Optical computing device

Abstract

The product of multi-bit input optical signals is calculated. [Solution] An optical arithmetic circuit 10 uses a star coupler 13 to optically perform a discrete cosine transform on a first input optical signal to generate a first intermediate optical signal. An optical arithmetic circuit 20 uses a star coupler 23 to optically perform a discrete cosine transform on a second input optical signal to generate a second intermediate optical signal. A multiplication circuit 30 generates a third intermediate optical signal indicating the product of the first and second intermediate optical signals. An optical arithmetic circuit 40 uses a star coupler 42 to optically perform an inverse discrete cosine transform on the third intermediate optical signal to generate an output optical signal indicating the product of the first and second input optical signals without carries. An output circuit 50 performs a carry operation on the output optical signal to generate and output an output signal indicating the product of the first and second input optical signals.

Description

[Technical Field]

[0001] The present disclosure relates to optical computing devices. [Background technology]

[0002] As the volume of information and communication traffic and the processing volume of computer systems continue to increase dramatically, and the demands for functionality continue to become more sophisticated, conventional computing technologies based on silicon integrated circuits are beginning to reach their limits in terms of processing speed and power consumption due to the limitations of semiconductor miniaturization.

[0003] Photonic computing is a promising new computing technology that can complement or replace conventional computing technologies due to the excellent properties of light, such as coherence, broadband, multiplexing, and quantum properties. To realize photonic computing, various optical processing elements and optical transmission elements have been developed.

[0004] For example, Patent Document 1 discloses an optical orthogonal frequency division multiplexing signal demultiplexing circuit equipped with a slab-type star coupler. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent No. 5400004 Summary of the Invention [Problem to be solved by the invention]

[0006] In conventional photonic computing technology, optical arithmetic circuits operate two bits at a time. To process more bits, optical arithmetic circuits must be connected in multiple stages. However, as the number of connected devices increases, the size, delay time, and power consumption increase. Therefore, there is a need to process multiple bits without excessive increases in size, delay time, and power consumption.

[0007] An object of the present disclosure is to provide an optical arithmetic device capable of calculating the product of multi-bit input optical signals without excessively increasing the size, delay time, and power consumption. [Means for solving the problem]

[0008] An optical arithmetic device according to a first aspect of the present disclosure, a first optical arithmetic circuit including a first star coupler, the first optical arithmetic circuit generating 2N-1 first intermediate optical signals by optically performing a discrete cosine transform on N (N≧2) first input optical signals using the first star coupler; a second optical arithmetic circuit including a second star coupler, the second optical arithmetic circuit generating 2N-1 second intermediate optical signals by optically performing a discrete cosine transform on N second input optical signals using the second star coupler; a multiplication circuit that generates 2N-1 third intermediate optical signals, each of which represents a product of one of the 2N-1 first intermediate optical signals and a corresponding one of the 2N-1 second intermediate optical signals; a third optical arithmetic circuit including a third star coupler, the third optical arithmetic circuit using the third star coupler to optically perform an inverse discrete cosine transform on the 2N-1 third intermediate optical signals to generate 2N-1 first output optical signals, each of which represents a product of the first and second input optical signals without carries; and an output circuit that generates and outputs 2N output signals that indicate the product of the first and second input optical signals by performing a carry operation on the 2N-1 first output optical signals.

[0009] According to the optical arithmetic device according to the second aspect of the present disclosure, in the optical arithmetic device according to the first aspect, the first star coupler has 4N-2 input ports and 2N-1 output ports; the first optical arithmetic circuit generates 2N third input optical signals from the N first input optical signals by duplicating each of the N first input optical signals, and further comprises a first input waveguide that inputs the 2N third input optical signals to 2N of the input ports of the first star coupler, respectively, and no optical signals are input to the remaining 2N-2 of the input ports of the first star coupler; the second star coupler has 4N-2 input ports and 2N-1 output ports; the second optical arithmetic circuit generates 2N fourth input optical signals from the N second input optical signals by duplicating each of the N second input optical signals, and further comprises second input waveguides that input the 2N fourth input optical signals to 2N of the input ports of the second star coupler, respectively, and no optical signals are input to the remaining 2N-2 of the input ports of the second star coupler; the third star coupler has 2N-1 input ports and 4N-2 output ports; The third optical operational circuit further includes output waveguides that generate the 2N-1 first output optical signals by multiplexing two second output optical signals output from the 4N-2 output ports of the third star coupler.

[0010] According to the optical arithmetic device according to the third aspect of the present disclosure, in the optical arithmetic device according to the first or second aspect, The multiplication circuit 2N-1 directional couplers each coupling one of the 2N-1 first intermediate optical signals with a corresponding one of the 2N-1 second intermediate optical signals; 2N-1 first optical receivers that convert output signals of the 2N-1 directional couplers into 2N-1 first electrical signals, respectively; and 2N-1 optical intensity modulators that convert the 2N-1 first electrical signals into the 2N-1 third intermediate optical signals, respectively.

[0011] According to the optical arithmetic device according to the fourth aspect of the present disclosure, in the optical arithmetic device according to one of the first to third aspects, The output circuit a second optical receiver that converts the 2N-1 first output optical signals into 2N-1 electrical signals, respectively; and processing circuitry for converting the 2N-1 electrical signals into 2N output signals by performing a carry operation on the 2N-1 electrical signals. [Effects of the Invention]

[0012] According to one aspect of the present disclosure, it is possible to calculate the product of multi-bit input optical signals without excessively increasing the size, delay time, and power consumption. [Brief explanation of the drawings]

[0013] [Figure 1] 1 is a diagram showing a configuration of an optical arithmetic device 100 according to an embodiment. [Figure 2] FIG. 2 is a diagram showing a detailed configuration of the optical arithmetic circuit 10 of FIG. [Figure 3] FIG. 2 is a diagram showing a detailed configuration of the optical arithmetic circuit 20 in FIG. [Figure 4] FIG. 2 is a diagram illustrating a detailed configuration of a multiplication circuit 30 in FIG. [Figure 5] FIG. 2 is a diagram showing detailed configurations of an optical arithmetic circuit 40 and an output circuit 50 in FIG. [Figure 6] 2 is a diagram for explaining the operation of star couplers 13, 23, and 42 in FIG. 1. FIG. [Figure 7] FIG. 5 is a diagram showing detailed configurations of a directional coupler 31 and a balanced optical receiver 32 shown in FIG. [Figure 8] 1. FIG. 4 is a first part of a table showing operations by the optical operation device of FIG. [Figure 9] 2 is a second part of a table showing operations by the optical operation device of FIG. 1; [Figure 10] 2 illustrates the operation of an exemplary input optical signal by the optical operation device of FIG. 1; DETAILED DESCRIPTION OF THE INVENTION

[0014] Hereinafter, optical arithmetic devices according to embodiments of the present disclosure will be described with reference to the drawings. The same reference numerals denote similar components throughout the drawings.

[0015] [Embodiment] [Configuration of optical computing device] 1 is a diagram showing the configuration of an optical arithmetic device 100 according to an embodiment. The optical arithmetic device 100 includes an optical arithmetic circuit 10, an optical arithmetic circuit 20, a multiplication circuit 30, an optical arithmetic circuit 40, and an output circuit 50. The optical arithmetic device 100 multiplies 3-bit input optical signals f and g together and outputs a 6-bit output optical signal h.

[0016] Multiplying 3-bit input optical signals f and g together is just one example, and the optical arithmetic device according to the embodiment may similarly process 2-bit input optical signals or 4-bit or more input optical signals.

[0017] The optical arithmetic circuit 10 includes a star coupler 13, which optically performs a discrete cosine transform on an input optical signal f to generate a first intermediate optical signal. The input optical signal f consists of three bits f(1), f(2), and f(3), where f(3) is the most significant bit and f(1) is the least significant bit. The first intermediate optical signal includes five signals s11, s12, s13, s14, and s15.

[0018] The optical arithmetic circuit 20 includes a star coupler 23, which optically performs a discrete cosine transform on the input optical signal g to generate a second intermediate optical signal. The input optical signal g consists of three bits g(1), g(2), and g(3), where g(3) is the most significant bit and g(1) is the least significant bit. The second intermediate optical signal includes five signals s21, s22, s23, s24, and s25.

[0019] Bits f(1), f(2), f(3), g(1), g(2), g(3) are pulses in phase with each other, whose non-zero amplitude represents a "1" and whose zero amplitude represents a "0".

[0020] The multiplication circuit 30 generates third intermediate optical signals s31, s32, s33, s34, and s35, each representing the product of one of the first intermediate optical signals s11, s12, s13, s14, and s15 and a corresponding one of the second intermediate optical signals s21, s22, s23, s24, and s25.

[0021] The optical arithmetic circuit 40 includes a star coupler 42, which is used to optically perform an inverse discrete cosine transform on the third intermediate optical signals s31, s32, s33, s34, and s35 to generate output optical signals s41, s42, s43, s44, and s45. The output optical signals s41, ..., s45 represent the respective digits of the product of the input optical signals f and g without a carry operation. In other words, the output optical signals s41, ..., s45 represent the respective digits of the product of the 3-bit input optical signals f and g calculated without a carry operation. s45 is the most significant digit, s44 is the next most significant digit, and so on, with s41 being the least significant digit. Each of the output optical signals s41, ..., s45 has a relative amplitude of 0, 1, 2, or 3, thereby representing one of the values ​​"0," "1," "2," and "3."

[0022] Output circuit 50 performs a carry operation on output optical signals s41, ..., s45 to generate and output output signal h, which indicates the product of input optical signals f and g. Output signal h consists of six bits h(1), h(2), h(3), h(4), h(5), and h(6), where h(6) is the most significant bit, h(5) is the next most significant bit, and so on, with h(1) being the least significant bit.

[0023] FIG. 2 is a diagram showing a detailed configuration of the optical arithmetic circuit 10 shown in FIG. 1. The optical arithmetic circuit 10 includes input waveguides 11-1 to 11-3, demultiplexers 12-1 to 12-3, a star coupler 13, and intermediate waveguides 14-1 to 14-5. The star coupler 13 is a slab-type star coupler formed on an optical waveguide substrate. The star coupler 13 has ten input ports p1 to p10 and five output ports q1 to q5. As will be described later with reference to FIG. 6, the star coupler 13 optically performs a discrete Fourier transform on signals input to the input ports p1 to p10 and outputs the converted signals from the output ports q1 to q5. The input waveguides 11-1 to 11-3 have Y-shaped paths that are demultiplexed into two by the demultiplexers 12-1 to 12-3, and transmit the bits f(1), f(2), and f(3) of the input optical signal f, respectively. As a result, input waveguide 11-1 duplicates bit f(1) and inputs the duplicated bit f(1) to two input ports p1 and p10 of star coupler 13, respectively. Similarly, input waveguide 11-2 duplicates bit f(2) and inputs the duplicated bit f(2) to two input ports p2 and p9 of star coupler 13, respectively. Similarly, input waveguide 11-3 duplicates bit f(3) and inputs the duplicated bit f(3) to two input ports p3 and p8 of star coupler 13, respectively. Intermediate waveguides 14-1 to 14-5 transmit the signals output from output ports q1 to q5 of star coupler 13 as first intermediate optical signals s11, s12, s13, s14, and s15. The first intermediate optical signals s11, s12, s13, s14, and s15 are the discrete cosine transforms of bits f(1), f(2), and f(3) of the input optical signal f.

[0024] For ease of explanation, the input optical signal f further includes virtual bits f(4) and f(5). Bit f(4) is the most significant bit adjacent to bit f(3), and bit f(5) is the most significant bit adjacent to bit f(4). Bits f(4) and f(5) have a fixed value of "0." To transmit bits f(4) and f(5), the optical processing circuit 10 further includes virtual input waveguides 11-4 and 11-5 and virtual demultiplexers 12-4 and 12-5. The virtual input waveguide 11-4 replicates bit f(4) and inputs the replicated bit f(4) to two input ports p4 and p7 of the star coupler 13, respectively. The virtual input waveguide 11-5 replicates bit f(5) and inputs the replicated bit f(5) to two input ports p5 and p6 of the star coupler 13, respectively. In reality, it is not necessary to provide the optical operational circuit 10 with the virtual input waveguides 11-4 and 11-5 and the virtual demultiplexers 12-4 and 12-5.

[0025] 3 is a diagram showing a detailed configuration of the optical arithmetic circuit 20 of FIG. 1. The optical arithmetic circuit 20 includes input waveguides 21-1 to 21-3, demultiplexers 22-1 to 22-3, a star coupler 23, and intermediate waveguides 24-1 to 24-5. The star coupler 23 has a configuration similar to that of the star coupler 13, and has ten input ports p1 to p10 and five output ports q1 to q5. The star coupler 23 optically performs a discrete Fourier transform on signals input to the input ports p1 to p10 and outputs the converted signals from the output ports q1 to q5. The input waveguides 21-1 to 21-3 have Y-shaped paths that are demultiplexed into two by the demultiplexers 22-1 to 22-3, and transmit bits g(1), g(2), and g(3) of the input optical signal g, respectively. As a result, input waveguide 21-1 duplicates bit g(1) and inputs the duplicated bit g(1) to two input ports p1 and p10 of star coupler 23, respectively. Similarly, input waveguide 21-2 duplicates bit g(2) and inputs the duplicated bit g(2) to two input ports p2 and p9 of star coupler 23, respectively. Similarly, input waveguide 21-3 duplicates bit g(3) and inputs the duplicated bit g(3) to two input ports p3 and p8 of star coupler 23, respectively. Intermediate waveguides 24-1 to 24-5 transmit the signals output from output ports q1 to q5 of star coupler 23 as second intermediate optical signals s21, s22, s23, s24, and s25. The second intermediate optical signals s21, s22, s23, s24, and s25 are the discrete cosine transforms of bits g(1), g(2), and g(3) of the input optical signal g.

[0026] For ease of explanation, the input optical signal g further includes virtual bits g(4) and g(5). Bit g(4) is the most significant bit adjacent to bit g(3), and bit g(5) is the most significant bit adjacent to bit g(4). Bits g(4) and g(5) have a fixed value of "0." To transmit bits g(4) and g(5), the optical processing circuit 20 further includes virtual input waveguides 21-4 and 21-5 and virtual demultiplexers 22-4 and 22-5. The virtual input waveguide 21-4 replicates bit g(4) and inputs the replicated bit g(4) to two input ports p4 and p7 of the star coupler 23, respectively. The virtual input waveguide 21-5 replicates bit g(5) and inputs the replicated bit g(5) to two input ports p5 and p6 of the star coupler 23, respectively. In reality, it is not necessary to provide the optical operational circuit 20 with the virtual input waveguides 21-4 and 21-5 and the virtual demultiplexers 22-4 and 22-5.

[0027] 4 is a diagram showing a detailed configuration of the multiplication circuit 30 shown in FIG. 1. The multiplication circuit 30 includes directional couplers 31-1 to 31-5, balanced optical receivers 32-1 to 32-5, a pulse light source 33, a demultiplexer 34, and optical intensity modulators 35-1 to 35-5. The directional couplers 31-1 to 31-5 each combine one of the first intermediate optical signals s11, s12, s13, s14, and s15 with a corresponding one of the second intermediate optical signals s21, s22, s23, s24, and s25. Each of the balanced optical receivers 32-1 to 32-5 converts two output signals of the corresponding directional couplers 31-1 to 31-5 into one electrical signal. The pulse light source 33 generates optical pulses and inputs the optical pulses to the optical intensity modulators 35-1 to 35-5 via the demultiplexer 34. The optical intensity modulators 35-1 to 35-5 modulate optical pulses with the electrical signals output from the balanced optical receivers 32-1 to 32-5, thereby converting the electrical signals into third intermediate optical signals s31, s32, s33, s34, and s35, respectively.

[0028] In FIG. 4, thick solid lines indicate paths of optical signals, and thick dashed lines indicate paths of electrical signals.

[0029] In this specification, the directional couplers 31-1 to 31-5 are also collectively referred to as "directional couplers 31," and the balanced optical receivers 32-1 to 32-5 are also collectively referred to as "balanced optical receivers 32."

[0030] FIG. 5 is a diagram showing the detailed configuration of the optical arithmetic circuit 40 and the output circuit 50 in FIG.

[0031] The optical operational circuit 40 includes intermediate waveguides 41-1 to 41-5, a star coupler 42, output waveguides 43-1 to 43-5, and multiplexers 44-1 to 44-5. The star coupler 42 has a configuration in which the inputs and outputs of the star couplers 13 and 23 are inverted, and has five input ports q1 to q5 and ten output ports p1 to p10. The star coupler 42 optically performs an inverse discrete Fourier transform on signals input to the input ports q1 to q5, and outputs the converted signals from the output ports p1 to p10. The intermediate waveguides 41-1 to 41-5 transmit third intermediate optical signals s31, s32, s33, s34, and s35, which are input to the input ports q1, q2, q3, q4, and q5 of the star coupler 42, respectively. The output waveguides 43-1 to 43-5 have Y-shaped paths that are respectively multiplexed by multiplexers 44-1 to 44-5. The output waveguides 43-1 to 43-5 transmit output optical signals output from output ports p1 to p10 of the star coupler 42, and multiplex the output optical signals two by two to generate output optical signals s41, s42, s43, s44, and s45. The output waveguide 43-1 multiplexes the output optical signals output from output ports p1 and p10 to generate output optical signal s41. The output waveguide 43-2 multiplexes the output optical signals output from output ports p2 and p9 to generate output optical signal s42. The output waveguide 43-3 multiplexes the output optical signals output from output ports p3 and p8 to generate output optical signal s43. The output waveguide 43-4 multiplexes the output optical signals output from the output ports p4 and p7 together to generate an output optical signal s44, and the output waveguide 43-5 multiplexes the output optical signals output from the output ports p5 and p6 together to generate an output optical signal s45.

[0032] The output circuit 50 includes photoreceivers 51-1 to 51-5 and a processing circuit 52. The photoreceivers 51-1 to 51-5 convert the output optical signals s41, s42, s43, s44, and s45 into electrical signals, respectively. The electrical signals output from the photoreceivers 51-1 to 51-5, like the output optical signals s41, ..., s45, are each represented with no carry in each digit of the number representing the product of the input optical signals f and g. The processing circuit 52 performs a carry operation on the electrical signals output from the photoreceivers 51-1 to 51-5 to convert the electrical signals into an output signal h.

[0033] In FIG. 5, thick solid lines indicate paths of optical signals, and thick dashed lines indicate paths of electrical signals.

[0034] The input waveguides 11-1 to 11-3 and 21-1 to 21-3 have the same length. The intermediate waveguides 14-1 to 14-5 and 24-1 to 24-5 have the same length. The intermediate waveguides 41-1 to 41-5 have the same length. The output waveguides 43-1 to 43-5 have the same length.

[0035] The optical arithmetic circuit 10, the optical arithmetic circuit 20, the directional couplers 31-1 to 31-5, and the optical arithmetic circuit 40 may be formed on one or more optical waveguide substrates.

[0036] [Star coupler configuration] Figure 6 is a diagram for explaining the operation of star couplers 13, 23, 42 in Figure 1. In the example of Figure 6, for the sake of explanation, reference is made to star coupler 60 having M (M is an even number equal to or greater than 6) input ports p1 to pM and M output ports q1 to qM.

[0037] In FIG. 6, reference numerals 61-1 to 61-M indicate waveguides to which optical signals are input, and reference numerals 62-1 to 62-M indicate waveguides from which optical signals are output.

[0038] Star coupler 60 has an incident surface S1 formed along an arc having a predetermined radius R and centered at a first point P1, and an exit surface S2 formed along an arc having a radius R and centered at a second point P2 on incident surface S1. First to Mth input ports p1 to pM are arranged in order on incident surface S1. First to Mth output ports q1 to qM are arranged in order on exit surface S2 corresponding to the first to Mth input ports p1 to pM, respectively, and at positions symmetrical to the first to Mth input ports p1 to pM with respect to midpoint P3 between first point P1 and second point P2. The positions of the M input ports and M output ports q1 to qM are set so that the relative phase shift φ(k, m) when an optical signal propagates from the kth input port pk to the mth output port qm (1≦k≦M, 1≦m≦M) satisfies the following equation, based on the phase shift when the optical signal propagates over a distance equal to the radius R in the star coupler 60.

[0039]

number

[0040] Moreover, the relative phase shift amount φ(k,m) satisfies the following equation.

[0041]

number

[0042] where n s is the refractive index of the slab-type optical waveguide that constitutes the star coupler 60, λ is the wavelength of the light, and D is the distance from the input port pk to the output port qm. m is the angle of the line passing through point P2 and output port qm with respect to the line passing through point P2 and point P1. k is the angle of the line passing through point P1 and input port pk with respect to the line passing through points P1 and P2.

[0043] The positions of the input ports p1 to pM and the output ports q1 to qM are set so that the right-hand side of equation (2) is equal to the right-hand side of equation (1).

[0044] Furthermore, phase shifters 63-1 to 63-M are inserted in the waveguides 61-1 to 61-M, respectively, so that a phase offset α(k)=(π / M)(M-1)(k-1) is applied to the optical signal input to the input port pk.

[0045] The phase offset may be imparted by increasing or decreasing the length of the waveguide, changing the refractive index of the waveguide by the thermo-optic effect, the electro-optic effect, ultraviolet light trimming, trimming using the photoelastic effect, or the like.

[0046] The phase difference Δφ between the optical signal propagating from the kth input port pk to the mth output port qm and the optical signal propagating from the k+1th input port p(k+1) to the mth output port qm is expressed as Δφ={φ(k+1,m)+α(k+1)}-{φ(k,m)+α(k)}=2π(m-1) / M. Therefore, when an optical signal I(k) is input to input port pk, the optical signal O(m) output from output port qm is expressed by the following equation.

[0047]

number

[0048] According to equation (3), it can be seen that the output optical signal is the discrete Fourier transform of the input optical signal.

[0049] As described with reference to FIG. 2, the optical operational circuit 10 includes a star coupler 13 having ten input ports p1 to p10 and five output ports q1 to q5, input waveguides 11-1 to 11-3 connected to the input ports p1 to p3 and p8 to p10, and intermediate waveguides 14-1 to 14-5 connected to the output ports q1 to q5. For the sake of explanation, FIG. 6 shows M output ports q1 to qM, equal to the number of input ports p1 to pM. However, in this embodiment, as shown in FIG. 2, only half the number of output ports q1 to q5 of the star coupler 13, i.e., half the number of input ports p1 to p10, are used. Phase shifters are inserted in the intermediate waveguides 14-1 to 14-5, respectively, to impart a phase offset π(m-1) / 10 to the optical signal output from output port qm (1≦m≦5). The denominator of the phase offset equation represents twice the number of output ports q1 to q5. In this case, the intermediate optical signal s(n) appearing in the intermediate waveguide 14-n is expressed by the following equation:

[0050]

number

[0051] Here, C1 = 1 / √2, C2 = C3 = C4 = C5 = 1. An attenuator, for example, a Mach-Zehnder interferometer type attenuator, is further inserted in the intermediate waveguide 14-1 so that the intermediate optical signal s(1) is attenuated by a coefficient C1 = 1 / √2 relative to the intermediate optical signals s(2), ..., s(5).

[0052] According to equation (4), the intermediate optical signal s(n) is the discrete cosine transform of the input optical signal f(k). The optical arithmetic circuit 10 optically performs the discrete cosine transform on the input optical signal f using the star coupler 13 to generate the first intermediate optical signals s11, ..., s15.

[0053] Similarly to the optical arithmetic circuit 10, the optical arithmetic circuit 20 also uses a star coupler 23 to optically perform a discrete cosine transform on the input optical signal g to generate second intermediate optical signals s21, . . . , s25.

[0054] On the other hand, star coupler 42 of optical arithmetic circuit 40 has a configuration in which the inputs and outputs of star couplers 13 and 23 are inverted, and performs an inverse discrete Fourier transform on the input optical signals. Phase shifters are inserted in each of intermediate waveguides 41-1 to 41-5 so as to impart a phase offset of -π(m-1) / 10 to the optical signals input to input ports qm (1≦m≦5) of star coupler 42. Optical arithmetic circuit 40 uses star coupler 42 to optically perform an inverse discrete cosine transform on third intermediate optical signals s31, ..., s35 to generate output optical signals s41, ..., s45.

[0055] [Configuration of directional coupler and balanced receiver] FIG. 7 is a diagram showing a detailed configuration of the directional coupler 31 and the balanced optical receiver 32 shown in FIG.

[0056] The directional coupler 31 includes waveguides 71 and 72 and phase shifters 73 and 74. The waveguides 71 and 72 have a coupling region 75 optically coupled to each other over a predetermined length. When an optical signal travels from the waveguide 71 to the waveguide 72, or vice versa, through the coupling region 75, the phase of the optical signal is shifted by π / 2. The phase shifter 73 is inserted between the input port and the coupling region 75 in the waveguide 71 and shifts the phase of the transmitted optical signal by π / 2. The phase shifter 74 is inserted between the coupling region 75 and the output port in the waveguide 72 and shifts the phase of the transmitted optical signal by π / 2. A first intermediate optical signal sa is input to the input port of the waveguide 71, and a second intermediate optical signal sb is input to the input port of the waveguide 72. Optical signals sc and sd are output from the output ports of the waveguides 71 and 72, respectively.

[0057] The balanced optical receiver 32 includes optical receivers 81 and 82. The optical receivers 81 and 82 receive the optical signals sc and sd and generate an electrical signal se. The electrical signal se is proportional to the difference in intensity between the optical signals sc and sd.

[0058] When the electric field E1(t) of the first intermediate optical signal sa is expressed as E1(t) = E1exp(jωt) and the electric field E2(t) of the second intermediate optical signal sb is expressed as E2(t) = E2exp(jωt), the electric signal se is expressed as a current I in the following equation:

[0059]

number

[0060] Here, E1 denotes the electric field amplitude of the first intermediate optical signal sa, E2 denotes the electric field amplitude of the second intermediate optical signal sb, ω is the angular frequency of the optical signal, and a is a constant.

[0061] As can be seen from equation (5), the electrical signal se is proportional to the product of the electric field amplitudes of the two intermediate optical signals sa and sb.

[0062] [Operation of the optical computing device] When r (1≦r≦5) represents each of the output optical signals s41, . . . , s45, the output optical signals s41, .

[0063]

number

[0064] As mentioned above, f(4) = f(5) = g(4) = g(5) = 0. Therefore, the output optical signals s41, ..., s45 are calculated from equation (6) as follows:

[0065]

number

number

number

number

number

[0066] The discrete cosine transform of the convolution of input optical signals f and g is equal to the product of the discrete cosine transforms of input optical signals f and g. Therefore, the product of input optical signals f and g can be calculated by calculating the product of the discrete cosine transforms of input optical signals f and g and then calculating its inverse discrete cosine transform.

[0067] Fig. 8 is a first part of a table showing the operations performed by the optical arithmetic unit of Fig. 1. Fig. 9 is a second part of a table showing the operations performed by the optical arithmetic unit of Fig. 1. According to Figs. 8 and 9, the product h of 3-bit input optical signals f and g having arbitrary values ​​can be correctly calculated by calculating the product of input optical signals f and g having binary values ​​without a carry operation to obtain output optical signals s45, ..., s41, and performing a carry operation on output optical signals s45, ..., s41 to obtain output signal h having a binary value.

[0068] 10 is a diagram illustrating an exemplary input optical signal operation by the optical operation device of FIG. 1. When a first input optical signal f=111 and a second input optical signal g=110 are input, the intermediate results of the multiplication of the input optical signals f and g (lines 3 to 5 of FIG. 10) are added together without a carry operation (addition without a carry), to obtain an output optical signal {s45, s44, s43, s42, s41}={1,2,2,1,0}. Next, a carry operation is performed on the output optical signal {1,2,2,1,0}, to obtain the final product of the input optical signals f and g, f×g=h={1,0,1,0,1,0}.

[0069] According to the optical arithmetic device 100 of the embodiment, it is possible to calculate the product of multi-bit input optical signals using a relatively small-scale circuit.

[0070] Furthermore, signals are processed at the speed of light in the optical arithmetic circuits 10, 20, and 40. Therefore, the optical arithmetic device 100 according to the embodiment can perform calculations faster than when products are calculated using electrical circuits, and the delay time due to calculations can be reduced.

[0071] Furthermore, the optical arithmetic device 100 according to the embodiment has a small circuit scale and can therefore operate with low power consumption.

[0072] In this way, the optical arithmetic device 100 according to the embodiment can calculate the product of 3-bit input optical signals without excessive increases in size, delay time, and power consumption.

[0073] [Other embodiments] As described above, f(4) = f(5) = g(4) = g(5) = 0, and therefore it is not necessary to actually provide the virtual input waveguides 11-4, 11-5, 21-4, and 21-5 and the virtual demultiplexers 12-4, 12-5, 22-4, and 22-5 in the optical arithmetic circuits 10 and 20. Therefore, the size of the optical arithmetic circuits 10 and 20 can be reduced below the size of the optical arithmetic circuit 40.

[0074] Furthermore, according to the optical arithmetic device of the embodiment, it is possible to process 2-bit input optical signals or 4-bit or more input optical signals in the same manner. In this case, the optical arithmetic device includes a first optical arithmetic circuit, a second optical arithmetic circuit, a multiplication circuit, a third optical arithmetic circuit, and an output circuit as described below. The first optical arithmetic circuit includes a first star coupler and uses the first star coupler to optically perform a discrete cosine transform on N (N≧2) first input optical signals to generate 2N−1 first intermediate optical signals. The second optical arithmetic circuit includes a second star coupler and uses the second star coupler to optically perform a discrete cosine transform on N second input optical signals to generate 2N−1 second intermediate optical signals. The multiplication circuit generates 2N−1 third intermediate optical signals, each of which represents the product of one of the 2N−1 first intermediate optical signals and a corresponding one of the 2N−1 second intermediate optical signals. The third optical arithmetic circuit includes a third star coupler, and uses the third star coupler to optically perform an inverse discrete cosine transform on the 2N-1 third intermediate optical signals to generate 2N-1 first output optical signals, each of which represents a product of the first and second input optical signals without a carry. The output circuit performs carry operations on the 2N-1 first output optical signals to generate and output 2N output signals that represent the product of the first and second input optical signals.

[0075] Also in this case, the first star coupler has 4N-2 input ports and 2N-1 output ports. The first optical processing circuit generates 2N third input optical signals from the N first input optical signals by duplicating each of the N first input optical signals, and further includes first input waveguides that input the 2N third input optical signals to 2N of the input ports of the first star coupler, respectively. No optical signals are input to the remaining 2N-2 of the input ports of the first star coupler. The second star coupler has 4N-2 input ports and 2N-1 output ports. The second optical processing circuit generates 2N fourth input optical signals from the N second input optical signals by duplicating each of the N second input optical signals, and further includes second input waveguides that input the 2N fourth input optical signals to 2N of the input ports of the second star coupler, respectively. No optical signals are input to the remaining 2N-2 of the input ports of the second star coupler. The third star coupler has 2N-1 input ports and 4N-2 output ports, and the third optical processing circuit further includes output waveguides that generate 2N-1 first output optical signals by multiplexing two by two the second output optical signals output from the 4N-2 output ports of the third star coupler.

[0076] In this case, the multiplication circuit includes 2N-1 directional couplers, 2N-1 first optical receivers, and 2N-1 optical intensity modulators as follows: The 2N-1 directional couplers each combine one of the 2N-1 first intermediate optical signals with a corresponding one of the 2N-1 second intermediate optical signals; The 2N-1 first optical receivers convert output signals of the 2N-1 directional couplers into 2N-1 first electrical signals, respectively; and the 2N-1 optical intensity modulators convert the 2N-1 first electrical signals into 2N-1 third intermediate optical signals, respectively.

[0077] In this case, the output circuit also includes a second photodetector that converts the 2N-1 first output optical signals into 2N-1 electrical signals, respectively, and a processing circuit that converts the 2N-1 electrical signals into 2N output signals by performing carry operations on the 2N-1 electrical signals.

[0078] In this way, according to the optical computing device of the embodiment, it is possible to calculate the product of input optical signals of any number of bits without excessively increasing the size, delay time, and power consumption, as in the case of calculating the product of 3-bit input optical signals. [Industrial Applicability]

[0079] An optical arithmetic device according to an embodiment of the present disclosure is applicable to the field of photonic computing, particularly hybrid optoelectronic computing, and is also applicable to speeding up multi-bit integer multiplication required for scientific and engineering calculations, public key calculations in public key cryptography, and the like. [Explanation of symbols]

[0080] 10 Optical calculation circuit 11-1~11-3 Input waveguide 11-4, 11-5 Virtual input waveguide 12-1~12-3 Duplexer 12-4,12-5 Virtual splitter 13 Star Coupler 14-1~14-5 Intermediate waveguide 20 Optical calculation circuit 21-1~21-3 Input waveguide 21-4, 21-5 Virtual input waveguide 22-1~22-3 Duplexer 22-4, 22-5 Virtual splitter 23 Star Coupler 24-1~24-5 Intermediate waveguide 30 Multiplication Circuit 31-1~31-5 Directional coupler 32-1~32-5 Balanced Receiver 33 Pulsed Light Source 34 Duplexer 35-1~35-5 Optical Intensity Modulator 40 Optical calculation circuit 41-1~41-5 Intermediate waveguide 42 Star Coupler 43-1~43-5 Output waveguide 44-1~44-5 Multiplexer 50 Output circuit 51-1~51-5 Receiver 52 Processing circuit 60 Star Coupler 61-1~61-M Waveguide 62-1~62-M Waveguide 63-1~63-M Phase shifter 71,72 Waveguide 73,74 Phase shifter 75 Combined area 81,82 Receiver 100 Optical calculation device

Claims

1. a first optical arithmetic circuit including a first star coupler, the first optical arithmetic circuit generating 2N-1 first intermediate optical signals by optically performing a discrete cosine transform on N (N≧2) first input optical signals using the first star coupler; a second optical arithmetic circuit including a second star coupler, the second optical arithmetic circuit generating 2N-1 second intermediate optical signals by optically performing a discrete cosine transform on N second input optical signals using the second star coupler; a multiplication circuit that generates 2N-1 third intermediate optical signals, each of which represents a product of one of the 2N-1 first intermediate optical signals and a corresponding one of the 2N-1 second intermediate optical signals; a third optical arithmetic circuit including a third star coupler, the third optical arithmetic circuit using the third star coupler to optically perform an inverse discrete cosine transform on the 2N-1 third intermediate optical signals to generate 2N-1 first output optical signals each representing a product of the first and second input optical signals without carries; an output circuit that generates and outputs 2N output signals indicating a product of the first and second input optical signals by performing a carry operation on the 2N−1 first output optical signals; Optical computing device.

2. the first star coupler has 4N-2 input ports and 2N-1 output ports; the first optical arithmetic circuit generates 2N third input optical signals from the N first input optical signals by duplicating each of the N first input optical signals, and further comprises a first input waveguide that inputs the 2N third input optical signals to 2N of the input ports of the first star coupler, respectively, and no optical signals are input to the remaining 2N-2 of the input ports of the first star coupler; the second star coupler has 4N-2 input ports and 2N-1 output ports; the second optical arithmetic circuit generates 2N fourth input optical signals from the N second input optical signals by duplicating each of the N second input optical signals, and further comprises a second input waveguide that inputs the 2N fourth input optical signals to 2N of the input ports of the second star coupler, respectively, and no optical signals are input to the remaining 2N-2 of the input ports of the second star coupler; the third star coupler has 2N-1 input ports and 4N-2 output ports; the third optical arithmetic circuit further includes an output waveguide that generates the 2N-1 first output optical signals by multiplexing two by two second output optical signals output from the 4N-2 output ports of the third star coupler; 2. The optical computing device according to claim 1.

3. The multiplication circuit 2N−1 directional couplers each coupling one of the 2N−1 first intermediate optical signals with a corresponding one of the 2N−1 second intermediate optical signals; 2N-1 first optical receivers that convert output signals of the 2N-1 directional couplers into 2N-1 first electrical signals, respectively; and (2N-1) optical intensity modulators that convert the (2N-1) first electrical signals into the (2N-1) third intermediate optical signals, respectively.

2. The optical computing device according to claim 1.

4. The output circuit a second optical receiver that converts the 2N-1 first output optical signals into 2N-1 electrical signals, respectively; a processing circuit for converting the 2N-1 electrical signals into 2N output signals by performing a carry operation on the 2N-1 electrical signals; 2. The optical computing device according to claim 1.

Images