Method of manufacture and photonic metasurface for performing computationally intensive mathematical calculations

Abstract

According to aspects of the embodiments described herein, an optical computing device (100) includes a plurality of input waveguides (101), a metasurface (103) in contact with the plurality of input waveguides, and a plurality of output waveguides (105) in contact with the transformation metasurface. The optical computing device can be configured to perform a mathematical operation, which can be a matrix multiplication. A computer-implemented method (300) of designing an optical computing device including a plurality of input waveguides, a metasurface, and a plurality of output waveguides includes exciting each input waveguide one at a time (303) and measuring energy of the input region and the output region to determine a contribution of the current input waveguide. The sum of the contributions of all input waveguides (307) is compared to a target transformation (315) to determine a loss value for updating a set of design parameters (317).

Description

Technical Field

[0001] This application relates to optical computing. More specifically, this application relates to the design and use of photonic metasurfaces. Background Technology

[0002] The rapid development of new technologies, such as machine learning, artificial intelligence, and big data analytics, requires more computing power than ever before. However, with Moore's Law slowing dramatically, central processing units (CPUs) and graphics processing units (GPUs) based on electronic computing hardware and architectures are facing performance limits. Therefore, there is a strong expectation for alternative computing platforms to rebuild Moore's Law.

[0003] Compared to their electronic counterparts, optical computing devices offer a promising solution due to their ultrafast speed and ultra-low power consumption. By intelligently engineering the scattering, propagation, and light-matter interactions of light, it is possible to create custom optical computing devices capable of computationally intensive mathematical operations. These concepts have been explored from various angles, ranging from more classical Fourier optics to more recently, calculus metasurfaces designed using Green's functions.

[0004] Linear algebra is fundamental to computation in the digital domain. Therefore, the ability to perform linear algebraic operations using metasurfaces is of great significance when considering optical computation. Based on the implementation described in this paper, a metasurface with significantly reduced engineering footprint is proposed through topology optimization. This metasurface is realized by utilizing light scattering and light-matter interactions to perform arbitrary universal matrix multiplication.

[0005] Currently, there are several popular optical computing platforms: Optalysys: 4F Optical Correlator The so-called "4-f" optical correlator uses lenses to perform Fourier transforms and then convolve the two signals in the spatial frequency domain. It requires a physical device footprint of four focal lengths, making it rather bulky compared to the thin metasurface we've designed here. In terms of accuracy, these devices are similar to conventional electronic computers, only offering 10 to 100 times higher throughput for a few specialized problems.

[0006] LightOn: Random Projection

[0007] By utilizing the scattering phenomenon through complex media, random projection offers a potentially large number of operations per cycle. Therefore, a single cycle can generate 10-1... 12 One MAC, but this throughput requires significant overhead in terms of device footprint (approximately one meter) and frequency (approximately 100 Hz). Similarly, a lengthy calibration procedure is necessary for each sample. Therefore, in its current state, this technology is likely still highly specialized.

[0008] LightMatter: Mach-Zehnder Interferometer (MZI)

[0009] Computation using a Mach-Zehnder interferometer (MZI) relies on a network of phase shifters connected by waveguides, which allows for MAC (Macro-Analog Shifting) operation. Many MZI cells can be compiled together to achieve complex operations similar to those in analog electronics. Each cell is typically micrometers in size, so the rather complex device occupies a footprint similar to that of a computer chip. Power consumption is as low as 3W, significantly better than that of a computer, while providing 10W per cycle. 6 MAC address.

[0010] Neuromorphic computing

[0011] The electronic neuromorphic computer TrueNorth has a MAC rate of 2.5 kHz, and each MAC requires 4.9 μm. 2 It also offers 5-bit precision. The photonic design operates faster, up to 20 GHz, but requires 200 μm per MAC. 2 (Also provides approximately 5 digits of precision).

[0012] In summary, most existing optical computing platforms offer higher throughput compared to their electronic counterparts. However, they remain relatively bulky in size, making them less competitive in terms of device footprint. Improved solutions are needed to deliver the required throughput within a reduced device footprint. Summary of the Invention

[0013] According to various aspects of the embodiments described herein, an optical computing device includes a plurality of input waveguides, a photonic metasurface in contact with the plurality of input waveguides, and a plurality of output waveguides in contact with the transform metasurface. The optical computing device can be configured to perform mathematical operations, including but not limited to matrix multiplication. The plurality of input waveguides can be configured to receive electromagnetic (EM) signals, wherein the power level of the EM signal at each input waveguide represents the numerical value of a vector used as an input to the optical computing device. Furthermore, the phase of the EM signal at a given input waveguide can be considered as representing the sign of the numerical value. According to some embodiments, an optical computing device having eight input waveguides may include a photonic metasurface with a thickness of approximately 3 μm; for 16 input waveguides, the thickness of the photonic metasurface may be approximately 4 μm; and for a device having 32 input waveguides, the photonic metasurface may have a thickness of approximately 12 μm.

[0014] In another embodiment, a computer-implemented method for designing an optical computing device having multiple input waveguides, a photonic metasurface, and multiple output waveguides includes: determining a target transformation of the optical computing device; performing multiple optimization steps for designing the photonic metasurface, each step including sequentially exciting the input waveguides, measuring the energy of the input and output regions to determine the contribution of the current input waveguide, summing the contributions of all input waveguides, comparing the summed contributions with the target transformation to determine a loss function value, and updating a set of design parameters based on the loss function value. The set of design parameters may be updated according to optimization to minimize the loss function value. In some embodiments, optimization may be performed based on a finite-memory BFGS algorithm. In some embodiments, computation may be performed in parallel to determine the contributions of at least two input waveguides. According to embodiments, the target transformation is defined as a mathematical operation. As a non-limiting example, the mathematical operation may be matrix multiplication. According to embodiments, the target transformation is scaled by a normalization factor to conform to physical constraints.

[0015] In some implementations, the loss function includes terms for applying a target electric field to the input region of the optical computing device and terms for applying a target electric field to the output region of the optical computing device. Applying the target electric field to the input region of the optical computing device reduces the backscattering effect of the input electromagnetic (EM) signal in the input region of the optical computing device. To produce optical computing devices with improved smaller footprints, for an optical computing device with 8 input waveguide channels, the photonic metasurface can be designed with a thickness of approximately 3 μm. For an optical computing device with 16 input waveguide channels, the photonic metasurface can be designed with a thickness of approximately 4 μm, and for an optical computing device with 32 input waveguide channels, the photonic metasurface can be designed with a thickness of approximately 12 μm. Attached Figure Description

[0016] The foregoing and other aspects of the invention will be best understood from the following detailed description when read in conjunction with the accompanying drawings. For the purpose of illustrating the invention, currently preferred embodiments are shown in the drawings; however, it should be understood that the invention is not limited to the specific means disclosed. The drawings include the following illustrations: Figure 1 An optical computing device according to an embodiment of the present disclosure is shown.

[0017] Figure 2 An illustration shows a reverse design optimization for determining the topology of an optical metasurface for an optical computing device according to an embodiment of the present disclosure.

[0018] Figure 3 A process flowchart for optimizing the topology of an optical computational metasurface according to various aspects of embodiments of the present disclosure is shown.

[0019] Figure 4 The diagram illustrates the topology of an optimized metasurface with two input waveguides and two output waveguides, the comparison of the simulation matrix with the target transformation matrix, and the procedure for determining the actual transformation matrix based on EM simulation.

[0020] Figure 5 The final metasurface structure obtained from reverse design is shown.

[0021] Figure 6 An exemplary computing environment is shown that can be used to implement embodiments of the present invention. Detailed Implementation

[0022] According to embodiments of this disclosure, the following are provided: Figure 1 The optical computing device 100 shown is an optical computing device 100. The optical computing device 100 includes multiple (n) input waveguides a1-a1. n 101. Input waveguide 101 is configured to receive the optical signal from each input waveguide. Additionally, there are multiple (m) output waveguides b1-b m 105. A transform core 103 is connected between the input waveguide 101 and the output waveguide 105. This transform core manipulates the optical signal received in the input waveguide 101 and creates a set of optical signals in each output waveguide 105. The optical computing device 100 can be programmed to perform operations by a universal matrix T∈R. n×m The transformation operation represents the transformation operation such that b = Ta, where a = [ a 1, ..., a n ] T and b=[ b 1, ..., b m ] T These are associated with the EM waves in input waveguide 101 and output waveguide 105, respectively. The magnitudes of the components in vectors a and b are reflected in the power of the EM signal transmitted to each waveguide. Furthermore, the signs of the vector components are indicated by the relative phase between waveguides 101 and 105. That is, if the first EM signal in a given waveguide is considered "positive," then any waveguide containing an EM signal in phase with the first signal is considered to represent a positive value, while those waveguides containing an EM signal out of phase with the first signal are considered to represent a negative value.

[0023] Transformer 103 includes an optical metasurface that controls the propagation of an optical signal from input waveguide 101 to output waveguide 105. The optical metasurface of transformer 103 can be selected to produce a set of values ​​representing the components of the forming vector. Thus, for a given operation or transform, a set of values ​​can be encoded as an EM signal into input waveguide 101, and transformer 103 can be selected such that a predetermined set of output values ​​is produced at output waveguide 105.

[0024] According to a specific embodiment of this disclosure, the transformation core 103 can be designed using reverse engineering concepts. The following will now be described... Figure 1 Details of the reverse engineering of the optical computing device shown.

[0025] As discussed above, the transform core 103 includes a photonic metasurface that guides and scatters light provided as an optical input signal. The topology of the optical metasurface determines the light transmission characteristics of the photonic metasurface. By controlling this topology, the characteristics of the transform core 103 can be designed to perform the desired function. According to one embodiment of this disclosure, a proposed topology optimization for designing a new computing device can be expressed as: Equation (1) Where ρ is the design variable vector, ε is the dielectric constant distribution interpolated from the design variable ρ, and the loss function... It has the following specific forms: , Equation (2) in, and The electric field E is extracted at the input and output regions, respectively. z Extraction operator, and These are the target EM fields that need to be applied in the input and output regions when the i-th input waveguide is excited by element excitation. n is the number of input waveguides, and w is a weighting factor that can be selected by the user to favor the physical phenomenon of backscattering, where light enters the input waveguides and is reflected back through the input region, and the output is the energy used to generate the desired transformed output.

[0026] Apply target field to input region To minimize the backscattering of the input EM wave, and to apply the target field Ensure the target transformation is achieved. Before performing topology optimization, the target transformation matrix T needs to be calculated based on the information encoded in the i-th column of matrix T. .

[0027] Figure 2 A diagram illustrating the reverse design optimization for determining the topology of an optical metasurface for an optical computing device according to embodiments of this disclosure is provided. For each input waveguide in input region 201, one of the plurality of input waveguides is excited, while all other input waveguides remain unexcited. In each step of the optimization, each input waveguide is excited one by one until all input waveguides are excited. At each excitation, the associated EM field is calculated based on the target matrix T. and The fitting loss attributable to each input waveguide is summed according to the loss function in equation (2) above. (See reference...) Figure 2 Scenario A illustrates the excitation of the first input waveguide a1207. The excitation is directed to waveguide a... l An EM signal is applied so that light from the EM signal is scattered across the entire optical metasurface of design scenario 205, generating an output signal at each output waveguide in output region 203. This optimization is performed on a computer that performs simulations of the optical computing device. In the second simulation step shown in scenario B, the fourth input waveguide 209 is excited while all other input waveguides remain unexcited. An EM signal is applied to the fourth waveguide 209 so that light from the EM signal is scattered across the entire optical metasurface simulated as design scenario 205. This process applies the EM signal to one input waveguide one by one until the last input waveguide 211 is excited, while all other input waveguides remain unexcited. This is based on the last waveguide 211 providing the output value at output region 203 and the output waveguide. Each individual waveguide channel on the input side is excited. In particular, in each topology optimization step, each individual waveguide channel is excited by unit excitation (while keeping other waveguide channels unexcited). The associated EM field is then calculated based on the target matrix T. and The fitting loss contributions from each input waveguide are summed to obtain the loss function defined in equation (2) above. Automatic differentiation is used to calculate the exact gradient of the loss function, and the design variable ρ is updated through optimization. In one implementation, the finite-memory BFGS algorithm can be used to update the design variable. This procedure is then repeated until convergence of the design variable is achieved. The design variable ρ is referenced to a design point in the optical metasurface. The volume of the optical metasurface can be considered as a voxel within the volume. As an example, each voxel can be assigned a binary value. The first binary value indicates the presence of material in the voxel, while the second binary value indicates the absence of material. Once the optimized design is obtained, the target transformation matrix T can be achieved for any universal input vector according to the superposition principle.

[0028] Figure 3This is a flowchart of a process for optimizing the topology of an optical computational metasurface according to various aspects of embodiments of the present disclosure. Method 300 is used to optimize the optical metasurface. The optimization step begins 301 by applying a unit excitation to one of a plurality of input waveguides while keeping all other input waveguides unexcited 303. With the selected input waveguide excited, an input EM signal is transmitted through the simulated optical metasurface to generate simulated outputs at a plurality of output waveguides. The simulated outputs represent the contribution of the selected (i-th) input waveguide and are read to capture the contribution of the selected input waveguide to the fitting loss used to evaluate and optimize the current design of the optical computing device 305. The plurality of input waveguides are evaluated one by one to identify the fitting loss contribution of each input waveguide. As each input waveguide is evaluated, its contribution is added to the current sum of the overall design 307. Next, it is determined whether all input waveguides have been evaluated in the current optimization step 309. If not 311, the next input waveguide is excited while keeping all other input waveguides unexcited 303.

[0029] If all input waveguides 313 have been excited in the current optimization step, the loss function of equation (2) is calculated to determine how close the current design alternative is to producing an output consistent with the target transformation. To minimize the loss function, optimization steps are performed to generate a set of updated design parameters. The design parameters represent the physical aspects of the optical metasurface and control the propagation of light through the optical metasurface and thus the output level at the output waveguide. Based on the calculated loss function, the design parameters are updated by optimization to minimize the loss function 317. The analysis results regarding the updated design parameters are monitored after each optimization step. The convergence of the updated design parameters is then checked 319. If the design parameters have converged 323, the topology optimization ends 325. If the design parameters have not yet converged 321, the next topology optimization step 301 is performed.

[0030] Regarding the aforementioned reverse design workflow, the associated computational cost varies linearly with the total number of input waveguides n, which is the number of columns in the target transformation matrix T. However, the implementation can be easily parallelized, allowing computations for each individual waveguide to be performed in parallel, thereby significantly improving the efficiency of the reverse design process.

[0031] Here, we provide a illustrative example to further demonstrate how reverse engineering can be implemented. In this illustrative example, we will use 2... 2. Matrix T is considered as: Equation (3) Figure 4This diagram illustrates the topology of an optimized metasurface 401 with two input waveguides 403 and two output waveguides 405, compares the simulation matrix 407 with the target transformation matrix 409, and illustrates the procedure for determining the actual transformation matrix based on EM simulation. When waveguide 1 is excited 411, the energy field in the output region 413 is captured 415. Next, waveguide 2 is excited 417, and the output field at the output region 413 is captured 419. The captured output fields of each input waveguide are summed to produce the simulation result 407. The simulation result 407 is compared with the target 409. The design parameters of the photonic metasurface 401 are updated, and another optimization step is performed to attempt to minimize the difference between the simulation result 407 and the target result 409.

[0032] Matrix normalization

[0033] A universal transformation matrix can take any value in its components. However, a physical metasurface must satisfy constraints imposed by the laws of physics. In practice, a metasurface cannot physically represent a matrix of any size because a metasurface cannot generate energy out of thin air. Therefore, the target matrix 309 must be scaled by a normalization factor before it can be encoded into a physical device using a reverse engineering procedure.

[0034] We can assume that the reverse engineering process generates a metasurface with zero fitting error. If this set of input waveguides is excited with an arbitrary input vector 'a', the energy level of the output waveguide will be the vector 'Ta'. The total input energy (summing the energy associated with each input waveguide) is equal to: Equation (4) The total output energy (summing the energy associated with each output waveguide) is equal to: Equation (5) Where s is the normalization factor derived from this analysis.

[0035] Energy cannot be generated spontaneously. Therefore, for any choice of a, En in ≥ En out The following conditions are met: Equation (6) For any real-valued input vector a.

[0036] consider The singular value decomposition (SVD) of the matrix, where U and V are unitary (rotation) matrices. It is a diagonal matrix containing the singular values ​​of T. The SVD of T can be applied to the above conditions to obtain: Equation (7) For any real-valued input vector a, we can see that if we take s as... The reciprocal of the maximum absolute value of |Σ|, which corresponds to the singular value of T with the maximum absolute value, will always satisfy the above condition and ensure the energy conservation of any input signal.

[0037] Design Examples

[0038] As a demonstration example, consider the Gaussian operator T. LOG The Laplace operator. The 1D Laplace operator of the Gaussian operator is given as a combination of the Laplace and Gaussian operators: T LOG =T L T G T L It is the matrix representation of the Laplace operator and T G It is the matrix representation of the Gaussian smoothing operator.

[0039] refer to Figure 5 Consider a housing 501 with 16 waveguide channels on the input and output sides. The dimensions of the design domain are set to 4λ × 8λ, where the first dimension is the thickness of the metasurface, the second dimension is the height of the metasurface, and λ = 1000 nm is the wavelength of the input wave in free space. The width of each individual waveguide is λ / 5, which allows only the propagation of the first transverse EM (TEM) mode.

[0040] Figure 5 The final metasurface construction obtained from reverse engineering is shown. Once metasurface 501 is obtained from reverse engineering, a frequency domain EM simulation is performed on metasurface 501, and the transformation encoded into the metasurface is calculated (i.e., the simulation readout process). Input signal 505 is applied to the simulated metasurface to generate the simulated transform output 503. Figure 503 shows the simulated transform versus the target transform T. LoG A comparison is also provided. The absolute difference 511 between each component of the analog transformation 509 and the target transformation 507 is also provided. As can be seen from the comparison, the reverse-designed metasurface 501 achieves the target transformation 507 with good accuracy (i.e., a maximum error of about 1%).

[0041] Furthermore, to demonstrate the full functionality of the optimized metasurface, consider, for example... Figure 5 The randomly selected input signal 505 is shown. An EM simulation is performed on the metasurface 501 by exciting each input waveguide channel according to the curve of the generated input signal 505, recording the amplitude and phase of the output waveguide 503, and converting them into vectors for both systems. A comparison of the simulated output vector 509 and the target output vector 507 (calculated through analysis) is shown. A good correlation 511 is observed between the simulation results and the target output.

[0042] Using the reverse design workflow proposed in this disclosure, any desired real-valued matrix multiplication can be encoded very accurately into a miniaturized photonic metasurface. Compared to advanced optical computing platforms such as MZI, the photonic metasurface obtained from the reverse design concept of this disclosure has the unique advantage of a significantly reduced device footprint (a size reduction of up to an order of magnitude while achieving the same computational task based on estimation). This is essentially because the reverse-designed metasurface achieves the target transformation through many-to-many coupling of the input-to-output waveguide channels, rather than the pairwise coupling used in MZI.

[0043] A promising potential application is the realization of fully optical neural networks using inversely designed photonic metasurface computing devices. The proposed inverse design enables any matrix multiplication on the photonic metasurface. Furthermore, with the aid of input and output waveguide channels, the resulting optical computing device is highly modular and can be easily coupled to other photonic metasurface computing devices via the input and output waveguides. This creates the possibility of encoding fully trained neural networks into photonic metasurface networks (in conjunction with nonlinear optical devices, which can also be inversely designed in principle). An advantage is that the realized optical neural networks can easily perform instantaneous predictions (at the speed of light) with minimal energy consumption. This makes them extremely attractive for applications requiring real-time prediction and control, such as vowel recognition, image detection, and decision-making in autonomous driving environments.

[0044] Quantitative comparison of equipment footprint and MZI

[0045] The most functionally similar existing optical computing platform to the one proposed in this IDF (i.e., performing matrix operations) is the MZI. However, initial analysis shows that a reverse-designed metasurface can achieve the same functionality as the MZI, but with a significantly reduced device footprint due to its many-to-many coupling characteristics.

[0046] More specifically, MZI requires pairwise coupling between input channels rather than all-to-all coupling. If matrix T is unitary (i.e., all singular values ​​are 1), then performing the transformation T requires dispersing information over a large neighborhood and using waveguides to guide the optical flow between different channels. The number of 2x2 couplers required for N channels is... This increases rapidly with the number of channels: 8 channels require 28, 16 channels require 120, 32 channels require 496, and 64 channels require 2016. In optimal packaging, the footprint increases linearly with the number of channels in both the lateral and propagation directions. Each MZI can be 10 x 100 μm in size, so a device with 32 channels would require approximately 0.32–3.2 mm on each side.

[0047] This becomes even worse when considering non-unitary T, where its SVD must be performed to achieve this. In this case, besides from singular values... In addition to scaling, we must also implement two unitary operators V. T And U. This is necessary because the interferometer does not leak any significant power, so there is no way to reduce the signal within the MZI grid. Therefore, for non-unitary operators, the device needs slightly more than twice the area in the propagation direction.

[0048] Conversely, the device of this invention, designed via topology optimization, can be implemented with only 0.5 μm of waveguide per channel in the lateral direction (wavelength λ is considered to be 1 μm). No strict scaling rule has been identified to determine how much material is needed in the propagation direction, but it has been found that 3 μm for 8 channels, 4 μm for 16 channels, and 12 μm for 32 channels represent at least an order of magnitude reduction in area compared to MZI. This indicates that the proposed metasurface significantly outperforms the MZI architecture in terms of area footprint.

[0049] Compare with recent work on reverse engineering of photonic metasurfaces

[0050] The recent work by Qu, Y, et al., “Inverse Design of an Integrated-nonphotonics Optical Neural Network,” Scientific Bulletin, 65(14), pp. 1177-1183, introduces an inverse design concept for designing optical scattering units using input and output waveguide channels for optical neural network applications. The inverse design method and concept proposed in this disclosure employ different approaches with different concepts and are more versatile in terms of capability. In particular, while other works only consider the signal intensity in the waveguide channel to encode matrix operators, the inverse design concept introduced in this disclosure utilizes both the signal intensity and phase information in the waveguide channel to encode a general matrix transformation. Furthermore, the inverse design procedure is also different; that is, the concept in previous work requires exciting all input waveguides together in a coupled manner, while the implementation of this disclosure excites each input waveguide individually in a decoupled manner. Therefore, the inverse design concept introduced in previous work can only encode unitary matrices (i.e., satisfying...). The reverse design concept can encode any real-valued matrix (even those with more than one singular value), while the rotation matrix of the reverse design concept can encode any real-valued matrix (even those with more than one singular value). Furthermore, the proposed reverse design framework appears to have better scalability compared to previous work. In embodiments of this disclosure, a device with 50 input and output waveguide channels can be implemented, while the largest device implemented in previous work contained only 9 input and output waveguide channels.

[0051] Figure 6An exemplary computing environment 600 in which embodiments of the present invention can be implemented is shown. Computers and computing environments (such as computer system 610 and computing environment 600) are known to those skilled in the art and are therefore briefly described herein.

[0052] like Figure 6 As shown, the computer system 610 may include communication mechanisms, such as a system bus 621 or other communication mechanisms for transmitting information within the computer system 610. The computer system 610 also includes one or more processors 620 coupled to the system bus 621 for processing information.

[0053] Processor 620 may include one or more central processing units (CPUs), graphics processing units (GPUs), or any other processor known in the art. More generally, a processor as used in this invention is a device for executing machine-readable instructions stored on a computer-readable medium to perform a task, and may include any or a combination of hardware and firmware. The processor may also include memory storing machine-readable instructions executable for performing the task. The processor operates according to information by manipulating, analyzing, modifying, transforming, or transmitting information used by an executable program or information device and / or by routing information to an output device. The processor may use or include the capabilities of, for example, a computer, controller, or microprocessor, and is modulated using executable instructions to perform special functions not performed by a general-purpose computer. The processor may be coupled (electrically coupled and / or include executable components) to any other processor to enable interaction and / or communication therebetween. A user interface processor or generator is a known element that includes electronic circuitry or software, or a combination of both, for generating a display image or portions thereof. The user interface includes one or more display images to enable user interaction with the processor or other devices.

[0054] Continue to refer to Figure 6The computer system 610 also includes a system memory 630 coupled to a system bus 621 for storing information and instructions executed by the processor 620. The system memory 630 may include computer-readable storage media in the form of volatile and / or non-volatile memory, such as read-only memory (ROM) 631 and / or random access memory (RAM) 632. RAM 632 may include other dynamic storage devices (e.g., dynamic RAM, static RAM, and synchronous DRAM). ROM 631 may include other static storage devices (e.g., programmable ROM, erasable PROM, and electrically erasable PROM). Furthermore, the system memory 630 may be used to store temporary variables or other intermediate information during instruction execution by the processor 620. A basic input / output system 633 (BIOS), containing basic routines such as those facilitating the transfer of information between components within the computer system 610 during startup, may be stored in the ROM 631. RAM 632 may contain data and / or program modules that are readily accessible and / or currently in operation by the processor 620. The system memory 630 may also include, for example, an operating system 634, an application program 635, other program modules 636, and program data 637.

[0055] Computer system 610 also includes a disk controller 640 coupled to system bus 621 to control one or more storage devices, such as magnetic hard disk 641 and removable media drives 642 (e.g., floppy disk drives, optical disk drives, tape drives, and / or solid-state drives), for storing information and instructions. Storage devices can be added to computer system 610 using appropriate device interfaces, such as Small Computer System Interface (SCSI), Integrated Device Electronics (IDE), Universal Serial Bus (USB), or FireWire.

[0056] Computer system 610 may also include a display controller 665 coupled to system bus 621 to control a display or monitor 666, such as a cathode ray tube (CRT) or liquid crystal display (LCD), for displaying information to a computer user. The computer system includes an input interface 660 and one or more input devices, such as a keyboard 662 and a clicking device 661, for interacting with the computer user and providing information to processor 620. Clicking device 661 may be, for example, a mouse, light pen, trackball, or pointer, for transmitting directional information and command selections to processor 620 and for controlling cursor movement on display 666. Display 666 may provide a touchscreen interface that allows input to supplement or replace the transmission of directional information and command selections by clicking device 661. In some embodiments, a user-wearable augmented reality device 667 may provide input / output functionality that allows the user to interact with both the physical and virtual worlds. Augmented reality device 667 communicates with display controller 665 and user input interface 660, allowing the user to interact with virtual items generated by display controller 665 within augmented reality device 667. Users can also provide gestures that are detected by augmented reality device 667 and transmitted as input signals to user input interface 660.

[0057] Computer system 610 may perform some or all of the processing steps of embodiments of the present invention in response to processor 620 executing one or more sequences of one or more instructions contained in memory (such as system memory 630). Such instructions may be read into system memory 630 from another computer-readable medium (such as magnetic hard disk 641 or removable media drive 642). Magnetic hard disk 641 may contain one or more data storage and data files used by embodiments of the present invention. The data storage contents and data files may be encrypted to improve security. Processor 620 may also be employed in a multiprocessing device to execute one or more sequences of instructions contained in system memory 630. In alternative embodiments, hard-wired circuitry may be used in place of or in combination with software instructions. Therefore, the embodiments are not limited to any particular combination of hardware circuitry and software.

[0058] As described above, computer system 610 may include at least one computer-readable medium or memory for holding instructions programmed according to embodiments of the invention and for containing data structures, tables, records, or other data described herein. As used herein, the term "computer-readable medium" refers to any medium that participates in providing instructions to processor 620 for execution. Computer-readable media may take many forms, including but not limited to non-transient, non-volatile, volatile, and transmission media. Non-limiting examples of non-volatile media include optical discs, solid-state drives, magnetic disks, and magneto-optical discs, such as magnetic hard disk 641 or removable media drive 642. Non-limiting examples of volatile media include dynamic memory, such as system memory 630. Non-limiting examples of transmission media include coaxial cables, copper wires, and optical fibers, including wires constituting system bus 621. Transmission media may also take the form of sound waves or light waves, such as those generated during radio wave and infrared data communication.

[0059] The computing environment 600 may also include a computer system 610 operating in a networked environment using logical connections to one or more remote computers, such as remote computing device 680. Remote computing device 680 may be a personal computer (laptop or desktop), mobile device, server, router, network PC, peer-to-peer device, or other public network node, and typically includes many or all of the elements described above with respect to computer system 610. When used in a networked environment, computer system 610 may include a modem 672 for establishing communication over a network 671, such as the Internet. Modem 672 may be connected to system bus 621 via user network interface 670 or via another suitable mechanism.

[0060] Network 671 can be any network or system generally known in the art, including the Internet, intranet, local area network (LAN), wide area network (WAN), metropolitan area network (MAN), direct connection or cascade, cellular telephone network, or any other network or medium capable of facilitating communication between computer system 610 and other computers (e.g., remote computing device 680). Network 671 can be wired, wireless, or a combination thereof. Wired connections can be implemented using Ethernet, Universal Serial Bus (USB), RJ-6, or any other wired connection generally known in the art. Wireless connections can be implemented using Wi-Fi, WiMAX and Bluetooth, infrared, cellular networks, satellite, or any other wireless connection method generally known in the art. Additionally, several networks can operate independently or communicate with each other to facilitate communication within network 671.

[0061] As used herein, an executable application includes code or machine-readable instructions for regulating the processor to perform predetermined functions (such as those of an operating system, contextual data acquisition system, or other information processing system) in response to user commands or input. An executable program is a piece of code or machine-readable instructions, a subroutine, or other distinct code segment or portion of an executable application for performing one or more specific processes. These processes may include receiving input data and / or parameters, performing operations on the received input data and / or performing functions in response to received input parameters, and providing the resulting output data and / or parameters.

[0062] As used in this invention, a graphical user interface (GUI) includes one or more display images generated by a display processor and enabling a user to interact with the processor or other devices and associated data acquisition and processing functions. The GUI also includes an executable program or executable application. The executable program or executable application modulates the display processor to generate signals representing the GUI display images. These signals are provided to a display device, which displays images for the user to view. Under the control of the executable program or executable application, the processor manipulates the GUI display images in response to signals received from an input device. In this way, the user can interact with the display images using an input device, thereby enabling the user to interact with the processor or other devices.

[0063] The functions and process steps described herein can be executed automatically, fully or partially in response to user commands. Automatically executed activities (including steps) are performed in response to one or more executable instructions or device operations, without requiring direct user initiation.

[0064] The systems and processes shown in the accompanying drawings are not unique. Other systems, processes, and menus can be derived from the principles of the invention to achieve the same purpose. Although the invention has been described with reference to specific embodiments, it should be understood that the embodiments and variations shown and described herein are for illustrative purposes only. Modifications to the present design can be made by those skilled in the art without departing from the scope of the invention. As described herein, different systems, subsystems, agents, managers, and processes can be implemented using hardware components, software components, and / or combinations thereof. Unless the claim element is explicitly stated using the phrase “means for…”, it should not be interpreted in this invention in accordance with the provisions of paragraph 6 of 35 USC 112.

Claims

1. A computer-implemented method for designing an optical computing device having multiple input waveguides, a photonic metasurface, and multiple output waveguides, wherein, The plurality of input waveguides are configured to receive an optical signal at each input waveguide, the method comprising: Determine the target transformation of the optical computing device; Perform multiple optimization steps to determine the topology of the photonic metasurface, each step including: One by one, the input waveguides are excited; The energy of the input and output regions is measured to determine the contribution of the current input waveguide, wherein the metasurface guides and scatters light provided as an optical input signal, and wherein the topology of the metasurface determines the optical transmission characteristics of the photonic metasurface. Sum the contributions of all input waveguides; The summed contribution is compared with the target transformation to determine the loss function value; and The design parameter set is updated based on the loss function value.

2. The computer-implemented method according to claim 1 further includes: The design parameter set is updated based on optimization to minimize the loss function value.

3. The computer-implemented method according to claim 2, wherein, The optimization is performed based on the finite memory BFGS algorithm.

4. The computer-implemented method according to claim 1, wherein, Parallel computation is used to determine the contributions of at least two input waveguides.

5. The computer-implemented method according to claim 1, further comprising: The target transformation is defined as a mathematical operation.

6. The computer-implemented method according to claim 5, wherein, The mathematical operation in question is matrix multiplication.

7. The computer-implemented method according to claim 6, wherein, The target transformation is scaled by a normalization factor.

8. The computer-implemented method according to claim 1, wherein, The loss function includes terms for applying the target electric field to the input region of the optical computing device and terms for applying the target electric field to the output region of the optical computing device.

9. The computer-implemented method according to claim 8, wherein, The term used to apply a target electric field to the input region of the optical computing device reduces the effect of backscattering of the input electromagnetic (EM) signal in the input region of the optical computing device.

10. The computer-implemented method according to claim 1, further comprising: For an optical computing device with eight input waveguide channels, the photonic metasurface is designed to have a thickness of 3 μm.

Images