Vector vortex light generation method based on polarization coded diffraction neural network

Abstract

The invention belongs to the field of optics and photonics, and discloses a vector vortex light generation method based on a polarization coding diffraction neural network, which changes the characteristic that the traditional diffraction neural network cannot modulate polarization information by adding a linearly uncontrollable polarization array layer in the traditional diffraction neural network. An input Gaussian beam is converted into a vector vortex beam needing to be output through a phase modulation layer of a polarization diffraction neural network, independent modulation of vector beams in the x direction and the y direction is achieved, an output result is optimized through an error back propagation algorithm, the output result is matched with an expected value, and the output result is output. And finally, generating vector vortex beams with different orbital angular momentum (OAM) modes in the x and y polarization directions. The method has the regulation and control flexibility and low power consumption of the diffraction neural network, is expected to be applied to the fields of optical communication, quantum information, optical calculation and the like, and provides a new solution for the generation of a vector light field.

Description

Technical Field

[0001] The invention belongs to the field of optics and photonics, and specifically relates to a method for generating vector vortex light based on a polarization-coded diffraction neural network. Background Art

[0002] Vector vortex beam (VVB) combines the characteristics of vortex beam and vector beam, and has a spiral phase wavefront carrying orbital angular momentum (OAM). -ilθ and polarization state. Its different OAM modes and polarization states can carry information, and different OAM modes are orthogonal to each other, providing a new degree of freedom for realizing high-capacity, high-speed, and large-scale optical communications. In recent years, the generation schemes of vector vortex light are mainly the following: liquid crystal phase plate method (S-LCD), spatial light modulator method (SLM), metasurface generation method, etc. They have been widely used to generate vector vortex light. The first two have high control flexibility but often have large light loss and low efficiency. Although metasurface technology has outstanding advantages in integration and high efficiency, it is difficult to adjust flexibly and has high cost. Therefore, traditional generation methods cannot have both flexibility and efficiency.

[0003] In summary, the existing vector vortex light generation technology has limitations, and the traditional D 2 NNs are insensitive to the polarization state of light and cannot modulate light in its polarization dimension. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a method for generating vector vortex light based on polarization-coded diffractive neural network. On the basis of the traditional diffractive neural network, the present invention adds a layer of linear non-adjustable polarization array, which can help D 2 NNs can achieve independent modulation in the x and y directions, and possess the control flexibility and low power consumption of diffractive neural networks. They are expected to be applied in optical communications, quantum information, optical computing and other fields, providing a new solution for the generation of vector light fields.

[0005] In order to achieve the above object, the present invention is achieved through the following technical solutions:

[0006] The present invention is a method for generating vector vortex light based on a polarization-coded diffractive neural network. 2 NNs) include the implementation of polarization-coded diffraction neural networks (PD 2 NNs) is a polarization array layer that performs independent polarization modulation in the x and y directions. Specifically, the vector vortex light generation method specifically includes the following steps:

[0007] Step 1: A Gaussian beam with a wavelength of λ is incident on a polarization-coded diffractive neural network (PD 2 NNs) on the input plane;

[0008] Step 2: Arrange the incident points on the input plane in accordance with the MPSK constellation diagram, and include n incident points at different positions, representing vector vortex lights with different topological charge values;

[0009] Step 3, pre-setting the polarization array layer;

[0010] Step 4: Place the polarization array layer pre-set in step 3 on the diffraction deep neural network (D 2 NNs) diffraction layers, and the distance between them and the previous diffraction layer is d p ;

[0011] Step 5: Input the Gaussian beam into the polarization coded diffraction neural network (PDNN) at different spatial positions. 2 NNs), each position input corresponds to a polarization coded diffraction neural network (PD 2 The vector vortex light output of orbital angular momentum (OAM) modes with different topological charges on the output plane of the NNs can be obtained by multiplying with the Jones matrix of the polarization array layer to obtain vector light fields with different polarization modulations in the x and y directions on the output plane.

[0012] Step 6: In the polarization coded diffraction neural network (PD 2 When training NNs, the mean square error loss function is used to calculate the difference between the predicted vector light field output and the target beam, i.e., the vector vortex light. The polarization-coded diffraction neural network (PDDN) is evaluated by minimizing the mean square error loss function (MSE). 2 NNs), using the stochastic gradient descent algorithm to train the polarization coded diffraction neural network (PD 2 NNs) for optimization;

[0013] Step 7, generating different vector vortex beams by integrating the vector light fields in the x and y directions;

[0014] Step 8: Perform Stokes operation on the generated vector vortex beam to obtain the light field distribution in each polarization direction and verify the generated vector vortex beam;

[0015] Step 9: Use the normalized overlap integral and the mean structural similarity index (SSIM) to measure the quality of the vector vortex beam generated in step 7.

[0016] A further improvement of the present invention is that step 5 specifically includes the following steps:

[0017] Step 5.1: Use a Gaussian beam as input and place it at the input plane spatial position (x m ,y m ,z m ) is expressed as:

[0018]

[0019] Among them, E x represents the light field in the x direction and E y represents the light field in the y direction, (x m ,y m ,z m ) represents the spatial position, m represents the mth incident Gaussian point, E x and E y In the entire polarization-coded diffractive neural network (PD 2 NNs) in parallel;

[0020] Step 5.2: Generate the same complex value modulation for two orthogonal polarization states. The propagation between the diffraction layers is based on the Rayleigh-Sommerfeld diffraction equation. If the distance between the input to the diffraction layer and the diffraction layer is set to d, then the modulated light field reaching the l+1th layer after passing through the lth diffraction layer is expressed as:

[0021]

[0022] Where p represents the polarization state of the modulated light field E at (x, y, z), S represents all pixels in the lth layer, and t l+1 represents the modulation of the l+1th diffraction layer, the modulation can be phase modulation, amplitude modulation and amplitude-phase modulation, represents the diffraction propagation function between the lth layer and the (l+1)th layer at point m;

[0023] Step 5.3: Use Jones matrix J θ Represents each polarization unit in the polarization array layer, θ represents the angle between the polarization axis and the x-axis, then the modulation of the input vector light field by the polarization unit at (x, y, z) in the polarization array layer is expressed as:

[0024] E out (x,y,z)=J θ (x,y,z)E in (x,y,z)

[0025] Among them, E in and E out They represent the vector light field before and after polarization modulation. For each vector light field, there are two components in the x and y directions. At this time, the Jones matrix J θ, producing different modulations for the x and y direction components.

[0026] A further improvement of the present invention is that in step 8, on the output plane, the vector characteristics of the output vector vortex light field can be effectively displayed by using the Stokes parameter, which is manifested as different intensity distributions of the vector light field in different polarization directions. The Stokes parameter is a four-dimensional vector containing S0, S1, S2, and S3, which is used to describe any vector light field, and its expression is:

[0027]

[0028] Among them, I H ,I V Represents the horizontal and vertical strengths, I 45° ,I 135° Respectively represent the intensity of polarization directions at 45° and 135° to the x-axis, I R ,I L They represent the right-circular and left-circular polarization intensities respectively. For each mode of vector vortex light field, the light field in the corresponding polarization direction is obtained after Stokes operation. When the light field intensity distribution in each polarization direction has different petal distributions, it is verified that the generated light field is a vector vortex light field.

[0029] A further improvement of the present invention is that in step 9, the quality of the vector vortex beam generated in step 7 is tested using the normalized overlap integral and the average structural similarity index (SSIM), specifically, the overlap integral is normalized to the range of [0, 1], 1 represents complete overlap, and 0 represents no overlap, which is expressed as:

[0030]

[0031] Among them, E 1 [n] and E 2 [n] represents the light field value of the nth pixel on the output plane of the target light field and the predicted light field, E 2 * [n] indicates E 2 [n] takes conjugate, represents the square of the overlap sum, is the product of the light field intensity, used for normalization;

[0032] The calculation of the mean structural similarity index (SSIM) is expressed as:

[0033]

[0034] Among them, μ 1 Represents the target light field E 1 [n] Mean value of amplitude, μ 2Represents the predicted light field E 2 [n] Mean value of amplitude, σ 1 2 Indicates E 1 [n] Variance of amplitude, σ 2 2 Indicates E 2 [n] Variance of amplitude, σ 12 Indicates E 1 [n], E 2 [n] Covariance of amplitude, C 1 and C 2 is a constant to prevent division by zero errors.

[0035] A further improvement of the present invention is that the polarization coded diffractive neural network includes an input plane, an output plane, a diffractive layer and a polarization coded diffractive neural network (PDD) 2 NNs) are polarization array layers that perform independent polarization modulation in the x and y directions.

[0036] A further improvement of the present invention is that: the polarization array layer in step 3 includes a plurality of polarization units, and each of the polarization units includes two polarization angles of 0° and 90°.

[0037] The beneficial effect of the present invention is that: on the basis of the traditional diffractive neural network, the present invention adds a layer of linear non-adjustable polarization array layer, which can help D 2 NNs achieve independent modulation in the x and y directions. Specifically, x and y polarized light with the same intensity will produce the same modulation in the polarization-insensitive diffraction layer, but when passing through the polarization array layer, different polarization modulations will be generated in the x and y directions by multiplying with the Jones matrix of the polarization array layer, thus obtaining different predicted vector light fields. By matching the network prediction results with the output target, the generation of vector vortex light is achieved.

[0038] At the same time, the vector light beam generated by the present invention performs light field similarity analysis to obtain a vector vortex light beam with the required index.

[0039] The present invention provides a new idea for efficiently and flexibly realizing the generation of vector vortex light under all-optical computing.

[0040] The polarization diffraction network design proposed in the present invention is not based on birefringence, anisotropy or polarization-sensitive materials, but is based on standard isotropic diffraction materials while adding a layer of linear polarizer, which makes the present invention simpler in terms of material selection, manufacturing and amplification. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a schematic diagram of the model of the polarization-coded diffractive nerve proposed in the present invention.

[0042] Figure 2 Schematic diagram of the structure of the polarization array layer.

[0043] Figure 3 The output vector vortex light intensity diagram in the x and y polarization directions and the light intensity diagram expressed by Stokes parameters are obtained through simulation.

[0044] Figure 4 It is a schematic diagram of input plane division. DETAILED DESCRIPTION

[0045] The following will disclose the embodiments of the present invention with drawings. For the purpose of clear description, many practical details will be described together in the following description. However, it should be understood that these practical details should not be used to limit the present invention. That is to say, in some embodiments of the present invention, these practical details are not necessary.

[0046] like Figure 1 and Figure 4 As shown, the present invention is a method for generating vector vortex light based on a polarization-coded diffractive neural network, wherein the polarization-coded diffractive neural network (PD 2 NNs) include the implementation of polarization-coded diffraction neural networks (PD 2 NNs) is a polarization array layer that performs independent polarization modulation in the x and y directions. Specifically, the vector vortex light generation method specifically includes the following steps:

[0047] Step 1: A Gaussian beam with a wavelength of 633.8 nm, i.e., red light, is incident on a polarization-coded diffraction neural network (PDDN). 2 NNs) on the input plane;

[0048] Step 2: In order to obtain vortex beams of different modes and avoid crosstalk between modes, the incident points on the input plane are arranged according to the MPSK constellation diagram, which contains a total of 8 input Gaussian beams. Each input Gaussian beam at a different spatial position corresponds to an output of a different mode, such as Figure 4 For example, if the center of the image is taken as the origin of the coordinate system, then the first Gaussian point is located directly above the origin, and the corresponding output is a vector vortex beam with l=1. The l values ​​of the corresponding output vector vortex beams increase successively as the Gaussian points rotate counterclockwise.

[0049] The polarization coded diffractive neural network includes an input plane, an output plane, four diffraction layers and a polarization array layer. The polarization array layer is the key to achieve independent modulation in the x and y directions. Specifically, the present invention is based on D 2 NNs incorporate non-trainable, pre-set polarization array layers, such as Figure 2As shown in FIG. 1 , the polarization array layer is composed of multiple polarization units, each of which contains two polarization angles of 0° and 90°. The polarization array layer is placed at D 2 NNs diffraction layers, and the distance between them and the previous diffraction layer is 0.

[0050] Step 3: Input the Gaussian beam into the polarization coded diffraction neural network (PDN) at different spatial positions. 2 NNs), each position input corresponds to a polarization coded diffraction neural network (PD 2 The vector vortex light output of orbital angular momentum (OAM) modes with different topological charges on the output plane of the NNs can be obtained by multiplying with the Jones matrix of the polarization array layer to obtain vector light fields with different polarization modulations in the x and y directions on the output plane.

[0051] Specifically, the present invention uses a Gaussian beam as input, and the spatial position is (x m ,y m ,z m ) is expressed as:

[0052]

[0053] Among them, E x Polarization-coded diffraction neural network (PD 2 NNs) light field in the x direction, E y Polarization-coded diffraction neural network (PD 2 NNs) light field in the y direction, x m ,y m ,z m represents the spatial position, m represents the mth incident Gaussian point, E x and E y In the entire polarization-coded diffractive neural network (PD 2 NNs) in parallel;

[0054] Since the trainable diffraction layer is insensitive to the polarization state, the same complex value modulation is generated for the two orthogonal polarization states. The propagation between the diffraction layers is based on the Rayleigh-Sommerfeld diffraction equation. If the distance between the input to the diffraction layer and the diffraction layer is set to 3 cm, then the modulated light field that reaches the l+1th layer after passing through the lth diffraction layer is expressed as:

[0055]

[0056] Where p represents the polarization state of the modulated light field E at (x, y, z), S represents all pixels in the lth layer, t represents the diffraction layer modulation, and W represents the diffraction propagation function between layers. For the linear polarization element, i.e., the polarization array layer, the Jones matrix J is used. θ Each polarization unit in the polarization array layer is modeled, and θ represents the angle between the polarization axis and the x-axis. Then, the replica modulation of a polarization unit at (x, y, z) in the polarization array layer, i.e., the modulated light field, is expressed as:

[0057] E out (x,y,z)=J θ (x,y,z)E in (x,y,z)

[0058] Among them, E in and E out They represent the vector complex fields before and after polarization modulation. Each vector complex field has two orthogonal components in the x and y directions. At this time, the Jones matrix J θ , producing different modulations for the two orthogonal polarization states.

[0059] The polarization array layer is modeled. In order to solve the defect that the diffraction layer itself is insensitive to polarization, the present invention adds a polarization array layer that can produce different modulations for the x and y polarization components in the middle of the diffraction network, such as Figure 2 For the polarization array layer used, the Jones matrix is ​​used mathematically to model each polarization unit in the polarization array layer:

[0060]

[0061] Among them, the value of θ is 0° and 90°. The polarization array layer is composed of multiple polarization units, each of which contains two polarization angles of 0° and 90°. These two different polarizations are combined into a 2x2 period in space and repeated in 4 periods in each direction, extending into a square area, such as Figure 2 The linear polarizer is placed between the second and third layers of the diffraction layer, and the distance d from the second layer p is 0, that is, attached to the third layer. Physically, this polarization array layer can be modeled using a linear polarizer. At this time, each Jones matrix is ​​multiplied by the light field modulated by the previous layer, so that PD 2 NNs can modulate vector beams in the polarization dimension, solving the problem of traditional D 2 Disadvantages of NNs.

[0062] Step 4: On the output plane, in the polarization coded diffractive neural network (PD 2When training NNs, the mean square error loss function is used to calculate the difference between the predicted vector light field output and the target beam, i.e., the vector vortex light. The polarization-coded diffraction neural network (PDDN) is evaluated by minimizing the mean square error loss function (MSE). 2 NNs), using the stochastic gradient descent algorithm to train the polarization coded diffraction neural network (PD 2 NNs) for optimization;

[0063] The loss function during backpropagation is defined as follows:

[0064]

[0065] Among them, E 1 [n] and E 2 [n] represents PD 2 Complex amplitude distribution of target light field and predicted light field of NNs, E 1x [n], E 2x [n] and E 1y [n], E 2y [n] represents the x and y polarization components of the light field respectively. The total loss function can be decomposed into the sum of the light field losses in the x and y polarization directions. n is the number of pixels in the diffraction layer, and b represents the number of samples in a batch during network training.

[0066] The structure of the polarization-coded diffraction deep neural network is built using 4 diffraction layers, each of which is 128×128 in size, and the distance between the input plane and the diffraction layer and between the diffraction layers is set to 3 cm. The size of the polarization layer is 128×128 and is placed between the second and third layers.

[0067] Preparation of input and output data sets: the input data set consists of 1600 sample data generated by the fundamental mode Gaussian function; the output data set consists of two Laguerre-Gaussian beams with opposite topological charge l and opposite circular polarization states, generating a total of 1600 sample data. The above data is only used in the training phase. For the test phase of the embodiment of the present invention, one sample is generated for testing.

[0068] The light field similarity index uses the normalized overlap integral and the average structural similarity index (SSIM) to test the quality of the generated vector vortex beam. The overlap integral is used to calculate the spatial overlap of two light fields. The overlap integral is usually normalized to the range of [0,1], where 1 represents complete overlap and 0 represents no overlap. It can be expressed as:

[0069]

[0070] Among them, E 1 [n] and E 2[n] represents the light field value of the nth pixel on the output plane of the target light field and the predicted light field, E 2 * [n] indicates E 2 [n] is conjugated, the numerator represents the square of the overlap sum, and the denominator is the product of the light field intensity, which is used for normalization.

[0071] The calculation of the mean structural similarity index (SSIM) is expressed as:

[0072]

[0073] Among them, μ 1 Represents the target light field E 1 [n] Mean value of amplitude, μ 2 Represents the predicted light field E 2 [n] Mean value of amplitude, σ 1 2 Indicates E 1 [n] Variance of amplitude, σ 2 2 Indicates E 2 [n] Variance of amplitude, σ 12 Indicates E 1 [n], E 2 [n] Covariance of amplitude, C 1 and C 2 is a constant to prevent division by zero errors.

[0074] The final simulation results, in order to show PD 2 The internal generalization ability of NNs to generate vector vortex light is tested by inputting Gaussian light at different spatial positions into the network according to the 8PSK constellation distribution. 2 Can NNs generate vector vortex beams of corresponding modes? The normalized intensity map obtained is shown in Figure 3 As shown in the figure, it can be seen that for each OAM mode of vector vortex light, PD 2 In order to test the quality of vector vortex light generation, the normalized overlap integral and structural similarity index between the output light field and the target light field were calculated. The normalized overlap integral between each output vector vortex light and the target value and the average structural similarity index (SSIM) in the test phase both reached 0.99.

[0075] The vector characteristics of the light beam are verified. In order to more intuitively describe the polarization state of the generated vector vortex light, the Stokes parameter is used on the output plane to effectively display the vector characteristics of the output vector vortex light field, which is manifested as different intensity distributions of the vector light field in different polarization directions. The Stokes parameter is a four-dimensional vector containing S0, S1, S2, and S3, which is used to describe any vector light field. Its expression is:

[0076]

[0077] Among them, I H ,I V Represents the horizontal and vertical strengths, I 45° ,I 135° Respectively represent the intensity of polarization directions at 45° and 135° to the x-axis, I R ,I L Represent the right and left circular polarization intensities respectively. The normalized intensity map obtained is shown in Figure 3 For each mode of vector vortex light field, the light field in the corresponding polarization direction is obtained after Stokes operation. When the light field intensity distribution in each polarization direction has different petal distributions, it is verified that the generated light field is a vector vortex light field.

[0078] The above description is only an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A method for generating vector vortex light based on polarization coded diffraction neural network, characterized in that: The polarization coded diffractive neural network (PD 2 NNs) include the implementation of polarization-coded diffraction neural networks (PD 2 NNs) is a polarization array layer that performs independent polarization modulation in the x and y directions. Specifically, the vector vortex light generation method specifically includes the following steps: Step 1: A Gaussian beam with a wavelength of λ is incident on a polarization-coded diffractive neural network (PD 2 NNs) on the input plane; Step 2: Arrange the incident points on the input plane in accordance with the MPSK constellation diagram, and include n incident points at different positions, representing vector vortex lights with different topological charge values; Step 3, pre-setting the polarization array layer; Step 4: Place the polarization array layer pre-set in step 3 on the diffraction deep neural network (D 2 NNs) diffraction layers, and the distance between them and the previous diffraction layer is d p ; Step 5: Input the Gaussian beam into the polarization coded diffraction neural network (PDNN) at different spatial positions. 2 NNs), each position input corresponds to a polarization coded diffraction neural network (PD 2 The vector vortex light output of orbital angular momentum (OAM) modes with different topological charges on the output plane of the NNs is multiplied by the Jones matrix of the polarization array layer to obtain vector light fields with different polarization modulations in the x and y directions on the output plane; Step 6: In the polarization coded diffraction neural network (PD 2 When training NNs, the mean square error loss function is used to calculate the difference between the predicted vector light field output and the target beam, i.e., the vector vortex light. The polarization-coded diffraction neural network (PDDN) is evaluated by minimizing the mean square error loss function (MSE). 2 NNs), using the stochastic gradient descent algorithm to train the polarization coded diffraction neural network (PD 2 NNs) for optimization; Step 7, generating different vector vortex beams by integrating the vector light fields in the x and y directions; Step 8: Perform Stokes operation on the generated vector vortex beam to obtain the light field distribution in each polarization direction and verify the generated vector vortex beam; Step 9: Use the normalized overlap integral and the mean structural similarity index (SSIM) to measure the quality of the vector vortex beam generated in step 7.

2. The method for generating vector vortex light based on polarization coded diffraction neural network according to claim 1, characterized in that: The step 5 specifically includes the following steps: Step 5.1: Use a Gaussian beam as input and place it at the input plane spatial position (x m ,y m ,z m ) is expressed as: Among them, E x represents the light field in the x direction and E y represents the light field in the y direction, (x m ,y m ,z m ) represents the spatial position, m represents the mth incident point, E x and E y In the entire polarization-coded diffractive neural network (PD 2 NNs) in parallel; Step 5.2: Generate the same complex value modulation for two orthogonal polarization states. The propagation between the diffraction layers is based on the Rayleigh-Sommerfeld diffraction equation. If the distance between the input to the diffraction layer and the diffraction layer is set to d, then the modulated light field reaching the l+1th layer after passing through the lth diffraction layer is expressed as: Where p represents the polarization state of the modulated light field E at (x, y, z), S represents all pixels in the lth layer, and t l+1 represents the modulation of the l+1th diffraction layer, represents the diffraction propagation function between the lth layer and the (l+1)th layer at point m; Step 5.3: Use Jones matrix J θ Represents each polarization unit in the polarization array layer, θ represents the angle between the polarization axis and the x-axis, then the modulation of the input vector light field by the polarization unit at (x, y, z) in the polarization array layer is expressed as: E out (x,y,z)=J θ (x,y,z)E in (x,y,z) Among them, E in and E out They represent the vector light field before and after polarization modulation. For each vector light field, there are two components in the x and y directions. At this time, the Jones matrix J θ , producing different modulations for the x and y direction components.

3. The method for generating vector vortex light based on polarization coded diffraction neural network according to claim 1, characterized in that: In step 8, on the output plane, the Stokes parameter is used to effectively display the vector characteristics of the output vector vortex light field, which is manifested as different intensity distributions of the vector light field in different polarization directions. The Stokes parameter is a four-dimensional vector containing S0, S1, S2, and S3, which is used to describe any vector light field, and its expression is: Among them, I H ,I V Represents the horizontal and vertical strengths, I 45° ,I 135° Respectively represent the intensity of polarization directions at 45° and 135° to the x-axis, I R ,I L They represent the right-circular and left-circular polarization intensities respectively. For each mode of vector vortex light field, the light field in the corresponding polarization direction is obtained after Stokes operation. When the light field intensity distribution in each polarization direction has different petal distributions, it is verified that the generated light field is a vector vortex light field.

4. The method for generating vector vortex light based on polarization coded diffraction neural network according to claim 1, characterized in that: In step 9, the normalized overlap integral and the average structural similarity index (SSIM) are used to check the quality of the vector vortex beam generated in step 7. Specifically, the overlap integral is normalized to the range of [0, 1], where 1 represents complete overlap and 0 represents no overlap, which is expressed as: Among them, E1[n] and E2[n] represent the light field value of the target light field and the expected output plane at the nth pixel, and E2 * [n] represents the conjugate of E2[n], represents the square of the overlap sum, is the product of the light field intensity, used for normalization; The calculation of the mean structural similarity index (SSIM) is expressed as: Among them, μ1 represents the mean value of the amplitude of the target light field E1[n], μ2 represents the mean value of the amplitude of the predicted light field E2[n], and σ1 2 represents the variance of the amplitude of E1[n], σ2 2 represents the variance of the amplitude of E2[n], σ 12 represents the covariance of the amplitudes of E1[n] and E2[n], where C1 and C2 are constants.

5. A method for generating vector vortex light based on polarization coded diffraction neural network according to any one of claims 1 to 4, characterized in that: The polarization coded diffractive neural network comprises an input plane, an output plane, a diffractive layer and a polarization coded diffractive neural network (PDD) 2 NNs) are polarization array layers that perform independent polarization modulation in the x and y directions.

6. The method for generating vector vortex light based on polarization coded diffraction neural network according to claim 5, characterized in that: The polarization array layer in step 3 includes a plurality of polarization units, and each of the polarization units includes two polarization angles of 0° and 90°.

Images