prior art, the present invention provides a magnetic random number generator, a magnetic probability calculation integrated device and a probability generation method. The random numbers generated by the magnetic random number generator can reduce energy consumption, improve the generation rate and thermal stability of random numbers, and can be further used for probability calculation of random numbers.

[0007] To achieve the above objectives, according to a first aspect of the present invention, a magnetic random number generator is provided, comprising: a pulse input module, a random number generation module, and an output module, wherein the input terminal of the random number generation module is connected to the pulse input module, and the output terminal of the random number generation module is connected to the output module through an adhesive layer;

[0008] The random number generation module is used to generate random pulses by utilizing the bistable characteristics of a strong nonlinear effect, under the condition of receiving a pulse signal from the pulse input module, and these random pulses are received by the output module. The random number generation module is a magnetic waveguide or thin film with a high nonlinear frequency shift coefficient and low damping, deposited on an insulating substrate. The high nonlinear frequency shift coefficient is 0.5–6 GHz, and the low damping coefficient is 5 × 10⁻⁶. -4 ~4×10 -3 .

[0009] Preferably, the material of the magnetic waveguide or thin film with high nonlinear frequency shift coefficient and low damping includes yttrium iron garnet ferrite material, cobalt iron boron alloy material, iron cobalt, and nickel iron alloy material.

[0010] Preferably, the random number generation module has a thickness of 10nm-10μm and a width of 100nm-100μm.

[0011] Preferably, both the pulse input module and the output module are metal electrode layers, selected from Au, Cu, or Pt electrodes, with a width of 2-5 μm and a thickness of 100-500 nm.

[0012] Preferably, the material of the adhesive layer is selected from Ti, Ta or Cr, with a width of 2-5 μm and a thickness of 3-10 nm.

[0013] Preferably, the random number generation module is fabricated by forming a magnetic thin film with a high nonlinear frequency shift coefficient and low damping on the insulating substrate using a thin film deposition process, and then forming a magnetic waveguide or thin film with a high nonlinear frequency shift coefficient and low damping using an electron beam exposure and etching process.

[0014] Preferably, both the pulse input module and the output module are fabricated by depositing a metal electrode layer on the random number generation module using a lift-off process.

[0015] According to a second aspect of the present invention, a magnetic probability calculation integrated device is provided, comprising a plurality of sub-magnetic random number generators coupled to a main magnetic random number generator, wherein the output module of the sub-magnetic random number generator serves as the pulse input module of the main magnetic random number generator; wherein the sub-magnetic random number generators and the main magnetic random number generator are magnetic random number generators as described above.

[0016] According to a third aspect of the present invention, a probability generation method is provided, which operates using the above-described magnetic particle random number generator, comprising: applying a microwave pulse current signal to a pulse generation module, switching a magnetic particle between a ground state and an excited state, then transmitting a spin wave signal of the excited state in a waveguide or thin film of the random number generation module, outputting from an output module and detecting the output magnetic particle probability.

[0017] According to a fourth aspect of the present invention, a probability generation method is provided, employing a magnetic probability calculation integrated device as described above, comprising: simultaneously applying a microwave pulse current signal to the pulse generation module of each sub-magnetic random number generator, outputting and detecting magnetic probabilities from the output module of the main magnetic random number generator; wherein the output probability detected by the output module of the magnetic probability calculation integrated device is the product of the output magnetic probabilities of each sub-magnetic random number generator.

[0018] In summary, compared with the prior art, the above-described technical solutions conceived by this invention mainly possess the following technical advantages:

[0019] 1. The pulse input module used in the magnetic random number generator of the present invention is a magnetic waveguide or thin film with a high nonlinear frequency shift coefficient and low damping, deposited on an insulating substrate. The input end of this material is connected to the pulse input module, and the output end is connected to the output module. After the random number generation module receives the pulse signal, it can generate random pulses and transmit them to the output module by virtue of the bistable characteristics of the strong nonlinear effect. It is worth noting that the output magnetic pulses do not need to be converted into electronic form and can directly interact with other magnetic pulses, effectively reducing the number of magnetic particle-electron conversions in the device and avoiding the generation of Joule heating. This makes it easier and more feasible to construct low-power magnetic probabilistic integrated devices, providing strong support for reducing device power consumption and improving operating efficiency.

[0020] 2. The random number generation module of the present invention is preferably made of yttrium iron garnet ferrite material. In addition to meeting the requirements of high nonlinear frequency shift coefficient and low damping magnetic waveguide or thin film, the material also has good gyromagnetic effect, ultra-low damping significantly increases the spin wave propagation distance, supports multi-level logic gate cascading, strong switching effect with large nonlinear response, low saturation magnetic field makes it highly applicable in engineering, and the nanofabrication process is mature and compatible, making it particularly suitable for the preparation of the magnetic random number generator of the present invention.

[0021] 3. The magnetic random number generator of this invention features a nanometer-scale size, significantly improving integration density and demonstrating broad application prospects in large-scale production. It can meet the integration requirements of large-scale integrated circuits and magnetic circuits, reducing production costs and improving production efficiency. Simultaneously, this generator has the potential to perform AND operations at the hardware level, further expanding its application scope and value in the field of logic operations. It provides a fundamental component for building high-performance, multifunctional magnetic systems, helping to promote the development of magnetic devices towards higher integration and intelligence.

[0022] 4. The random magnetic pulses generated by the magnetic random number generator of the present invention can reach GHz. Its high-speed pulse generation capability ensures the high efficiency of random number generation, and can quickly respond and output a large number of high-quality random numbers, meeting the requirements of various application scenarios with high requirements for random number generation speed, such as key generation in cryptography and Monte Carlo simulation, and providing high-speed and stable data support for the technological development and application in related fields. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the magnetic random number generator provided by the present invention.

[0024] Figure 2 A process flow diagram for manufacturing a magnetic random number generator according to the present invention is provided for the present invention.

[0025] Figure 3 The schematic diagram of the working principle of the magnetic random number generator provided by the present invention.

[0026] Figure 4 The microwave power-probability diagram provided by this invention illustrates the probability switching diagram of magneton intensity with adjustable microwave power.

[0027] Figure 5 A schematic diagram of the structure of the magnetic probability calculation integrator made from the magnetic random number generator provided by the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0029] In a first aspect of the present invention, a schematic diagram of a magnetic random number generator is provided. (See diagram below.) Figure 1As shown, the present invention is a novel magnetic random number generator that generates random pulses based on the bistable characteristics of strong nonlinear effects. Its structure includes: a pulse input module 11, a random number generation module 12, and an output module 13.

[0030] The input terminal of the random number generation module 12 is connected to the pulse input module 11, and the output terminal is connected to the output module 13. The pulse input module 11 provides a preset pulse signal and applies it to the random number generation module 12, using two microwave pulse signals with different frequencies and powers to generate the pulse signal. The output module 13 receives random number sequences in real time and detects the random probability; it can be connected to electrical instruments such as an oscilloscope to detect the random probability in real time.

[0031] The random number generation module 12 is a magnetic waveguide or thin film deposited on an insulating substrate, possessing a high nonlinear frequency shift coefficient and low damping. This module 12 generates random pulses and outputs them to the output module 13 based on the bistable characteristics exhibited by the strong nonlinear effect, under the condition of receiving a pulse signal from the pulse input module 11. The strong nonlinear effect of the random number generation module is that, under a high-power pump pulse microwave signal, the large precession angle of the magnetic moment dominates, causing a frequency shift phenomenon. The system has two stable states (a low-amplitude ground state and a high-amplitude excited state). By imparting a certain amount of energy to the ground state through a bias pulse microwave signal to overcome the energy barrier, the magnetic particle randomly switches between the bistable states to generate random pulses.

[0032] In this invention, the materials used for magnetic waveguides or thin films with high nonlinear frequency shift coefficients and low damping include, but are not limited to, yttrium iron garnet ferrite materials, cobalt iron boron alloy materials, iron-cobalt, and nickel-iron alloy materials, which possess excellent gyromagnetic effects, low resonant linewidth, and high resistance. This magnetic material employs a straight waveguide structure design with a thickness of 10 nm-10 μm and a width of 100 nm-100 μm.

[0033] In this embodiment, the random number generation module 12 is preferably made of yttrium iron garnet ferrite material (YIG). This material has good gyromagnetic effect, ultra-low damping that significantly increases the propagation distance of spin waves, supports multi-level logic gate cascading, strong exchange action with large nonlinear response, low saturation magnetic field that makes it highly applicable in engineering, and mature and compatible nanofabrication process.

[0034] In this invention, the thickness of the random number generation module is 10nm-10μm; the width is 100nm-100μm.

[0035] In this invention, the materials of the pulse input module and the output module are both conductive metal electrode materials, selected from Au, Cu or Pt electrodes, with a width of 2-5 μm and a thickness of 100-500 nm.

[0036] In this invention, the adhesive layer has a width of 2-5 μm and a thickness of 5-10 nm. The adhesive layer material is selected from Ti, Ta or Cr to enhance the adhesion between the gold layer and the surface of the magnetic material, prevent the Au layer from peeling off due to thermal or mechanical stress, and avoid the adhesive layer atoms from penetrating into the magnetic material and affecting the magnetic properties.

[0037] In this invention, the pulse input module includes one or two microwave antennas. The invention also includes using two microwave antennas as the pulse input module, and is not limited to using a single microwave antenna as the pulse generation module as in this example. For instance, a high-power and high-frequency pump microwave pulse current signal is applied to the pulse generation module. Due to the bistable characteristic of the nonlinear effect, the magnet is in the ground state. Further application of a trigger microwave pulse current signal triggers a random switching between the magnet's ground and excited states. The core principle is to achieve the switching between the magnet's ground and excited states through the nonlinear effect of the magnet, thereby realizing a random pulse signal.

[0038] This invention provides a method for preparing a magnetic random number generator, used to prepare the aforementioned magnetic random number generator. The following will describe... Figure 2 For example, the preparation process will be explained in detail, including:

[0039] S1. Thin film layers are grown on insulating substrates using liquid phase epitaxy or magnetron sputtering thin film deposition techniques. Insulating substrates are selected and thoroughly cleaned before use. Thin film deposition is then performed on the cleaned insulating substrate, ultimately forming a thin film of uniform thickness. Liquid phase epitaxy, through single-crystal epitaxial growth, offers atomic-level interface control and sub-nanometer thickness precision. Magnetron sputtering thin film deposition, due to its high cleanliness, vacuum level, and large-area uniformity, can also be used to prepare magnetic nanolayered films. Insulating substrates such as GGG, SGGG, and SiO2 are used, with a thickness of 500-1000 μm.

[0040] S2. Waveguides are formed using electron beam exposure and etching processes. Photoresist is coated on the aforementioned thin film layer, and after electron beam exposure and development, the waveguide pattern is defined. A metal mask is selected for metal layer deposition, and after photoresist removal and cleaning, argon ion beam etching and chemical etching are performed to obtain the final waveguide.

[0041] Among them, electron beam lithography is used to accelerate the Ar+ ions generated by the ionization of Ar atoms under an electric field to impact the sample surface, causing the atoms of the substrate to detach from the substrate surface, thereby reducing the thickness of the sample.

[0042] Furthermore, in this waveguide fabrication process, the thickness of the metal mask, the energy of Ar ions and the incident angle are optimized, and argon ion beam etching and chemical etching methods are used to effectively reduce the influence of the deposited metal layer film, reduce the damage of the metal layer to the structure, thereby reducing the risk of increased damping of the magnetic material caused by it, which is not conducive to signal transmission, and finally obtaining a high-quality, low-damping waveguide.

[0043] The S3 lift-off process is used to fabricate electrodes and an adhesive layer to obtain the desired random number generator structure. Photoresist is coated onto the sample, and then the electrodes are patterned using a maskless UV lithography machine. When the photoresist at the electrode locations is exposed to UV light, it changes from being insoluble in the developer to being soluble. The unexposed areas, except for the electrodes, remain insoluble in the developer. The exposed sample is then immersed in the developer, leaving an opaque pattern. An adhesive layer of a predetermined thickness and metal electrode material are then deposited on the developed substrate using thermal evaporation. The sample is then immersed in a resist stripper for peeling. The stripper dissolves and removes the photoresist from the sample, leaving the desired metal electrode pattern. Finally, cleaning and inspection are performed to obtain the desired random number generator structure.

[0044] In a second aspect of the invention, a magnetic probability calculation integrated device is provided, comprising a plurality of sub-magnetic random number generators coupled to a main magnetic random number generator, wherein the output module of the sub-magnetic random number generator serves as the pulse input module of the main magnetic random number generator; the sub-magnetic random number generators and the main magnetic random number generator are magnetic random number generators as described in any of the preceding claims.

[0045] In a third aspect of the invention, a probability generation method is provided, employing the aforementioned magnetic particle random number generator, comprising: applying a high-power (10-20 dBm) and frequency (5-7 GHz) pump microwave pulse current signal to a pulse generation module; since the magnetic particle is in the ground state due to its bistable characteristics with high nonlinear effects; continuing to apply another trigger microwave pulse current signal to trigger random switching of the magnetic particle between the ground state and the excited state; subsequently, the spin wave signal of the excited state is transmitted in a magnetic waveguide or thin film with deep nonlinear effects and low damping in the random number generation module; and the magnetic particle probability is output from the output module and detected.

[0046] Furthermore, since the trigger microwave pulse power and the output magnetron probability conform to the Sigmoid function relationship, the probability of the output random number can be further adjusted.

[0047] In a fourth aspect of the invention, a probability generation method is provided, employing the aforementioned integrated device for calculating magnetic particle probabilities, comprising: simultaneously applying a microwave pulse current signal to the pulse generation module of each sub-magnetic particle random number generator, thereby each sub-magnetic particle random number generator having a different output magnetic particle probability, triggering a sigmoid function relationship between the microwave pulse power and the magnetic particle probability; outputting and detecting the magnetic particle probability from the output module of the integrated device; wherein the output probability detected by the output module of the integrated device is the product of the output magnetic particle probabilities of each sub-magnetic particle random number generator, thereby realizing probability calculation.

[0048] Furthermore, the microwave pulse current signals applied simultaneously to the pulse generation modules of each sub-magnetic random number generator can be the same or different.

[0049] Example 1:

[0050] The following example illustrates a random number generator. It includes a pulse input module, a random number generation module, and an output module. The input of the random number generation module is connected to the pulse input module, and the output of the random number generation module is connected to the output module. The random number generation module is deposited on a 500 μm thick (111) crystal orientation gadolinium gallium garnet (GGG) single-crystal substrate. Liquid phase epitaxy technology is used to precisely control the cooling rate to induce directional crystallization on the substrate surface. By adjusting the immersion time, a 44 nm yttrium iron garnet ferrite (YIG) thin film with ultra-low damping, high crystal quality, and uniform thickness is finally formed on the GGG substrate. Electron beam exposure and etching processes are used to form a 220 nm wide and 100 μm long YIG waveguide. GGG serves as the substrate, providing lattice matching to reduce epitaxial stress. Liquid phase epitaxy technology, through single-crystal epitaxial growth, atomic-level interface control, and sub-nanometer thickness precision, makes YIG a material system that can simultaneously meet the requirements of nanoscale integration and low loss. The electrodes of the pulse input and output modules consist of a 10nm thick Ti adhesive layer and a 100nm thick Au conductive layer, with an antenna width of 2μm. The Ti adhesive layer enhances the adhesion between the gold layer and the YIG waveguide surface, prevents the Au layer from peeling off due to thermal or mechanical stress, and suppresses interface diffusion (preventing Au atoms from penetrating into YIG and affecting magnetic properties). The Au conductive layer provides a low-resistance microwave current path.

[0051] Figure 3 This is a schematic diagram illustrating the working principle of this embodiment. Figure 3As shown, the pulse generation module fixes the microwave frequency and power of input microwave pulse 1, placing it in the ground state within the nonlinear magnetic bistable window. Due to the energy barrier between the ground and excited states, a spin wave signal cannot be directly excited. Then, a trigger microwave pulse 2 is applied, and its power is adjusted to provide energy, causing the magnetic particle to randomly switch between the ground and excited states, thus achieving random excitation of the magnetic pulse. The excited spin wave signal is then transmitted in the YIG waveguide of the random number generation module and detected in real-time by the microwave antenna output module at the other end.

[0052] Specifically, in this example, the pulse generation module receives microwave pulse 1 with a period of 1 μs, a width of 0.8 μs, a frequency of 5.02 GHz, and a power of 15 dBm. At this point, the pulse is in the ground state, at the bottom of the bistable window, and cannot excite a spin wave signal. Then, a trigger microwave pulse 2 with a frequency of 4.8 GHz, a power of 10 dBm, a period of 1 μs, and a width of 20 ns is applied. This input microwave pulse reduces the energy difference between the ground and excited states, achieving random excitation of the spin wave pulse signal. Due to the high quality and low damping of the YIG waveguide obtained from this fabrication process, the random pulse signal can be transmitted well. The output module, passing through the Au electrode on the right, receives the random pulse signal and outputs the probability.

[0053] Furthermore, in this embodiment, the probability of the random pulse signal can be adjusted by the power of the microwave in the pulse generation module. Figure 4 This is a graph showing the relationship between the output magnetron signal probability and the input trigger microwave pulse power. The relationship between the input trigger microwave pulse power and the output magnetron probability is strictly a Sigmoid function. Specifically, with microwave pulse 1 at a frequency of 5.02 GHz and a power of 15 dBm, and trigger microwave pulse 2 at a frequency of 4.8 GHz, adjusting only the microwave power of trigger microwave pulse 2 from 8 dBm to 16 dBm adjusts the output magnetron probability of the output module, achieving a random pulse probability of 0 and 1 from 0% to 100%. It should be noted that the same Sigmoid function relationship between the output magnetron probability and the input microwave pulse power can be achieved with different microwave pulse widths and periods for microwave pulses 1 and 2.

[0054] Example 2:

[0055] Figure 5 This is a schematic diagram of the structure of an integrated device for calculating the probability of magnetic particles, fabricated using the magnetic random number generator of this invention. Figure 5As shown, two magnetic random number generators 51 and 52 based on the present invention are connected by a Y-shaped waveguide. By adjusting the input microwave pulse delay time of the two magnetic random number generators 51 and 52, it is ensured that the magnetic pulses from the two magnetic random number generators 51 and 52 can arrive at the output module simultaneously for AND operation. Then, the random output probability of a single magnetic random number generator 51 and 52 is tested and recorded. When the random output probabilities of the two magnetic random number generators 51 and 52 are measured simultaneously, the random output probability of the multiplication operation can be obtained. Specifically, using the magnetic random number generator and its related parameters from Embodiment 1, based on the relationship between the input trigger microwave pulse power and the output magnetic pulse probability, the input trigger microwave pulse power of the magnetic random number generator 52 is adjusted to 12.6 dBm to achieve a random output magnetic pulse probability of 1 / 4; similarly, the input trigger microwave pulse power of the magnetic random number generator 51 is adjusted to 13.4 dBm to achieve a random output magnetic pulse probability of 3 / 4; then, when the two magnetic random number generators output magnetic pulse signals simultaneously, a random output pulse probability of 3 / 16 can be obtained in the output module, that is, the probability calculation of magnetic particles is realized through the magnetic particle probability calculation integrator of the present invention.

[0056] In this invention, a high-quality, high-speed, low-power magnetic random number generator was fabricated by optimizing the nano-etching process. By testing the output magnetic particle strength and using a sigmoid function for fitting, the relationship between the input microwave pulse power and the output magnetic particle probability was established. Furthermore, by optimizing the microwave pulse width and period, a high-speed, low-power magnetic random number generator was achieved. This magnetic random number generator scheme has a simple structure and is easy to integrate, expanding the functionality of spin wave (magnetic particle) based devices while significantly reducing device size and power consumption. Furthermore, a magnetic probability calculation integrator incorporating multiple magnetic random number generators was designed and fabricated. Since the magnetic particle pulses output by the magnetic random number generators do not need to be converted into electrons to directly interact with other magnetic particle pulses, the number of magnetic particle-electron conversions in the device is reduced, further facilitating the construction of a low-power magnetic probability calculation integrator.

[0057] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations. The above-described embodiments are merely preferred embodiments provided to fully illustrate this invention, and their scope of protection is not limited thereto. Equivalent substitutions or modifications made by those skilled in the art based on this invention are all within the scope of protection of this invention.

Claims

1. An integrated device for calculating the probability of magnetic particles, characterized in that, The system includes multiple sub-magnetic random number generators coupled to a main magnetic random number generator. The output modules of the sub-magnetic random number generators serve as pulse input modules of the main magnetic random number generator. A microwave pulse current signal is simultaneously applied to the pulse input module of each sub-magnetic random number generator, and the output module of the main magnetic random number generator outputs and detects the magnetic probability. Both the sub-magnetic random number generators and the main magnetic random number generator are magnetic random number generators. The magnetic random number generator includes a pulse input module, a random number generation module, and an output module. The random number generation module is connected to both the pulse input module and the output module via an adhesive layer. A microwave pulse current signal is applied to the pulse input module, causing the magnetic particle to switch between its ground state and excited state. The spin wave signal of the excited state then propagates in the waveguide or thin film of the random number generation module and is output from the output module, where the probability of the output magnetic particle is detected. The random number generation module is a magnetic waveguide or thin film with a high nonlinear frequency shift coefficient and low damping, deposited on an insulating substrate. The high nonlinear frequency shift coefficient is 0.5~6 GHz, and the low damping coefficient is 5 × 10⁻⁶ GHz. -4 ~4×10 -3 .

2. The integrated device for calculating the probability of magnetic particles according to claim 1, characterized in that, The materials used for the magnetic waveguides or thin films with high nonlinear frequency shift coefficients and low damping include yttrium iron garnet ferrite materials, cobalt iron boron alloy materials, and iron-cobalt and nickel-iron alloy materials.

3. The integrated device for calculating the probability of magnetic particles according to claim 1, characterized in that, The thickness of the random number generation module is 10nm-10. m, width 100nm-100 m.

4. The integrated device for calculating the probability of magnetic particles according to claim 1, characterized in that, Both the pulse input module and the output module are metal electrode layers, selected from Au, Cu, or Pt electrodes, with a width of 2-5 mm. m, with a thickness of 100-500nm.

5. The integrated device for calculating the probability of magnetic particles according to claim 1, characterized in that, The adhesive layer is made of Ti, Ta, or Cr, and has a width of 2-5 mm. m, with a thickness of 3-10nm.

6. The integrated device for calculating the probability of magnetic particles according to claim 1, characterized in that, The random number generation module is fabricated by forming a magnetic thin film with a high nonlinear frequency shift coefficient and low damping on the insulating substrate using a thin film deposition process, and then forming a magnetic waveguide or thin film with a high nonlinear frequency shift coefficient and low damping using an electron beam exposure and etching process.

7. The integrated device for calculating the probability of magnetic particles according to claim 1, characterized in that, Both the pulse input module and the output module are constructed by depositing an adhesion layer on the random number generation module and then using Lift. The metal electrode layer is deposited using an off process.

8. A probability generation method, characterized in that, The integrated device for calculating the probability of magnetic particles as described in claim 1 includes: simultaneously applying a microwave pulse current signal to the pulse input module of each sub-magnetic particle random number generator, and outputting and detecting the magnetic particle probability from the output module of the main magnetic particle random number generator; wherein the output probability detected by the output module of the integrated device for calculating the probability of magnetic particles is the product of the output magnetic particle probabilities of each sub-magnetic particle random number generator.

Images