CN113452449A - Optical device for unitary matrix calculation of time domain signals - Google Patents
Optical device for unitary matrix calculation of time domain signals Download PDFInfo
- Publication number
- CN113452449A CN113452449A CN202110771423.3A CN202110771423A CN113452449A CN 113452449 A CN113452449 A CN 113452449A CN 202110771423 A CN202110771423 A CN 202110771423A CN 113452449 A CN113452449 A CN 113452449A
- Authority
- CN
- China
- Prior art keywords
- optical
- modulator
- delay line
- unit
- input
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
- H04B10/50—Transmitters
- H04B10/516—Details of coding or modulation
- H04B10/54—Intensity modulation
- H04B10/541—Digital intensity or amplitude modulation
Abstract
The optical device for performing unitary matrix calculation on time domain signals directly realizes corresponding unitary matrix calculation by utilizing the propagation process of light, so that the number of modulators and delay lines which need to pass different optical signals when the optical signals are propagated in the device is the same, the unevenness caused by the insertion loss of components can be ignored, and the fidelity of a target unitary matrix is greatly improved. The method has the advantages that the calculation speed is the light speed, the calculation time delay is extremely low, and the fidelity of unitary matrix calculation is greatly improved. The invention provides a high-efficiency calculation means, and can greatly accelerate the development of the fields of artificial intelligence, quantum calculation, wireless communication, life science and the like.
Description
Technical Field
The invention relates to the field of optical calculation, in particular to an optical device for performing unitary matrix calculation on time domain signals.
Background
Unitary matrices (also called unitary matrices) are a special complex field matrix form. Its conjugate transpose is equal to its inverse. Due to such special properties, the unitary matrix mathematically represents an operation of arbitrarily rotating one complex vector on the hilbert space; and physically may represent a system that performs any lossless transform on an input signal. Since signals in a large number of physical systems can be represented by complex vectors, unitary matrices play an important role in modeling physical systems. For example, the optical diffraction and transmission process may be described by a unitary matrix; the entanglement and interaction of optical photons can also be described by a unitary matrix. The calculation of an nxn unitary matrix requires O (N) as a calculation quantity2) Speed of simulation using a conventional computerIt drops sharply as the complexity of the simulation system increases. In order to meet the important requirements of technologies such as future quantum computation, artificial intelligence, wireless communication, life science and the like, the construction of a unitary matrix efficient computation system becomes important.
Taking advantage of the nature of optical coherent propagation, Reck et al propose an optical apparatus for unitary matrix computation of spatial domain signals (see Michael Reck and Anton Zeilinger, "Experimental correlation of and random singular operator," Physics Review Letters, vol.73, No.1, pp.58-61, 1994). The basic principle is that two optical phase shifters are arranged in a Mach-Zehnder interferometer, so that a unitary matrix calculation unit with 2 multiplied by 2 of one dimension is realized, and then a plurality of unitary matrix calculation units are utilized to form a triangular periodically-arranged net structure, thereby being equivalent to high-dimensional unitary matrix calculation. Since the mach-zehnder interferometers in the device proposed by Reck et al are arranged according to a triangular period, when light passes through the device, the insertion loss caused by each mach-zehnder interferometer is inherited, so that the insertion loss accumulated by the light at different input ports is inconsistent, and the fidelity of the unitary matrix finally realized is influenced. To solve this problem, Clements et al propose a unitary matrix calculation device for rectangular periodic arrangement of mach-zehnder interferometers (see William r. elements, Peter c. humpheys, Benjamin j. metacalf, w. steven Kolthammer, and Ian a. walmsley, "optical design for elementary multi-interferometers," optical, vol.3, No.12, pp.1460-1465,2016). The basic unit of this device is the same as that proposed by Reck et al. However, the triangular periodic arrangement is changed into the rectangular periodic arrangement by different high-dimensional unitary matrix decomposition methods, so that the insertion loss of the light accumulation input by different ports is largely homogenized, and the fidelity of the finally realized unitary matrix is improved.
However, the above schemes all focus on unitary matrix computation of the spatial domain signal. In many fields, the signal representation is in the time domain, and the unitary matrix calculation cannot be performed on the time domain signal by using the device. To solve this problem, Motes et al propose a device for performing unitary matrix computation on time domain signals (see Keith r. Motes, Alexei Gilchrist, Jonathan p. dowling, and Peter p. rohde1, "Scalable Boson sampling with time-bin encoding using a loop-based architecture," Physics Review Letters, vol. 113, No.120501,2014). Compared with a spatial domain unitary matrix calculating device, the calculating unit of the device is a Mach-Zehnder modulator and an optical loop structure, and the structure equivalently realizes one column of Mach-Zehnder interferometers of triangular periodically-arranged Mach-Zehnder interferometers. Therefore, the time domain unitary matrix calculation device proposed by Motes et al is physically equivalent to a spatial domain unitary matrix calculation device with triangular periodic arrangement, and naturally cannot avoid the problem of low fidelity of the unitary matrix. Meanwhile, due to the fact that the Mach-Zehnder modulator is used, the insertion loss of the unit structure is much larger than that of a Mach-Zehnder interferometer, the inconsistency of the insertion loss accumulated by signals at different time points is larger, and the fidelity of the unitary matrix of the device is greatly reduced.
Therefore, in order to realize high fidelity unitary matrix calculation of time domain signals, a completely new device structure is required to be provided, so as to avoid the problem caused by nonuniform insertion loss.
Disclosure of Invention
The invention aims to provide an optical device for performing unitary matrix calculation on a time domain signal, which directly realizes the corresponding unitary matrix calculation by utilizing the propagation process of light, has the calculation speed of light speed and extremely low calculation time delay, and greatly improves the fidelity of the unitary matrix calculation. The device can provide a high-efficiency calculation means, and can greatly accelerate the development of the fields of artificial intelligence, quantum calculation, wireless communication, life science and the like.
The technical scheme of the invention is as follows.
On one hand, the invention provides an optical device for performing unitary matrix calculation on time domain signals, which is characterized in that when the dimensionality of an input time domain signal is N, wherein N is a natural number more than or equal to 3, the required components of the device comprise that the number of double-input double-output Mach-Zehnder modulators (hereinafter referred to as modulators) is N +2, the number of one-unit optical delay lines is 2, and the number of two-unit optical delay lines is N-1;
the modulator comprises two optical input ports, two interference arms and two double output ports, a 1:1 optical splitter is arranged between the two optical input ports and the two interference arms, a second 1:1 optical splitter is arranged between the two interference arms and the two double output ports, a first phase shifter is positioned on one arm of the two interference arms, a second phase shifter is positioned in front of one optical output port, the first phase shifter is used for configuring the splitting ratio of the two optical output ports of the modulator, and the second phase shifter is used for applying an extra phase to one of the optical output ports;
the optical delay line is divided into a unit optical delay line and two unit optical delay lines, and each unit optical delay line comprises an optical input port and an optical output port; assuming that one delay unit is delta t, one unit of light delay line will introduce delta t delay, and two unit of light delay line will introduce 2 x delta t delay;
the connection sequence between the parts of the device is as follows:
an input optical signal is input by an optical input port of the first modulator; the calculation result is output by an optical output port of the last modulator; the modulators are alternately connected with the optical delay lines, namely, one output port of the previous modulator is connected with the input port of the optical delay line, and then the input port of the optical delay line is connected with the other output port of the previous modulator, and the modulators are alternately connected until the output end of the last optical delay line is connected with the input port of the last modulator; counting from the signal input end, the first connected optical delay line must be a unit optical delay line, the last connected optical delay line must be a unit optical delay line, and all the intermediate optical delay lines are two unit optical delay lines.
The input optical signal expresses an N-dimensional complex vector distributed in time, the output optical signal represents the N-dimensional complex vector calculated by the unitary matrix, the light field complex amplitudes of N different moments represent the value of the N-dimensional complex vector, and the time interval of each light field complex amplitude is a delay unit delta t.
On the other hand, the invention also provides another optical device for performing unitary matrix calculation on the time domain signal, which is characterized by comprising double-input double-output Mach-Zehnder modulators (hereinafter referred to as modulators) and optical delay lines, wherein the number of the modulators is M, M is more than or equal to 3, the optical delay lines comprise 1 long optical delay line, a first unit optical delay line, a second unit optical delay line and M-3 two unit delay lines, the delay length of the long delay line is more than or equal to NxDeltat, and Deltat is the time interval of each optical field complex amplitude.
The connection mode between the parts is as follows:
when M is 3, the optical signal is input from an input port of a first modulator, an output port of the first modulator is connected with the input port of a first unit optical delay line, and the output port of the unit optical delay line is connected with an input port of a second modulator; one output port of the second modulator is connected with the input port of the second unit optical delay line, and the output port of the second unit optical delay line is connected with the other optical input port of the third modulator; one output port of the third modulator outputs the calculation result, the other output port of the third modulator is connected with one end of the long delay line, and the other end of the long delay line is connected with the unused input port of the first modulator;
when M is greater than 3, an optical signal is input from an input port of the first modulator, the modulators are alternately connected with delay lines, namely, an output port of each modulator is connected with one delay line, each delay line is connected to one input port of the next modulator, an unused output port of the previous modulator is connected to an unused input port of the next modulator, and so on, all the M modulators, 2 one-unit delay lines and M-3 two-unit delay lines are connected, wherein the first delay line is a first one-unit optical delay line, the last delay line is a second one-unit optical delay line, and the middle delay lines are two-unit delay lines. One of the output ports of the last modulator outputs the calculation result, the other output port is connected with one end of a long delay line, and the other end of the long delay line is connected with an unused input port of the first modulator. And one output port of the second unit optical delay line is connected with one end of the long delay line, and the other end of the long delay line is connected with the unused input port of the first unit optical delay line.
The invention can be a system formed by discrete devices, an integrated chip platform or a system formed by the discrete devices and the integrated chip in a mixed way; when the device is implemented by discrete devices, the connection mode between the components can adopt a mode of limiting a light beam propagation path by using optical fibers, space optical devices and the like; when implemented using an integrated chip, the connections between the components may be made using integrated waveguides.
The modulator is formed based on an optical phase shift and a Mach-Zehnder interference structure, and the optical phase shift can be realized by utilizing a lithium niobate Pockels effect and a carrier dispersion effect.
The optical delay line can be an optical fiber delay line, an optical waveguide delay line, a slow optical delay line, a component for space free propagation delay or a component for realizing delay effect by utilizing optical storage time.
The working principle of the device is briefly described as follows.
The unitary matrix is a square matrix and satisfies the property that the inverse matrix is equal to its conjugate transpose. An N × N unitary matrix can transform an N-dimensional input complex vector. The purpose of this device is to make the optical signal transmission process in this device equivalent to a unitary matrix transform of NxN by properly configuring several phase shifters in this device. The input optical signal of the device expresses N-dimensional complex vectors distributed on a time domain, the complex amplitudes of the optical fields at N different moments represent the values of the N-dimensional complex vectors, and the time interval of each complex amplitude of the optical field is a delay unit delta t. The output optical signal of the device represents an N-dimensional complex vector after unitary matrix transformation.
With matrix elimination, an N × N unitary matrix can be decomposed into a number of N × N multiplication results of the unitary matrix, which is called decomposition of the unitary matrix. Different matrix elimination orders correspond to different matrix decomposition methods.
The unitary matrix decomposition method adopted by the device can be seen in (W.R. elements, et al, optical design for elementary multi-port indicators, optical, vol.3, No.12, pp.1460-1465, 2016). An N × N unitary matrix U can be decomposed as a result of multiplying (N-1) × N/2 unitary matrices T, and the formula is expressed as:
we call U the target unitary matrix, Tm,nReferred to as the rotation matrix, D is a diagonal matrix. Let the element in the ith row and the jth column in the rotation matrix be ti,jRotation matrix Tm,nIs characterized by tm,m，tm,n，tn,m，tn,nThese 4 elements form a 2 × 2 small unitary matrix, and the elements at other positions are an identity matrix, i.e., all 1 on the diagonal. S is a set of (m, N) values, where there are (N-1) × N/2 elements, and the decomposition method in the reference can determine the S set corresponding to each N. An expanded form of expression (1) when N is 4 is given in the embodiment.
The modulator in the device has the function that two phase shifters are configured, so that a 2 x 2 unitary matrix calculation can be applied to the complex amplitude of two optical fields input simultaneously, and the process is equivalent to realizing a rotation matrix. When the complex amplitudes of the two input light fields respectively represent the m-th dimension and the N-th dimension of the N-dimensional complex vector, the rotation matrix realized by the modulator is Tm,n. When the two optical signals input at the next moment change, the configuration of the two phase shifters is changed, ensuring that the correct rotation matrix is applied. The effect of the delay line in this arrangement is to delay the incoming optical signal to different degrees to ensure that the values represented by the complex amplitudes of the two optical fields input to the modulator at the same time are in the S set.
Because the cascade connection of the modulators can be physically equivalent to the multiplication of a matrix, under the configuration of a reasonable optical delay line and a phase shifter, a plurality of rotation matrix multiplications described by the expression (1) can be realized, so that the whole device forms a target unitary matrix and can carry out unitary matrix calculation on an input optical signal.
There is an equivalent variation of the present device. The optical delay line consists of three modulators, two one-unit optical delay lines and one long delay line. The delay length of the long delay line must be equal to or greater than nxat. The connection mode between the parts is as follows: an optical signal is input from one of the input ports of the first modulator; one output port of the first modulator is connected with an input port of a unit optical delay line, and the output port of the unit optical delay line and the other optical output port of the first modulator are respectively connected with two input ports of the second modulator; one output port of the second modulator is connected with the input port of another unit optical delay line, and the output port of the unit optical delay line and the other optical output port of the second modulator are respectively connected with two input ports of a third modulator; one output port of the third modulator outputs the calculation result, the other output port is connected with a long delay line, and the output of the long delay line is connected to the unused input port of the first modulator.
The working principle of this deformation device is similar to the above-described process. The second modulator applies a rotation matrix T to the complex amplitude of the input light fieldm,n. After passing through the rotation matrix, the optical signal enters the long delay line and is input into the second modulator again to apply the corresponding rotation matrix again, and the effect of multiplying the rotation matrix is formed. After N cycles, the transformation of the target unitary matrix is realized, and the final result is output through an output port of a third modulator.
The technical advantages of the invention are as follows:
(1) the corresponding unitary matrix calculation is directly realized by utilizing the propagation process of light, the calculation speed is the light speed, and the calculation time delay is extremely low. Since unitary matrix calculation is a basic mathematical model in the fields of artificial intelligence, quantum calculation, wireless communication, life science and the like, the development of the fields can be greatly accelerated by the device.
(2) Has extremely high fidelity. Because the number of modulators and delay lines which need to pass through when different optical field complex amplitudes are transmitted in the device is the same, unevenness caused by insertion loss of components can be ignored, the output calculation result only has uniform insertion loss, and the fidelity of the target unitary matrix is greatly improved.
Drawings
Fig. 1 is a schematic diagram of an optical apparatus for performing unitary matrix calculation on a time domain signal according to an embodiment of the present invention. The figure shows the components required and one possible connection between the components when N is 4.
Fig. 2 is a schematic diagram of the calculation process of the optical apparatus for performing unitary matrix calculation on time domain signals according to the present invention. (a) The correspondence of the first modulator input signal to the output signal is shown. (b) The correspondence of the second modulator input signal to the output signal is shown. (c) The third modulator input signal to output signal correspondence is shown. (d) The correspondence of the last modulator input signal to the output signal is shown.
Fig. 3 is a schematic diagram of an embodiment of a transforming apparatus of an optical apparatus for performing unitary matrix calculation on time domain signals according to the present invention, which describes the components required by the transforming apparatus and a possible connection manner between the components.
Detailed Description
The technical solution of the present invention is described in detail below with reference to the accompanying drawings and examples, and a detailed embodiment and structure are given, but the scope of the present invention is not limited to the following examples.
The optical device for performing unitary matrix calculation on time domain signals comprises the following components that when the dimensionality of an input time domain signal is N, wherein N is a natural number which is more than or equal to 3, the number of double-input double-output Mach-Zehnder modulators (hereinafter referred to as modulators) is N +2, the number of one-unit optical delay lines 200 is 2, and the number of two-unit optical delay lines 300 is N-1;
the modulator 100 comprises two optical input ports, two interference arms and two dual output ports, wherein a 1:1 optical splitter 101 is arranged between the two optical input ports and the two interference arms, a second 1:1 optical splitter 101 is arranged between the two interference arms and the two dual output ports, a first phase shifter 102 is arranged on one arm of the two interference arms, a second phase shifter 103 is arranged in front of one optical output port, the first phase shifter 102 is used for configuring the splitting ratio of the two optical output ports of the modulator, and the second phase shifter 103 is used for applying an additional phase to one of the two optical output ports;
the optical delay line is divided into a unit optical delay line 200 and a unit optical delay line 300, which both comprise an optical input port and an optical output port; assuming that a delay unit is Δ t, the delay of Δ t will be introduced by the one-unit optical delay line 200, and the delay of 2 × Δ t will be introduced by the two-unit optical delay line 300;
the connection sequence between the parts of the device is as follows:
an input optical signal is input by an optical input port of the first modulator; the calculation result is output by an optical output port of the last modulator; the modulators are alternately connected with the optical delay lines, namely, one output port of the previous modulator is connected with the input port of the optical delay line, and then the input port of the optical delay line is connected with the other output port of the previous modulator, and the modulators are alternately connected until the output end of the last optical delay line is connected with the input port of the last modulator; counting from the signal input end, the first connected optical delay line must be a unit optical delay line 200, the last connected optical delay line must be a unit optical delay line 200, and all the optical delay lines in the middle are two unit optical delay lines 300.
The optical device for performing unitary matrix calculation on time domain signals comprises three modulators 100, two one-unit optical delay lines 200 and one long delay line 400, wherein the delay length of the long delay line 400 is required to be greater than or equal to NxDeltat, and the connection mode among the components is as follows: an optical signal is input from one input port of the first modulator 100; an output port of the first modulator is connected to the input port of the unit optical delay line 200, and the output port of the unit optical delay line 200 is connected to an input port of the second modulator 100; an output port of the second modulator 100 is connected to an input port of another unit optical delay line 200, and an output port of this unit optical delay line 200 is connected to another optical output port of the third modulator 100; one output port of the third modulator 100 outputs the calculation result, the other output port is connected to one end of the long delay line 400, and the other end of the long delay line 400 is connected to an unused input port of the first modulator 100.
Example 1
Referring to fig. 1, a connection mode of components and an input/output mode of an optical signal when N is 4 are shown. The number of modulators needed in the device is 6, one unit optical delay line is 2, and two unit optical delay lines is 3, N + 2. The following description of "upper" and "lower" are based on the pose position plotted in fig. 1, and do not describe the position of the actual system. An optical signal is input through the lower input port of the first modulator 100. After passing through the first modulator, a one-unit delay line 200 is disposed between the upper output port and the second modulator, and a two-unit optical delay line 300 is disposed between the lower output port of the second modulator and the third modulator. And by parity of reasoning, the connection of the whole device is completed. The internal structure of each modulator is depicted in the diagram of fig. 1, and comprises two 1:1 optical splitters 101, two interference arms exist between the two optical splitters, a first phase shifter 102 is arranged on one of the interference arms to control the optical phase on the interference arm, and a second phase shifter 103 is arranged in front of one of the output ports to control the extra phase of the output port. After the optical signal propagates in the device, the calculation result is output by the lower output port of the last modulator.
Fig. 2 shows a detailed calculation process of the present apparatus. According to the above description, a unitary matrix can be decomposed into multiplication results of several unitary matrices by a matrix elimination method. The decomposition method adopted by the device is a method in a reference (W.R. entities, actual, optimal design for elementary multi-port transfermeters, optical, vol.3, No.12, pp.1460-1465,2016). According to this method, an N-4 unitary matrix can be decomposed into the result of multiplying 6 rotation matrices and a diagonal matrix, i.e.
U＝DT2,3T3,4T1,2T2,3T3,4T1,2 (2)
Assuming that the input complex vector is x, the unitary matrix calculation result is
U·x＝DT2,3T1,2T3,4T2,3T1,2T3,4·x (3)
And according to the combination law of matrix multiplication, sequentially calculating the multiplication results of the matrix and the vector from right to left to obtain a final calculation result.
Referring to fig. 2(a), the input optical signal is divided into 4 time instants in the time domain, and the time instants are separated by a delay unit Δ t. The complex amplitude of the light field at each time represents a complex number, and the different labels represent the positions of the complex amplitudes of the light field in the complex number vector. Thus, the input optical signal represents a complex vector of [ r, g, and g]T. By controlling the first phase shifter 102 of the first modulator, the transmittance of the modulator can be changed, thereby distributing the signals at different times to different output optical ports. In fig. 2(a), the second output port is allocated to the upper output port, and the third output port is allocated to the lower output port. Therefore, the second modulator will be entered after passing through a unit delay line, and the third modulator will be entered directly. See fig. 2 (b). When entering the second modulator, the third and fourth are at the same time, and the first and second are at the same time. When the third modulator enters the second modulator, the second modulator is configured to realize a T3,4Rotating the matrix; when the second modulator is entered, the second modulator is configured to realize a T1,2The matrix is rotated. Thereby realizing multiplication of the latter two rotation matrices in expression (3). The lower output port of the second modulator is connected with a two-unit delay line, so that the time of the third modulator is delayed backward by two units and then is input into the third modulator, and the fourth modulator is directly input into the third modulator. Thus, referring to FIG. 2(c), both (c) and (c) will be input to the third modulator simultaneously. A third modulator is configured to implement a T2,3And (4) rotating the matrix, namely completing the multiplication of the third last rotation matrix in the expression (3). The above process is repeated until the multiplication of all rotation matrices in expression (3) is completed. Fig. 2(d) shows the input-output situation of the last modulator. At this time, the process of (I), (II), (III) has been completedThe multiplication of all rotation matrixes is realized, and the final calculation result can be obtained only by configuring one phase. The 4 results are combined to one output port by the switching effect of the last modulator.
Example 2
Fig. 3 is a schematic diagram of a variant embodiment of the apparatus, which includes 3 modulators, a first one-unit delay line 200, a second one- unit delay line 201, and 1 long delay line 400. The long delay line 400 connects the upper output port of the second unit delay line 201 with the upper input port of the first unit delay line 200. The calculation process is similar to that described above, an optical signal is input from the lower input port of the first modulator 100, the rotation matrix is calculated in the second modulator 101 after passing through the first modulator 100 and the first one-unit delay line 200, the calculation result is combined into the long delay line 400 in the third modulator 102, the long delay line returns to the first modulator 101 through a loop to realize the loop calculation, and after the number of loop times required by the calculation is completed (N is 4 times of loop time), the calculation result is combined to the lower output port by the last modulator and output.
Experiments show that the calculation speed of the method is the light speed, the calculation time delay is extremely low, and the fidelity of unitary matrix calculation is greatly improved. The invention provides a high-efficiency calculation means, and can greatly accelerate the development of the fields of artificial intelligence, quantum calculation, wireless communication, life science and the like.
Claims (6)
1. An optical device for performing unitary matrix calculation on time domain signals is characterized in that when the dimensionality of an input time domain signal is N, wherein N is a natural number which is more than or equal to 3, the optical device comprises a dual-input dual-output Mach-Zehnder modulator (modulator) and optical delay lines, the number of the modulators is N +2, and the optical delay lines comprise 2 one-unit optical delay lines (200) and N-1 two-unit optical delay lines (300);
the modulator comprises two optical input ports, two interference arms and two dual-output ports, a 1:1 optical splitter (101) is arranged between the two optical input ports and the two interference arms, a second 1:1 optical splitter (101) is arranged between the two interference arms and the two dual-output ports, a first phase shifter (102) is arranged on one arm of the two interference arms, a second phase shifter (103) is arranged in front of one optical output port, the first phase shifter (102) is used for configuring the splitting ratio of the two optical output ports of the modulator, and the second phase shifter (103) is used for applying extra phase to one optical output port;
the unit optical delay line (200) and the unit optical delay line (300) both comprise an optical input port and an optical output port; setting a delay unit as delta t, leading the delay of delta t into one unit light delay line (200), and leading the delay of 2 multiplied by delta t into two unit light delay lines (300);
the connection sequence between the parts of the device is as follows:
an input optical signal is input by an optical input port of a first modulator (100), and a calculation result is output by an optical output port of a last modulator;
the modulators are alternately connected with the optical delay lines, namely, one output port of the previous modulator is connected with the input port of the optical delay line, and then the input port of the optical delay line is connected with the other output port of the previous modulator, and the modulators are alternately connected until the output end of the last optical delay line is connected with the input port of the last modulator; from the signal input end, the first connected optical delay line is a unit optical delay line (200), the last connected optical delay line is a unit optical delay line (200), and all the optical delay lines in the middle are two unit optical delay lines (300).
2. An optical apparatus as claimed in claim 1, wherein the input optical signal represents a time-distributed N-dimensional complex vector, the output optical signal represents the N-dimensional complex vector after the unitary matrix calculation, the optical field complex amplitudes at N different time instants represent the values of the N-dimensional complex vector, and the time interval of each optical field complex amplitude is a delay unit Δ t.
3. An optical device for performing unitary matrix calculation on time domain signals is characterized by comprising double-input double-output Mach-Zehnder modulators (hereinafter referred to as modulators) and optical delay lines, wherein the number of the modulators is M, M is more than or equal to 3, the optical delay lines comprise 1 long optical delay line (400), a first one-unit optical delay line (200), a second one-unit optical delay line (201) and M-3 two-unit delay lines (300), the delay length of the long delay line (400) is more than or equal to NxDeltat, and Deltat is the time interval of each optical field complex amplitude.
The connection mode between the parts is as follows:
when M is 3, an optical signal is input from an input port of a first modulator (100), an output port of the first modulator (100) is connected with an input port of a first unit optical delay line (200), and an output port of the unit optical delay line (200) is connected with an input port of a second modulator (101); one output port of the second modulator (101) is connected with the input port of the second unit optical delay line (201), and the output port of the second unit optical delay line (201) is connected with the other optical input port of the third modulator (102); an output port of the third modulator (102) outputs the calculation result, the other output port of the third modulator (102) is connected with one end of the long delay line (400), and the other end of the long delay line (400) is connected with the unused input port of the first modulator (100);
when M >3, an optical signal is input from one input port of a first modulator (100), each modulator being alternately connected to a delay line, that is, one output port of each modulator is connected with one delay line, each delay line is connected with one input port of the next modulator, the output port which is not used by the previous modulator is connected with the input port which is not used by the next modulator, by analogy, all M modulators, 2 one-unit delay lines and M-3 two-unit delay lines are connected, wherein, the first delay line is a first unit optical delay line (200), the last delay line is a second unit optical delay line (201), the middle delay lines are all two unit delay lines (300), one output port of the last modulator is connected with one end of a long delay line (400), the other end of the long delay line (400) is connected to an unused input port of the first modulator.
4. An optical apparatus as claimed in any of claims 1 to 3, characterized in that the implementation is discrete device composition, integrated chip platform, or a mixture of discrete devices and integrated chip; when the device is realized by discrete devices, the connection mode between the components adopts a mode of limiting a light beam propagation path by using optical fibers, space optical devices and the like; when implemented using an integrated chip, the connections between the components are made using integrated waveguides.
5. An optical apparatus as claimed in any of claims 1 to 3, wherein said modulator is based on optical phase shift and Mach-Zehnder interference structures, and said optical phase shift can be realized by utilizing the Czochralski effect and the carrier dispersion effect of lithium niobate.
6. An optical apparatus as claimed in any one of claims 1 to 3, wherein said optical delay line is an optical fiber delay line, an optical waveguide delay line, a slow optical delay line, a spatial free propagation delay or a delay effect using optical storage time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110771423.3A CN113452449B (en) | 2021-07-08 | 2021-07-08 | Optical device for unitary matrix calculation of time domain signals |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110771423.3A CN113452449B (en) | 2021-07-08 | 2021-07-08 | Optical device for unitary matrix calculation of time domain signals |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113452449A true CN113452449A (en) | 2021-09-28 |
| CN113452449B CN113452449B (en) | 2022-05-17 |
Family
ID=77815359
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110771423.3A Active CN113452449B (en) | 2021-07-08 | 2021-07-08 | Optical device for unitary matrix calculation of time domain signals |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113452449B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017210550A1 (en) * | 2016-06-02 | 2017-12-07 | Massachusetts Institute Of Technology | Apparatus and methods for optical neural network |
| CN110197277A (en) * | 2019-05-13 | 2019-09-03 | 浙江大学 | Realize the optical neural network method of number identification |
| CN110516802A (en) * | 2019-07-03 | 2019-11-29 | 上海交通大学 | Photon convolutional neural networks framework based on optical delay line caching |
| CN111723337A (en) * | 2020-06-16 | 2020-09-29 | 上海交通大学 | Photon tensor core integrated circuit architecture for neural network training and neural network training method thereof |
| CN112232504A (en) * | 2020-09-11 | 2021-01-15 | 联合微电子中心有限责任公司 | Photon neural network |
| CN112232503A (en) * | 2020-06-09 | 2021-01-15 | 联合微电子中心有限责任公司 | Computing device, computing method, and computing system |
| CN112384748A (en) * | 2018-05-15 | 2021-02-19 | 轻物质公司 | Photon processing system and method |
| US20210133547A1 (en) * | 2020-11-17 | 2021-05-06 | Intel Corporation | High efficiency optical neural network |
| US20210191232A1 (en) * | 2019-12-18 | 2021-06-24 | Xanadu Quantum Technologies Inc. | Apparatus and methods for spatio-temporal implementation of arbitrary unitary transformations on optical modes |
-
2021
- 2021-07-08 CN CN202110771423.3A patent/CN113452449B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017210550A1 (en) * | 2016-06-02 | 2017-12-07 | Massachusetts Institute Of Technology | Apparatus and methods for optical neural network |
| CN112384748A (en) * | 2018-05-15 | 2021-02-19 | 轻物质公司 | Photon processing system and method |
| CN110197277A (en) * | 2019-05-13 | 2019-09-03 | 浙江大学 | Realize the optical neural network method of number identification |
| CN110516802A (en) * | 2019-07-03 | 2019-11-29 | 上海交通大学 | Photon convolutional neural networks framework based on optical delay line caching |
| US20210191232A1 (en) * | 2019-12-18 | 2021-06-24 | Xanadu Quantum Technologies Inc. | Apparatus and methods for spatio-temporal implementation of arbitrary unitary transformations on optical modes |
| CN112232503A (en) * | 2020-06-09 | 2021-01-15 | 联合微电子中心有限责任公司 | Computing device, computing method, and computing system |
| CN111723337A (en) * | 2020-06-16 | 2020-09-29 | 上海交通大学 | Photon tensor core integrated circuit architecture for neural network training and neural network training method thereof |
| CN112232504A (en) * | 2020-09-11 | 2021-01-15 | 联合微电子中心有限责任公司 | Photon neural network |
| US20210133547A1 (en) * | 2020-11-17 | 2021-05-06 | Intel Corporation | High efficiency optical neural network |
Non-Patent Citations (1)
| Title |
|---|
| 徐绍夫等: "Optical Convolutional Neural Network With", 《IEEE PHOTONICS TECHNOLOGY LETTERS》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113452449B (en) | 2022-05-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Rafayelyan et al. | Large-scale optical reservoir computing for spatiotemporal chaotic systems prediction | |
| Tang et al. | Integrated reconfigurable unitary optical mode converter using MMI couplers | |
| WO2021227767A1 (en) | All-optical diffractive neural network and system implemented on optical waveguide and/or optical chip | |
| WO2023201970A1 (en) | Computing chip, system, and data processing method | |
| Gu et al. | SqueezeLight: Towards scalable optical neural networks with multi-operand ring resonators | |
| CN111723337B (en) | Photon tensor core integrated circuit architecture for neural network training and neural network training method thereof | |
| JPH0643957A (en) | Optical information processor | |
| US20220263582A1 (en) | Balanced photonic architectures for matrix computations | |
| WO2023109065A1 (en) | Method and system for implementing recurrent neural network, and electronic device and storage medium | |
| CN110908428A (en) | Parallel optical computing system for efficiently realizing large-scale matrix operation | |
| CN113452449B (en) | Optical device for unitary matrix calculation of time domain signals | |
| CN113589880B (en) | Optical device for simultaneously performing unitary matrix calculation on time domain signal and space domain signal | |
| WO2021170107A1 (en) | Optical computing device and system, and computing method | |
| CN116112088B (en) | Adaptive modulation method for multi-input multi-output type photon device | |
| CN114966967B (en) | Space light field regulation and control device based on super structure surface of guided wave drive | |
| CN113900283B (en) | On-chip integrated optical signal processor based on thin film lithium niobate electro-optical modulator array | |
| Dang | P-ReTiNA: Photonic Tensor Core Based Real-Time AI | |
| Arrathoon et al. | A high data rate integrated optical vector matrix multiplier | |
| Nakajima | Parallel neuromorphic computing on space and wavelength division multiplexed photonic processor | |
| Donn | Optics education for now and future from an entropy perspective | |
| Matthes et al. | Performing Linear Operations Using Optical Complex Media | |
| CN116108905A (en) | Photoelectric device unit, photoelectric neural network and calculation module | |
| Hasnain | Photonic Hardware Accelerators for Reservoir Computing | |
| CN117787368A (en) | Reconfigurable photon reservoir based on micro-ring resonator and baud rate adaptation method | |
| CN116720561A (en) | Photon processor based on optical frequency comb and double micro-ring modulator |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |