Real-time manipulation device for microscopic particles

Abstract

Provided in the present disclosure is a real-time manipulation device for microscopic particles, comprising: a first laser, configured to generate a first laser beam; an optical tweezer array generation assembly, comprising: a spatial light modulator configured to modulate a wavefront of the first laser beam on the basis of received intermediate holograms to obtain a first modulated laser beam and modulate the wavefront of the first laser beam on the basis of a received target hologram to obtain a second modulated laser beam, and a focusing assembly configured to focus the first modulated laser beam to produce an intermediate optical tweezer array and configured to focus the second modulated laser beam to produce a target optical tweezer array; and a real-time processing assembly, configured to simultaneously determine a plurality of intermediate holograms on the basis of respective initial positions and respective final positions of a plurality of target microscopic particles, the real-time processing assembly is further configured to transmit the plurality of intermediate holograms to the optical tweezer array generation assembly in a time sequence, and the real-time processing assembly is further configured to transmit the target hologram to the optical tweezer array generation assembly.

Description

A real-time manipulation device for microscopic particles Technical Field

[0001] This disclosure relates to the fields of optics and quantum computing, and in particular to a device for real-time manipulation of microscopic particles. Background Technology

[0002] Neutral atom quantum computing systems based on optical tweezers arrays are strong contenders for large-scale universal quantum computers. A key technology in this field is the use of optoelectronic devices to form rapidly movable, large-scale optical tweezers arrays. Moving the optical tweezers array allows for the manipulation of atoms trapped within them. A significant application of this technology in neutral atom quantum computing experiments is atomic rearrangement. Atomic rearrangement is a process of manipulating atomic positions using optical tweezers. A camera identifies the positional distribution of atoms trapped by the optical tweezers, the tweezers that fail to trap atoms are closed, and then the tweezers that trap atoms are rapidly rearranged into the shape required for quantum algorithms, thus achieving atomic rearrangement.

[0003] The existing steps for realizing atomic rearrangement are as follows: First, a camera is used to collect the fluorescence of the trapped atoms to obtain information about the atomic array; then, a field-programmable gate array is used to identify the positional distribution of the trapped atoms, deduce the optical tweezers array movement method required for the rearrangement of the trapped atoms, and calculate the waveform; finally, a digital-to-analog converter is used to convert the digital waveform into an analog signal, which is then amplified by an analog front-end to drive an acousto-optic deflector to move the optical tweezers, thus completing the optical tweezers manipulation and realizing the atomic rearrangement.

[0004] However, acousto-optic deflectors have only one-dimensional parallelism. This is insufficient for the two-dimensional and three-dimensional atomic arrays commonly used in quantum computing and quantum simulation. The deflectors cannot move all atoms simultaneously, resulting in slow movement speeds for large-scale atomic arrays. Furthermore, the deflection angle range of acousto-optic deflectors is limited by device bandwidth, restricting the size of the optical tweezers arrays they can generate. Therefore, finding optoelectronic devices capable of manipulating larger-scale atomic arrays with higher parallelism has become an urgent need in this field. Summary of the Invention

[0005] In view of the above problems, this disclosure provides a real-time manipulation device for a microscopic particle array, the manipulation device comprising:

[0006] A first laser is configured to generate a first laser.

[0007] An optical tweezers array generating component is configured to receive multiple intermediate holograms and a target hologram, and to generate an intermediate optical tweezers array based on each intermediate hologram and a target optical tweezers array based on the target hologram. The optical tweezers array generating component includes:

[0008] A spatial light modulator is configured to modulate the wavefront of a first laser according to a received intermediate hologram to obtain a first modulated laser, and to modulate the wavefront of the first laser according to a received target hologram to obtain a second modulated laser; and

[0009] A focusing assembly is configured to focus a first modulated laser to generate an intermediate optical tweezers array, and to focus a second modulated laser to generate a target optical tweezers array; and

[0010] The real-time processing component is configured to simultaneously determine multiple intermediate holograms based on the initial and final positions of multiple target microparticles. The real-time processing component is also configured to transmit the multiple intermediate holograms to the optical tweezers array generation component in a time sequence. After the last intermediate optical tweezers array transmission is completed, the real-time processing component is also configured to transmit the target hologram to the optical tweezers array generation component.

[0011] Each intermediate optical tweezers array is configured to trap multiple target microparticles. Under the action of multiple successive intermediate optical tweezers, the multiple target microparticles gradually move and approach their respective final positions. After the last intermediate optical tweezers array traps multiple target microparticles, the target optical tweezers array is configured to trap multiple target microparticles to move them to their respective final positions.

[0012] According to embodiments of this disclosure, the real-time processing component includes:

[0013] The planning unit is configured to determine the parameter information of each intermediate optical tweezer array based on the initial and final positions of multiple target microparticles. The parameter information of the intermediate optical tweezers includes the positions of all intermediate optical tweezers in the intermediate optical tweezer array and the phase at the positions of all intermediate optical tweezers. The parameter information of different intermediate optical tweezer arrays is determined simultaneously.

[0014] The generation unit is configured to generate an intermediate hologram corresponding to the intermediate optical tweezers array based on the positions of all intermediate optical tweezers in each intermediate optical tweezers array and the phase at the positions of all intermediate optical tweezers; and

[0015] The transmission unit is configured to transmit multiple intermediate holograms and a target hologram sequentially to the optical tweezers array generating component;

[0016] In this configuration, each intermediate optical tweezer in the intermediate optical tweezers array is configured to move a target microparticle in a one-to-one correspondence, and each target optical tweezer in the target optical tweezers array is configured to move a target microparticle in a one-to-one correspondence, so that the target microparticle moves to its final position.

[0017] According to embodiments of this disclosure, the real-time processing component is further configured to transmit the initial hologram to the optical tweezers array generation component;

[0018] The optical tweezers array generating component is also configured to obtain an initial optical tweezers array based on the initial hologram; the initial optical tweezers array includes a plurality of initial optical tweezers, at least a portion of the initial optical tweezers in the initial optical tweezers array is used to trap microparticles, each initial optical tweezer is configured to trap one microparticle; the target microparticle is selected from the microparticles trapped by the initial optical tweezers array.

[0019] The real-time control device also includes:

[0020] The containment component is configured to provide microscopic particles and allows the initial optical tweezers array, intermediate optical tweezers array, and target optical tweezers array to enter.

[0021] According to embodiments of this disclosure, the real-time manipulation device further includes an imaging component configured to image target microscopic particles.

[0022] According to embodiments of this disclosure, the real-time processing component further includes:

[0023] The determining unit is configured to determine the initial position of the target microparticles based on the image formed by the target microparticles.

[0024] According to embodiments of this disclosure, the generation unit includes:

[0025] The inference module is configured to perform bilinear interpolation on the positions of all intermediate optical tweezers in each intermediate optical tweezers array and nearest-neighbor interpolation on the phases at the positions of all intermediate optical tweezers in each intermediate optical tweezers array, and input the processing results into a trained convolutional neural network to obtain the output result; and

[0026] The transformation module is configured to perform a Fourier transform on the output to obtain the hologram corresponding to the intermediate optical tweezers array.

[0027] According to embodiments of this disclosure, the real-time processing component further includes:

[0028] The hologram determination unit is configured to generate a transition hologram based on optical tweezers position samples using the Geisberg-Saxon algorithm.

[0029] The tag acquisition unit is configured to perform a Fourier transform on the target region of the transition hologram to obtain an intensity matrix loaded with the interference information of the target region and a phase matrix loaded with the interference information of the target region. The intensity matrix loaded with the interference information of the target region is used as the intensity matrix tag, and the phase matrix loaded with the interference information of the target region is used as the phase matrix tag.

[0030] The phase acquisition unit is configured to perform a Fourier transform on the transition hologram and extract the phase at the location of each optical tweezers position sample from the Fourier transform result to obtain an optical tweezers phase sample.

[0031] An interpolation unit is configured to perform bilinear interpolation on the optical tweezers position sample to obtain an intensity matrix sample; and is configured to perform nearest-neighbor interpolation on the optical tweezers phase sample to obtain a phase matrix sample.

[0032] The input / output unit is configured to input intensity matrix samples and phase matrix samples into a convolutional neural network model, and output intensity matrix prediction results and phase matrix prediction results; and

[0033] The adjustment unit uses the intensity matrix label, phase matrix label, intensity matrix prediction result, and phase matrix prediction result to adjust the parameters of the convolutional neural network until the loss function of the convolutional neural network meets the preset conditions, thus obtaining a trained convolutional neural network.

[0034] According to embodiments of this disclosure, the positions of all intermediate optical tweezers in each intermediate optical tweezers array and the phase at the positions of all intermediate optical tweezers are configured to satisfy the following conditions:

[0035] When the optical tweezers array trapping multiple target microparticles is switched, the target microparticles can move from the first optical tweezers to the second optical tweezers. The first optical tweezers and the second optical tweezers are the optical tweezers corresponding to the same target microparticle in the two optical tweezers arrays before and after the switch.

[0036] When the optical tweezers array trapping multiple target microscopic particles is switched, any two target microscopic particles and the optical tweezers they contain will not collide with each other during the movement.

[0037] According to embodiments of this disclosure, the target hologram is either a hologram generated by a real-time processing component based on the final positions of the various target microparticles, or a preset hologram.

[0038] According to embodiments of this disclosure, the planning unit is further configured to match multiple target microparticles with multiple target positions, wherein each target microparticle corresponds to a target position, and the target position matched with the target microparticle is used as the final position of the multiple microparticles.

[0039] According to embodiments of this disclosure, the real-time processing component can perform real-time processing based on the initial and final positions of multiple target microparticles, and can simultaneously determine multiple intermediate holograms. "Real-time" means that after obtaining the initial positions of the multiple target microparticles, multiple intermediate holograms that meet the requirements can be simultaneously determined based on their respective initial and final positions. Furthermore, the resulting multiple intermediate holograms are independent of each other; that is, the determination of a later intermediate hologram does not depend on a earlier intermediate hologram.

[0040] By using a spatial light modulator and focusing component to generate a series of optical tweezers arrays to sequentially trap target microscopic particles, the movement of these particles can be achieved. Since the optical tweezers arrays generated by the spatial light modulator and focusing component can be one-dimensional, two-dimensional, or three-dimensional, each optical tweezers array generated by the spatial light modulator and focusing component can simultaneously move multiple target microscopic particles, regardless of whether the microscopic particle array formed by multiple target microscopic particles is one-dimensional, two-dimensional, or three-dimensional. Therefore, the embodiments of this disclosure can achieve simultaneous movement of one-dimensional, two-dimensional, and three-dimensional microscopic particle arrays. Attached Figure Description

[0041] Figure 1 shows a schematic diagram of a real-time manipulation device for microscopic particles provided according to an embodiment of the present disclosure;

[0042] Figure 2 shows a schematic diagram of a real-time manipulation device for microscopic particles provided according to another embodiment of the present disclosure;

[0043] Figure 3 illustrates a schematic diagram of the microscopic particle movement provided according to an embodiment of the present disclosure;

[0044] Figure 4 illustrates a flowchart of training a convolutional neural network according to a specific embodiment of the present disclosure;

[0045] Figure 5 shows a schematic diagram of a real-time manipulation device for microscopic particles provided according to another embodiment of the present disclosure.

[0046] Figure Reference Numerals: 1 First Laser; 2 Optical Tweezers Array Generation Component; 21 Spatial Light Modulator; 22 Focusing Component; 3 Real-Time Processing Component; 31 Transmission Unit; 32 Planning Unit; 33 Generation Unit; 34 Determination Unit; 341 Acquisition Card; 342 Analysis Module; 4 Reception Component; 5 Imaging Component; 6 First Lens Component; 7 Dichroic Mirror; 8 Second Lens Component; 9 Imaging and Focusing Lens; 10 First Adjustable Mirror; 11 Second Adjustable Mirror; 12 Third Adjustable Mirror; 13 Fourth Adjustable Mirror; 14 Fifth Adjustable Mirror. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of this disclosure clearer, the following detailed description is provided in conjunction with specific embodiments and the accompanying drawings.

[0048] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0049] All terms used herein, including technical and scientific terms, have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0050] When using expressions such as "at least one of A, B, and C," the meaning should generally be interpreted in accordance with the understanding of a person skilled in the art. For example, "a system having at least one of A, B, and C" should include, but is not limited to, systems having A alone, having B alone, having C alone, having A and B, having A and C, having B and C, and / or having A, B, and C. Similarly, when using expressions such as "at least one of A, B, or C," the meaning should generally be interpreted in accordance with the understanding of a person skilled in the art. For example, "a system having at least one of A, B, or C" should include, but is not limited to, systems having A alone, having B alone, having C alone, having A and B, having A and C, having B and C, and / or having A, B, and C.

[0051] It should also be noted that the directional terms mentioned in the embodiments, such as "up," "down," "front," "back," "left," and "right," are only for reference to the directions in the accompanying drawings and are not intended to limit the scope of protection of this disclosure. Throughout the drawings, the same elements are represented by the same or similar reference numerals. Conventional structures or constructions will be omitted where they may cause confusion in understanding this disclosure.

[0052] Figure 1 shows a schematic diagram of a real-time manipulation device for microscopic particles provided according to an embodiment of the present disclosure.

[0053] As shown in Figure 1, the manipulation device includes: a first laser 1, an optical tweezers array generating component 2, and a real-time processing component 3.

[0054] A first laser 1 is configured to generate a first laser beam. An optical tweezers array generating component 2 is configured to receive a plurality of intermediate holograms and a target hologram, and to generate an intermediate optical tweezers array based on each intermediate hologram and a target optical tweezers array based on the target hologram. The optical tweezers array generating component 2 includes a spatial light modulator 211 and a focusing component 22. The spatial light modulator 211 is configured to modulate the wavefront of the first laser beam according to the received intermediate holograms to obtain a first modulated laser beam, and to modulate the wavefront of the first laser beam according to the received target hologram to obtain a second modulated laser beam. The focusing component 22 is configured to focus the first modulated laser beam to generate the intermediate optical tweezers array, and is configured to focus the second modulated laser beam to generate the target optical tweezers array. The real-time processing component 3 is configured to simultaneously determine multiple intermediate holograms based on the initial and final positions of multiple microparticles. The real-time processing component 3 is also configured to transmit the multiple intermediate holograms sequentially to the optical tweezers array generation component. After the transmission of the last intermediate hologram is completed, the real-time processing component 3 is further configured to transmit the target hologram to the optical tweezers array generation component. Each intermediate optical tweezers array is configured to trap multiple target microparticles. Under the action of the multiple intermediate optical tweezers, the multiple target microparticles gradually move and approach their respective final positions. After the last intermediate optical tweezers array traps the multiple target microparticles, the target optical tweezers array is configured to trap the multiple target microparticles to move them to their respective final positions, thus achieving the movement of the multiple target microparticles.

[0055] According to embodiments of this disclosure, the target microscopic particle can be a cell, a particle with a radius within hundreds of micrometers, such as colloidal particles, nanoparticles, biological samples, or a molecule (from large biological macromolecules to small diatomic molecules) or an atom encoding a quantum state in the field of quantum computing.

[0056] The spatial light modulator 21 used in the embodiments of this disclosure can be a high-speed phase-type liquid crystal spatial light modulator from Meadowlark, with a resolution of 1024*1024 and a maximum refresh rate of 2000Hz. This spatial light modulator 21 can be replaced with other types of spatial light modulators, such as electro-optic spatial light modulators, optically addressed spatial light modulators, etc. The refresh rate and resolution of the spatial light modulator can also be selected according to specific requirements.

[0057] According to embodiments of this disclosure, a hologram is an image that records light field information. If the spatial light modulator 21 used has both amplitude and phase modulation capabilities, a hologram that simultaneously records the amplitude and phase of the light wave can be used to modulate the wavefront of the laser based on the hologram. If the spatial light modulator 21 used only has phase modulation capabilities, a hologram that records the phase of the light wave can be used to modulate the wavefront of the laser based on the hologram. When a hologram is loaded onto the spatial light modulator 21, each pixel on the spatial light modulator 21 modulates the amplitude and phase of the incident light accordingly based on the information in the hologram. The spatial light modulator 21 is a device capable of changing the distribution of the light field. When light shines on the spatial light modulator 21, it modulates the wavefront of the incident light according to the information in the hologram, thereby changing the distribution of the light field, such as the phase distribution of the light field. This modulation effect allows the outgoing light field to acquire a one-dimensional, two-dimensional, or three-dimensional structure corresponding to a hologram after focusing, thus generating a one-dimensional, two-dimensional, or three-dimensional optical tweezers array. Depending on the resolution of the high-speed spatial light modulator, the number of optical tweezers generated can range from thousands to hundreds of thousands.

[0058] According to embodiments of this disclosure, optical tweezers are three-dimensional optical potential wells formed by a highly focused laser beam, the intensity of which approximates a Gaussian or Airy distribution. When the laser beam irradiates a target microscopic particle (such as an atom), the potential energy of the ground-state atom decreases due to the AC Stark effect; the stronger the light intensity, the greater the decrease in potential energy, thus creating an optical potential well. Because the intensity distribution of the focused beam approximates a Gaussian or Airy distribution, an approximate harmonic potential is formed near the deepest part of the optical potential well. The atom is bound to the center of the potential well, i.e., the target microscopic particle is trapped by the optical tweezers. For other microscopic particles, the binding force can be the AC Stark effect, or a combination of scattering force and gradient force; the common point is that the target microscopic particle is bound to the region of strongest light intensity. Therefore, when the light intensity decreases, the target microscopic particle may escape.

[0059] The optical tweezers array generated by the spatial light modulator 21 and focusing component 22 is used to trap target microscopic particles. By switching the hologram, the trapping position of the microscopic particles can be changed, thus enabling the movement of the target microscopic particles. Since the optical tweezers array generated by the spatial light modulator 21 and focusing component 22 can be one-dimensional, two-dimensional, or three-dimensional, each optical tweezers array generated by the spatial light modulator and focusing component can simultaneously move multiple target microscopic particles, regardless of whether the microscopic particle array formed by multiple target microscopic particles is one-dimensional, two-dimensional, or three-dimensional. Therefore, the embodiments of this disclosure can realize the simultaneous movement of one-dimensional, two-dimensional, and three-dimensional microscopic particle arrays.

[0060] According to embodiments of this disclosure, the real-time processing component 3 can perform real-time processing based on the initial and final positions of multiple target microparticles, and can simultaneously determine multiple intermediate holograms. "Real-time" means that after obtaining the initial positions of the multiple target microparticles, multiple intermediate holograms that meet the requirements can be simultaneously determined based on their respective initial and final positions. Furthermore, the resulting multiple intermediate holograms are independent of each other; that is, the determination of a later intermediate hologram does not depend on a earlier intermediate hologram.

[0061] Referring again to Figure 1, the real-time processing component 3 includes: a transmission unit 31, a planning unit 32, and a generation unit 33.

[0062] The planning unit is configured to determine the parameter information of each intermediate optical tweezer array based on the initial and final positions of multiple target microparticles. The parameter information of the intermediate optical tweezers includes the positions of all intermediate optical tweezers in the array and the phase at each position. The parameter information of different intermediate optical tweezer arrays is determined simultaneously. The generation unit 33 is configured to generate an intermediate hologram corresponding to each intermediate optical tweezer array based on the positions of all intermediate optical tweezers and the phase at each position. The intermediate holograms corresponding to different intermediate optical tweezer arrays are generated simultaneously. The transmission unit 31 is configured to transmit the initial hologram, multiple intermediate holograms, and the target hologram to the optical tweezer array generation component in a time sequence. Each intermediate optical tweezer in the intermediate optical tweezer array is configured to move one target microparticle in a one-to-one correspondence, and each target optical tweezer in the target optical tweezer array is configured to move one target microparticle in a one-to-one correspondence, so that the target microparticle moves to its final position.

[0063] Multiple target microscopic particles can be determined using an initial hologram, or not (only their positions are determined). When multiple target microscopic particles are determined using an initial hologram, the transmission component in the real-time processing component 3 is also configured to transmit the initial hologram to the optical tweezers array generation component 2. The optical tweezers array generation component 2 is also configured to obtain an initial optical tweezers array based on the initial hologram. Specifically, the spatial light modulator 21 performs a third modulation on the wavefront of the first laser based on the received initial hologram, and the focusing component 22 focuses the third-modulated laser to obtain the initial optical tweezers array. The initial optical tweezers array includes multiple initial optical tweezers, at least a portion of which are configured to trap microscopic particles, with each initial optical tweezer configured to trap one microscopic particle; the target microscopic particles are selected from those trapped by the initial optical tweezers array. The selection method can be specified or selected according to the planning unit.

[0064] According to embodiments of this disclosure, the initial optical tweezers array utilizes the trapping effect of optical tweezers on microscopic particles, enabling at least a portion of the optical tweezers to trap microscopic particles and achieve random loading of microscopic particles. After loading of microscopic particles is completed, the initial optical tweezers array has two states: one is optical tweezers internally trapping microscopic particles, and the other is empty optical tweezers that have failed to trap atoms. Empty optical tweezers can be closed in intermediate optical tweezers as needed, and excess optical tweezers trapping microscopic particles can also be closed in intermediate optical tweezers as needed. The microscopic particles leave the optical tweezers array position under the combined action of their own micro-motion velocity and gravity, thereby achieving the purpose of cleaning. Multiple intermediate optical tweezers and target optical tweezers utilize the trapping effect of microscopic particles to achieve overall movement of the trapped microscopic particle array when switching holograms. The loading of atoms and molecules by the optical tweezers array is random, while the loading of other microscopic particles besides atoms and molecules by the optical tweezers array can be deterministic, fully filled, or random. However, regardless of whether the microscopic particles can be randomly loaded, the movement or position of the microscopic particle array can be rapidly and precisely manipulated according to the method of this embodiment.

[0065] Figure 2 shows a schematic diagram of a real-time manipulation device for a microscopic particle array provided according to another embodiment of the present disclosure.

[0066] As shown in Figures 1 and 2, the real-time control device also includes a housing component 4.

[0067] When the initial optical tweezers array is not present, the containment component 4 is configured to contain multiple target microparticles, and the containment component 4 allows the intermediate optical tweezers array and the target optical tweezers array to enter.

[0068] According to embodiments of this disclosure, the multiple target microparticles are generated in different ways, and the sources of the target holograms are also different. The planning unit 32 uses the initial position and final position of each of the multiple target microparticles and the necessary optical tweezers phase information to determine the position of all intermediate optical tweezers of the multiple intermediate optical tweezers array and the phase at the position of each intermediate optical tweezer in different ways.

[0069] According to embodiments of this disclosure, when a specified initial hologram and / or a specified target hologram exist, determining the position of all intermediate optical tweezers in each intermediate optical tweezers array and the phase at the position of all intermediate optical tweezers requires not only utilizing the initial position and final position of each of the multiple target microparticles, but also utilizing the phase at the position of all initial optical tweezers in the initial optical tweezers array and / or the phase of all target optical tweezers in the target optical tweezers array.

[0070] According to embodiments of this disclosure, if there is no specified initial hologram and / or target hologram, when determining the position of all intermediate optical tweezers in each intermediate optical tweezers array and the phase at the position of all intermediate optical tweezers, in addition to using the initial position and final position of each of the multiple target microparticles, it is also necessary to randomly assign a phase value to the phase at each optical tweezer position of the first intermediate optical tweezers array and / or randomly assign a phase value to the phase at each optical tweezer position of the target optical tweezers array.

[0071] According to embodiments of this disclosure, and in conjunction with Figures 1 and 2, the aforementioned operating device further includes...

[0072] Imaging component 5 is configured to image target microscopic particles.

[0073] According to embodiments of this disclosure, the real-time processing component 3 further includes a determination unit 34, configured to determine the initial position of the target microparticles based on an image formed by the target microparticles. The determination unit 34 includes an acquisition card 341 and an analysis module 342. The acquisition card can be, for example, a Dalsa acquisition card, and the analysis module can be implemented, for example, through a GPU server. The GPU model of the server can be, for example, an NVIDIA GeForce RTX4090, and the CPU model can be, for example, an AMD EPYC 7763. The acquisition card is configured to acquire images of the target microparticles, and the analysis module is configured to analyze the acquired images to determine the initial position of the target microparticles. In Figure 2, multiple target microparticles are arranged in two layers, namely a first layer and a second layer. After the images of the two layers of target microparticles are acquired by the acquisition card 341, they are transmitted to the analysis module 342. The analysis module 342 analyzes the resulting images to obtain the initial positions of the multiple target microparticles. When multiple target microparticles are determined by an initial optical tweezers array, the analysis module distinguishes the initial optical tweezers by marking the array, for example, marking 1 at the initial optical tweezers that trap microparticles and marking 0 at the initial optical tweezers that do not trap microparticles.

[0074] In Figure 2, the position of each intermediate optical tweezer in the first intermediate optical tweezer array output from the planning unit 32 and the phase at that position are represented as follows: …, the position of each intermediate optical tweezer of the second intermediate optical tweezer and the phase at that position are represented as… …The position of each intermediate optical tweezer of the Nth intermediate optical tweezer and the phase at that position are represented as follows: ….in … Corresponding to the first target microscopic particle, … This corresponds to the second target microscopic particle, and so on.

[0075] The generation unit 33 can precisely control the position of each intermediate optical tweezer in the intermediate optical tweezer array and the phase at each position. At the same time, the calculation of each intermediate hologram is independent, not dependent on any parameters of the previous intermediate hologram, but only on the initial and final positions of all target microscopic particles. Therefore, the generation unit 33 can determine all intermediate holograms of the microscopic particle array movement process in parallel at the same time.

[0076] According to embodiments of this disclosure, the number of intermediate optical tweezers and the phase at each optical tweezer position in each intermediate optical tweezer array are configured to satisfy the following conditions, and the phase and position of all optical tweezers in each intermediate optical tweezer array are configured to satisfy the following two conditions.

[0077] The first condition is that when an optical tweezers array of multiple target microparticles switches, the microparticles in the array (composed of multiple target microparticles) can move from the first optical tweezer to the second. The first and second optical tweezers are the optical tweezers corresponding to the same target microparticle in the two arrays before and after the switch, respectively. In the switched state, the preceding and following optical tweezers are coherently superimposed. To satisfy this first condition, the minimum light intensity caused by interference in the switched state must not be too small. For example, assuming the light intensities of the preceding and following optical tweezers are almost the same, the minimum light intensity of the switched state's optical tweezers must be greater than or equal to 27% of the light intensity before and after the switch. To satisfy this light intensity condition, the position and phase changes of the optical tweezers corresponding to the same target microparticle in any two consecutive optical tweezers arrays during the switch cannot be too large.

[0078] Figure 3 illustrates a schematic diagram of microparticle movement according to an embodiment of the present disclosure.

[0079] As shown in Figure 3, the optical tweezers are assumed to be Gaussian spots. When switching holograms, the intensity of the potential wells of the two optical tweezers corresponding to the same microparticle in two consecutive optical tweezers arrays changes. In Figure 2, the plus (+) shading indicates the potential well of the optical tweezers before (the one preceding) the switching, and the slash ( / ) shading indicates the potential well of the optical tweezers after (the one following) the switching. The solid black line represents the total potential well intensity of the coherent superposition of the potential wells of the two optical tweezers, and the hollow circle represents the microparticle. From the time axis, it can be seen that as time increases, the potential well intensity of the earlier optical tweezers gradually decreases, while the potential well intensity of the later optical tweezers gradually increases. Parts (a) to (c) respectively represent the different effects produced by different optical tweezers switching distances Δr (for the positional changes of the two optical tweezers corresponding to the same microparticle in two consecutive optical tweezers arrays) and different optical tweezers switching phases Δφ (for the phase difference of the two optical tweezers corresponding to the same microparticle in two consecutive optical tweezers arrays). In part (a), Δr = w0 and Δφ = 0.1π. Part (a) shows that the switching distance between the optical tweezers is small, and the phase change is small, allowing the trapped microparticles to successfully transfer between the optical tweezers and potential wells. In part (b), Δr = 2.5w0 and Δφ = 0.1π. Part (b) shows that the distance between the two optical tweezers is large, and the overlap area of ​​the potential wells is small, preventing the trapped microparticles from passing through and causing them to be lost. In part (c), Δr = 2w0 and Δφ = 0π. The phase change between the two optical tweezers is close to π, and the overlapping area forms a potential barrier, making it difficult for the trapped microparticles to cross, resulting in their loss. Here, w0 represents the Gaussian waist radius of the optical tweezers.

[0080] In the embodiments of the invention, it is necessary to ensure that when switching holograms, the position and phase changes of the two optical tweezers corresponding to the same trapped atom in two consecutive optical tweezers arrays cannot be too large, thereby ensuring that the microparticle is not lost during movement. According to embodiments of this disclosure, for any two consecutive optical tweezers arrays of approximately Gaussian shape and a waist radius of about 1 micrometer, the position change of the optical tweezers corresponding to the same trapped microparticle can, for example, be less than or equal to 1 micrometer. The phase difference between the optical tweezers corresponding to the same trapped microparticle in any two consecutive optical tweezers arrays can, for example, be less than or equal to 0.5 rad.

[0081] The second condition is that when the optical tweezers array trapping multiple target microparticles switches, any two target microparticles and their respective optical tweezers in the array will not collide during movement. To satisfy this condition, the distance between any two optical tweezers trapping target microparticles in any optical tweezers array must not be less than the resolution limit of the optical tweezers. Each optical tweezer in the array is a superposition of an Airy disk and a Gaussian disk, the specific superposition method depending on the optical path setting. For each given optical path setting, the criterion for the resolution limit of the optical tweezers can be calculated. Collisions between microparticles usually occur with larger microparticles, and the energy generated by the collision causes particle loss. For small microparticles, due to the small collision cross-section, it is difficult to produce collisions in the usual sense; in this case, collisions between optical tweezers are more destructive. A collision between optical tweezers refers to two optical tweezers being so close that a microparticle can escape or redistribute within the optical tweezers potential well. Redistribution means that at least one microparticle moves from its original location to another optical tweezer.

[0082] According to an embodiment of this disclosure, the planning unit 32 performs multi-agent path finding for each target microparticle based on its initial and final positions. This means that the planning unit performs collision-free (no two particles will occupy the same position at the same time) shortest path finding for each target microparticle. Based on the planning results, the planning unit obtains the parameter information of the intermediate optical tweezers array that simultaneously satisfies the above two conditions.

[0083] According to embodiments of this disclosure, the planning unit 32 is further configured to match a plurality of target microparticles with a plurality of target positions, wherein each target microparticle corresponds to a target position, and the target position matched with the target microparticle is used as the final position of the plurality of microparticles.

[0084] In order to match multiple target micro-particles with multiple target positions, the planning unit 32 adopts the following matching principles for multiple target micro-particles: no collision, shortest total path, and minimum maximum value of a single movement path.

[0085] According to embodiments of this disclosure, in order to match multiple target microparticles with target positions and obtain parameters of multiple intermediate holograms based on the matching results, the planning unit 32 is based on an improved block-based Hungarian matching algorithm. This ensures that during the movement of the microparticle array from its initial position to the target position, the trajectory (path) of each microparticle is a straight line, and the paths between any two microparticles do not intersect, with the total movement distance being minimized. The improved block-based Hungarian matching algorithm specifically includes: First, dividing the microparticle array into multiple blocks, and applying the Hungarian algorithm to each block. During block division, the algorithm ensures that each block has enough microparticles to complete the movement. The improved block-based Hungarian algorithm allows the determination of the paths of microparticles in each region to be executed in parallel, accelerating the computation speed. Second, using linear segmentation, the path is divided into sufficiently small steps, such as 20 steps, so that the movement distance of each step is less than or equal to 1 micrometer. The matching algorithm can be a greedy algorithm or other matching algorithms. The path planning and segmentation algorithms can also be replaced with other collision-free path planning algorithms such as the Probabilistic Path Recognition (PRM) method. These algorithms can work on other computing devices to replace general-purpose computers (servers), such as small embedded devices, etc. Continuing to refer to Figures 1-2, according to embodiments of this disclosure, the generation unit 33 includes: an inference module and a transformation module.

[0086] The inference module is configured to perform bilinear interpolation on the positions of all intermediate optical tweezers in each intermediate optical tweezers array and nearest-neighbor interpolation on the phases at the positions of all intermediate optical tweezers in each intermediate optical tweezers array. The processing results are then input into a trained convolutional neural network to obtain the output. The transformation module is configured to perform a Fourier transform on the output. The result of the transform is a complex matrix. For spatial light modulators with adjustable amplitude and phase, this complex matrix is ​​the hologram; for phase-type spatial light modulators, the argument of this complex matrix is ​​used to obtain the hologram corresponding to the intermediate optical tweezers array.

[0087] According to the embodiments of this disclosure, the following detailed description of bilinear interpolation and nearest neighbor interpolation using formulas (1)-(4) is provided, taking the two-dimensional case as an example.

[0088] The arbitrary optical tweezers position sample and optical tweezers phase sample are expressed as Equation (1):

[0089] Where x and y are the position coordinates of any point in the optical tweezers position sample. Let x be the phase at x, y.

[0090] Intensity matrix samples are generated using bilinear interpolation. Any element A in the intensity matrix sample... uv It is represented by equation (2).

[0091] Phase matrix samples are generated using the nearest neighbor interpolation method. Any element Φ in the phase matrix sample... uv It is represented by equation (3).

[0092] Where [] is the Gaussian floor function, {} is the fractional part function, and [] and {} satisfy equation (4).

[0093] In equation (4), q is either x or y.

[0094] According to embodiments of this disclosure, by employing bilinear interpolation and nearest-neighbor interpolation, the resulting intermediate optical tweezers array can possess super-resolution characteristics.

[0095] According to embodiments of the present disclosure, the transmission unit is configured to transmit an initial hologram, an intermediate hologram, and a target hologram to the spatial light modulator 21 in a time sequence.

[0096] According to embodiments of this disclosure, the real-time processing component 3 further includes a hologram sample determination unit, a tag acquisition unit, a phase acquisition unit, an interpolation unit, an input / output unit, and an adjustment unit.

[0097] The hologram determination unit is configured to generate a transition hologram based on optical tweezers position samples using the Gerchberg-Saxton algorithm (GS algorithm). The label acquisition unit is configured to perform a Fourier transform on the target region of the transition hologram to obtain an intensity matrix and a phase matrix loaded with interference information from the target region. The intensity matrix loaded with interference information is used as the intensity matrix label, and the phase matrix loaded with interference information from the target region is used as the phase matrix label. The phase acquisition unit is configured to perform a Fourier transform on the transition hologram and extract the phase at the location of each optical tweezers position sample from the Fourier transform result, obtaining optical tweezers phase samples. The interpolation unit is configured to perform bilinear interpolation on the optical tweezers position samples to obtain intensity matrix samples; and to perform nearest-neighbor interpolation on the optical tweezers phase samples to obtain phase matrix samples. The input / output unit is configured to input the intensity matrix samples and phase matrix samples into a convolutional neural network model and output the intensity matrix prediction result and the phase matrix prediction result. The adjustment unit uses the intensity matrix label, phase matrix label, intensity matrix prediction result, and phase matrix prediction result to adjust the parameters of the convolutional neural network until the loss function of the convolutional neural network meets the preset conditions, thus obtaining a trained convolutional neural network.

[0098] According to embodiments of this disclosure, the difference between the predicted intensity matrix and predicted phase matrix generated by the convolutional neural network and the label is used as a loss function to optimize the parameters of the convolutional neural network model. The difference in the intensity matrix in the loss function can be calculated using, for example, the L1 norm, and the difference in the phase matrix can be calculated using, for example, the L2 norm.

[0099] The training and inference of the convolutional neural network model (AI model) used in this embodiment are run on a GPU server with two NVIDIA GeForce RTX 4090 graphics cards. Because the AI ​​model is lightweight, inference can also be performed using small embedded hardware such as FPGAs. More AI models can run simultaneously, independently inferring holograms, thus improving computational parallelism. Furthermore, other algorithms can be used to replace the AI ​​model as the core algorithm of the phase-controllable hologram calculation module, such as grating algorithms and conjugate gradient descent algorithms, as long as the algorithm has control over the position and phase of the generated optical tweezers. Users can choose a suitable processing method based on factors such as light utilization efficiency and computational overhead.

[0100] Figure 4 shows a flowchart of training a convolutional neural network according to a specific embodiment of the present disclosure.

[0101] As shown in Figure 4, the training process of the convolutional neural network is completed in several units, including the hologram sample determination unit, label acquisition unit, interpolation unit, phase acquisition unit, input-output unit, and adjustment unit.

[0102] First, a holographic sample determination unit is configured to acquire optical tweezer position samples with a resolution of 8192. The number of optical tweezer position samples is a random value n, denoted as p1, p2, ..., p n p i Let p represent the three-dimensional coordinates of the optical tweezers position sample, where i = 1, 2, 3…n. The hologram sample determination unit uses the Geisenberg-Saxon algorithm (GS algorithm) to obtain a hologram sample from n optical tweezers position samples, where n and p are the coordinates of the optical tweezers position sample. i For example, random sampling can be used, where n can be, for example, 3600 samples. Using the label acquisition unit, a Fourier transform is performed on the target region of the hologram sample, for example, the central (1024×1024) region, to obtain the intensity matrix and phase matrix of the target region. The intensity matrix carries the interference information of the target region, and the phase matrix carries the interference information of the target region. The intensity matrix carrying the interference information of the target region is used as the intensity matrix label, and the phase matrix carrying the interference information of the target region is used as the phase matrix label.

[0103] The interpolation unit performs bilinear interpolation on the optical tweezers position samples to obtain intensity matrix samples of size 1024×1024. The hologram generation unit performs a Fourier transform on the transition hologram (8192×8192) to obtain the phase value at the location of each optical tweezers sample. The nearest neighbor interpolation method is used to generate phase matrix samples of size 1024×1024. The input / output unit inputs the intensity matrix samples and phase matrix samples into the convolutional neural network model and outputs the intensity matrix prediction results and phase matrix prediction results. In the adjustment unit, the convolutional neural network is adjusted using intensity matrix labels and phase matrix labels until the loss function of the convolutional neural network meets preset conditions, resulting in a trained convolutional neural network.

[0104] According to embodiments of this disclosure, the ability to control the phase at the location of each intermediate optical tweezer in the intermediate optical tweezer array is unique to convolutional neural networks. The GS algorithm cannot control the phase at any location of an intermediate optical tweezer; it can only constrain the absolute phase change between two consecutive intermediate optical tweezers to not exceed a certain range. Here, two consecutive intermediate optical tweezers refer to the optical tweezers corresponding to the same microparticle in the two intermediate optical tweezer arrays before and after the switching of the array containing the microparticle. Furthermore, to achieve the above constraint effect, two prerequisites must be met: First, for two consecutive intermediate holograms, the second intermediate hologram must be iterated starting from the first hologram. Second, the number of iteration rounds must be limited to 3-5 rounds to prevent excessive phase change, but this limitation will sacrifice the quality of the hologram. In this embodiment, the movement of microparticles requires a small phase change. Under this premise, compared with the hologram obtained directly using the GS algorithm, using a convolutional neural network has at least four advantages: First, a convolutional neural network can control the phase of each intermediate optical tweezer during its position change, causing the optical tweezer phase to evolve towards the phase of the optical tweezer array corresponding to the target hologram. The phase transformation direction of the GS algorithm is random, and it can only constrain the absolute amount of phase change. Therefore, if a target hologram is specified, by the time the last intermediate hologram is accumulated, the phase will have a significant difference from the phase of the optical tweezers corresponding to the target hologram (almost completely random), leading to the loss of microparticles. Therefore, the GS algorithm cannot specify a target hologram or target array; it can only use the last intermediate optical tweezer for subsequent operations, which reduces the repeatability of subsequent experiments. Second, a convolutional neural network does not rely on iteration, thus it can independently compute all holograms simultaneously. The GS algorithm, because it relies on the previous hologram as the starting point for iteration, must complete the computation of one hologram before starting the next. Third, a convolutional neural network can support super-resolution. The resolution of the GS algorithm depends on the size of the computation matrix. Specifically, for a matrix of size N×N (8192×8192 in this example), the optical tweezers generated by the GS algorithm can only be located at the intersections of the N×N grid. However, for a super-resolution convolutional neural network of size N×N (1024×1024 in this example), the generated optical tweezers positions are continuous and can be selected either within or on the grid. Fourthly, under the same computational scale and hardware conditions, the computation speed of the convolutional neural network is at least one order of magnitude faster than the GS algorithm.

[0105] The computational scale of a hologram must also be modified according to the resolution of the spatial light modulator. The computational scale of the GS algorithm is generally taken as an integer multiple of the resolution, while the computational scale of the neural network is generally taken as the same as the resolution.

[0106] According to embodiments of this disclosure, the containment component 4 is also configured to clean up microparticles that were not trapped by the initial optical tweezers array.

[0107] According to embodiments of this disclosure, at least a portion of the microparticles trapped in the initial optical tweezers array can be selected to form a microparticle array based on the matching method of the planning module, or a portion of the microparticles trapped in the initial optical tweezers array can be actively cleaned up by the real-time processing component 3, so that the remaining microparticles form a microparticle array.

[0108] According to embodiments of this disclosure, when the microparticles are atoms, the containing component 4 can be, for example, a vacuum component. The vacuum component is configured to contain the microparticles. The vacuum component is also configured to receive an initial optical tweezers array to load at least a portion of the initial optical tweezers with microparticles, and to sequentially receive a plurality of intermediate optical tweezers arrays to move the microparticles loaded by the initial optical tweezers, and to receive a target optical tweezers array to move the loaded microparticles to a final position.

[0109] Figure 5 shows a schematic diagram of a real-time manipulation device for a micro-particle array provided according to another embodiment of the present disclosure.

[0110] Referring to Figures 1, 2, and 5, the manipulation device also includes a second laser. This second laser is configured to emit a second laser beam to excite the target microparticles, causing them to fluoresce. An imaging assembly then images the target microparticles based on the fluorescence.

[0111] According to embodiments of this disclosure, the first laser 1 is capable of generating a high-power 813nm laser beam.

[0112] According to embodiments of this disclosure, the focusing component 9 is also configured to collect fluorescence. The focusing component 9 may, for example, be a Special Optics NA=0.55 focusing component.

[0113] According to an embodiment of this disclosure, the manipulation device further includes a first lens assembly 6. The first lens assembly 6 is configured to change the beam size of the laser output from the spatial light modulator 21. The first lens assembly 6 consists of two lenses, which can, for example, be configured as a 3:1 beam expander.

[0114] According to embodiments of this disclosure, the manipulation device further includes: a dichroic mirror 7, configured to transmit the light beam output from the first lens assembly 6 so that the light beam output from the first lens assembly 6 enters the focusing assembly 9, and configured to transmit the fluorescence output from the focusing assembly 9 to the imaging assembly 5. The imaging assembly 5 employs an electron-multiplying charge-coupled device (EMCCD), also known as an electron-multiplying CCD camera, model Andor iXon Ultra 888, or may be replaced with other cameras or detector arrays, such as research-grade complementary metal-oxide-semiconductor (sCMOS, or research-grade CMOS camera), photodiode (APD) arrays, etc. Depending on the type of image acquisition device and receiver, the acquisition card or data readout device may need to be replaced accordingly.

[0115] According to embodiments of this disclosure, at least a portion of the functionality of one or more units in the real-time processing component 3 can be combined with at least a portion of the functionality of other units and implemented in one unit. According to embodiments of this disclosure, at least one unit in the real-time processing component 3 can be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging the circuitry, or implemented in any one of software, hardware, and firmware methods, or in a suitable combination of any of these. Alternatively, at least one ternary unit in the real-time processing component 3 can be at least partially implemented as a computer program unit that, when run, can perform corresponding functions. The real-time processing component 3 can use small embedded hardware (such as an FPGA).

[0116] According to embodiments of this disclosure, the manipulation device further includes a second lens assembly 8 and an imaging focusing lens 9. The second lens assembly 8 is configured to adjust the size of the microparticle array on the imaging plane of the imaging assembly 5, and consists of two lenses, which can, for example, be configured as a 1:1 beam expander. The microparticle fluorescence is collimated by the focusing assembly 22, reflected by the dichroic mirror 7, scaled by the second lens assembly 8, and finally focused onto the imaging assembly 5 by the imaging focusing lens 9 to complete the imaging of the microparticle array.

[0117] According to an embodiment of the present disclosure, the manipulation device further includes a plurality of adjustable mirrors arranged along the optical path. The plurality of adjustable mirrors include a first adjustable mirror 10, a second adjustable mirror 11, a third adjustable mirror 12, a fourth adjustable mirror 13, and a fifth adjustable mirror 14. The plurality of adjustable mirrors are configured to change the propagation direction of light.

[0118] The wavelength and power parameters of each laser, the magnification of the second lens assembly 8, the placement and mechanical structure of each adjustable reflector, and the field of view of the imaging focusing lens 9 in this embodiment are all designed according to the requirements of the actual optical tweezers system.

[0119] This embodiment of the invention can control the position and phase of each optical tweezer in each intermediate optical tweezer array, and calculate the required hologram based on the required optical tweezer position and phase. Furthermore, the calculation of each hologram is independent, requiring no iteration and depending solely on the information from the previous hologram, relying only on the user-inputted optical tweezer position and phase information. Therefore, this module can calculate all holograms of the optical tweezer movement process in parallel at the same time.

[0120] Therefore, the target optical tweezers array does not depend on the last intermediate hologram in the rearrangement process.

[0121] The manipulation device provided in this disclosure can simultaneously move two-dimensional or three-dimensional microscopic particle arrays, and the rearrangement time of the microscopic particles is independent of the size of the microscopic particle array. For an acousto-optic deflector with only one-dimensional parallel movement degree of freedom, the rearrangement time and the number of atoms in the array increase polynomially. The manipulation device based on the spatial light modulator in this application has three-dimensional parallel movement degree of freedom, and for a microscopic particle array with a certain number density, the rearrangement time does not increase with the size. In addition, the movement range of the embodiments of this disclosure is large. The deflection range of the spatial light modulator for laser light is much larger than that of the acousto-optic deflector.

[0122] The embodiments of this disclosure have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of this disclosure. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. The scope of this disclosure is defined by the appended claims and their equivalents. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of this disclosure, and all such substitutions and modifications should fall within the scope of this disclosure.

Claims

1. A real-time manipulation device for microscopic particles, comprising: A first laser is configured to generate a first laser. An optical tweezers array generating component is configured to receive a plurality of intermediate holograms and a target hologram, and to generate an intermediate optical tweezers array based on each intermediate hologram and a target optical tweezers array based on the target hologram. The optical tweezers array generating component includes: A spatial light modulator is configured to modulate the wavefront of the first laser according to the received intermediate hologram to obtain a first modulated laser, and to modulate the wavefront of the first laser according to the received target hologram to obtain a second modulated laser. A focusing component is configured to focus the first modulated laser to generate the intermediate optical tweezers array, and is configured to focus the second modulated laser to generate the target optical tweezers array. A real-time processing component is configured to simultaneously determine multiple intermediate holograms based on the initial and final positions of multiple target microparticles. The real-time processing component is also configured to transmit the multiple intermediate holograms to the optical tweezers array generating component in a time sequence. After the transmission of the last intermediate hologram is completed, the real-time processing component is also configured to transmit the target hologram to the optical tweezers array generating component. Each of the intermediate optical tweezers arrays is configured to trap the plurality of target microparticles. Under the action of the successive intermediate optical tweezers arrays, the plurality of target microparticles gradually move and approach their respective final positions. After the last intermediate optical tweezers array traps the plurality of target microparticles, the target optical tweezers array is configured to trap the plurality of target microparticles to move the plurality of target microparticles to their respective final positions.

2. The real-time control device according to claim 1, wherein, The real-time processing component includes: The planning unit is configured to determine the parameter information of each intermediate optical tweezer array based on the initial and final positions of multiple target microparticles. The parameter information of the intermediate optical tweezers includes the positions of all intermediate optical tweezers in the intermediate optical tweezer array and the phase at the positions of all intermediate optical tweezers. The parameter information of different intermediate optical tweezer arrays is determined simultaneously. The generation unit is configured to generate an intermediate hologram corresponding to the intermediate optical tweezers array based on the positions of all intermediate optical tweezers in each intermediate optical tweezers array and the phase at the positions of all intermediate optical tweezers; The transmission unit is configured to transmit the plurality of intermediate holograms and the target hologram to the optical tweezers array generating component in a time sequence; In this configuration, each intermediate optical tweezer in the intermediate optical tweezers array is configured to move a target microparticle in a one-to-one correspondence, and each target optical tweezer in the target optical tweezers array is configured to move a target microparticle in a one-to-one correspondence, so that the target microparticle moves to its final position.

3. The real-time control device according to claim 1, wherein, The real-time processing component is also configured to transmit the initial hologram to the optical tweezers array generating component; The optical tweezers array generating component is also configured to obtain an initial optical tweezers array based on the initial hologram; The initial optical tweezers array includes a plurality of initial optical tweezers, at least a portion of which are used to trap microscopic particles, and each initial optical tweezer is configured to trap one microscopic particle. The target microparticles are selected from the microparticles trapped by the initial optical tweezers array; The real-time control device also includes: An accommodating component is configured to provide the microparticles, the accommodating component allowing the initial optical tweezers array, the intermediate optical tweezers array, and the target optical tweezers array to enter.

4. The real-time control device according to claim 1, further comprising: An imaging component is configured to image the target microscopic particles.

5. The real-time manipulation device according to claim 4, wherein the real-time processing component further comprises: The determining unit is configured to determine the initial position of the target microparticle based on an image formed by the target microparticle.

6. The real-time control device according to claim 2, wherein, The generation unit includes: The inference module is configured to perform bilinear interpolation on the positions of all intermediate optical tweezers in each intermediate optical tweezers array and nearest neighbor interpolation on the phase at the positions of all intermediate optical tweezers in each intermediate optical tweezers array, and input the processing results into the trained convolutional neural network to obtain the output results. The transformation module is configured to perform a Fourier transform on the output result to obtain a hologram corresponding to the intermediate optical tweezers array.

7. The real-time control device according to claim 6, wherein, The real-time processing component also includes: The hologram determination unit is configured to generate a transition hologram based on optical tweezers position samples using the Geisberg-Saxon algorithm. The tag acquisition unit is configured to perform a Fourier transform on the target region of the transition hologram to obtain an intensity matrix loaded with the interference information of the target region and a phase matrix loaded with the interference information of the target region. The intensity matrix loaded with the interference information of the target region is used as the intensity matrix tag, and the phase matrix loaded with the interference information of the target region is used as the phase matrix tag. The phase acquisition unit is configured to perform a Fourier transform on the transition hologram and extract the phase at the location of each optical tweezers position sample from the Fourier transform result to obtain an optical tweezers phase sample. An interpolation unit is configured to perform bilinear interpolation on the optical tweezers position sample to obtain an intensity matrix sample; and is configured to perform nearest-neighbor interpolation on the optical tweezers phase sample to obtain a phase matrix sample. The input / output unit is configured to input the intensity matrix samples and the phase matrix samples into the convolutional neural network model and output the intensity matrix prediction result and the phase matrix prediction result. The adjustment unit uses the intensity matrix label, the phase matrix label, the intensity matrix prediction result, and the phase matrix prediction result to adjust the parameters of the convolutional neural network until the loss function of the convolutional neural network meets the preset conditions, thereby obtaining the trained convolutional neural network.

8. The real-time control device according to claim 2, wherein, The positions of all intermediate optical tweezers in each intermediate optical tweezers array and the phase at each position of the intermediate optical tweezers are configured to satisfy the following conditions: When the optical tweezers array trapping the multiple target microparticles is switched, the target microparticle can move from the first optical tweezer to the second optical tweezer. The first optical tweezer and the second optical tweezer are the optical tweezers corresponding to the same target microparticle in the two optical tweezers arrays before and after the switch. When the optical tweezers array trapping the multiple target microparticles is switched, any two target microparticles and their respective optical tweezers will not collide with each other during the movement.

9. The real-time control device according to claim 1, wherein, The target hologram is either a hologram generated by the real-time processing component based on the final positions of the multiple target micro-particles, or a preset hologram.

10. The real-time control device according to claim 2, wherein, The planning unit is also configured to match multiple target microparticles with multiple target positions, wherein each target microparticle corresponds to a target position, and the target position matched with the target microparticle is used as the final position of the multiple microparticles.

Images