JPH04129423A - Superconducting multi-threshold logic circuit - Google Patents

Abstract

PURPOSE:To improve the learning efficiency with simple constitution by adopting a threshold logic circuit having plural threshold so as to realize an optional logic function without taking complicated circuit constitution. CONSTITUTION:The circuit consists of plural input lines 102a-102n, an output line 103, plural weighting sections 101a-101n, a threshold section 100, and a control line 104 and the circuit system suitable for the purpose finally is built up in which a weight Wi is changed. Then a threshold logic circuit having plural threshold levels, i.e., lots of threshold employing a magnetic flux coupling Josephson element is adopted for the threshold level section 100. Thus, a complicated function is realized with simple constitution and the learning is implemented simply at a high speed.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Application of the Invention) The present invention relates to a circuit configuration of a neuron using a superconducting circuit, particularly a multi-threshold logic circuit using a flux-coupled Josephson element.

(Background of the Invention) Conventional computers are constructed with a logic circuit system combining AND or OR circuits. It is a well-known fact that these computers operate at extremely high speeds, exhibit performance that far exceeds human computing ability, and contribute to society. However, it has become increasingly clear that conventional computers are unsuitable for the recognition and judgment operations that humans perform on a daily basis. For this reason, for the purpose of constructing computers suitable for recognition and judgment, threshold logic circuits modeled after human brain cells (neurons) and computer system technology using them have been developed, for example, by Toshi Seri, ``neurons''. “Mathematics of Circuit Networks” Industrial Books,
1978, L. D., Jackle, L.

R.E., Howard, H.P., Graf, B.
, Straughn, and J.D., Denk.
er, ” Artificial neural n
networks for computing, J.
Internal of Vacuum Science
eTechnology B4(1), Jan/F
eb, 1986. pp. 61-63.

The operation of a threshold logic circuit according to the prior art will be explained below to clarify the positioning of the present invention.

FIG. 4 is a diagram showing the operation of a conventional threshold logic circuit. A threshold logic circuit is a circuit having a plurality of input lines 102 and at least one output line 103. Input line 102
The plurality of input signals Xi inputted through the weighting unit 101
and is compared with a predetermined threshold T in the threshold section 200. The threshold unit 200 outputs an output signal F to the output line 103 based on the comparison result. In a threshold logic circuit, a digital signal Xi of "0 or 1" is applied to a plurality of input terminals, and if the weighted sum ΣW i X i of the digital signals Xi exceeds a threshold T, the output becomes "l". ”
Otherwise, it performs a logical operation that becomes "O". Here, Wi represents weight. Therefore, the operation of the threshold logic circuit shown in FIG. 4, ie, the relationship between the input signal Xi and the output signal F, is expressed by equation (1).

F=OΣW i X i < TF=IΣ
Wi X i ≧T . Therefore, in order to configure a threshold logic circuit, it is necessary to have a function of changing the weight Wi and a function of performing weight addition of input signals and switching elements. Normally, this weight is controlled by a weight control signal input from a control line 104.

In the prior art, the threshold value T is limited to one value, and, for example, the threshold characteristic shown in FIG. 5 is employed as an approximation of equation (1). The threshold characteristic shown in FIG. 5 is expressed by equation (2).

F = 1 / (1+ e x p (ΣWiX
i-T)) (2) There is a limit to the logic functions that can be realized by a threshold logic circuit in which only one threshold is defined according to the prior art. For example, the exclusive OR function of two people is expressed by equation (3), but this function cannot be expressed by equation (1) no matter how Wi is changed.

F=0; (Xl=0.X2=0)or
(X, = 1.X2 = 1) F ” 1 : (X + ” 1 + X
220) Or(X,=O, Adopts a configuration method that connects. For this reason, if you try to realize a complex logic function using a threshold logic circuit according to the prior art, you will have to adopt a complicated configuration in which many threshold logic circuits are connected in series. . in this case,
Not only is the circuit configuration complicated, but the control for changing the weight of the threshold logic circuit is complicated, the algorithm is extremely complicated and enormous, and the learning speed is slow.

(Objective of the Invention) An object of the present invention is to provide a threshold logic circuit that can realize any logical function without a complicated circuit configuration, and to realize a neuron element with a simple configuration and high learning efficiency. be.

(Summary of the Invention) For this purpose, the present invention employs a threshold logic circuit having multiple threshold values using magnetic flux-coupled Josephson elements. The flux-coupled Josephson element has periodic threshold values, and this periodicity can be used to define a plurality of threshold values.

(Example of the invention) The present invention will be explained below using Examples.

FIG. 1 shows an embodiment of the invention. The threshold logic circuit according to the embodiment of FIG.
3) Multiple weight sections 101) Threshold section 100, control line 10
It consists of 4. The roles of the plurality of input lines 102) output lines 103) the plurality of weighting units 101) control line 104 are the same as those explained in FIG. 4. On the other hand, the threshold section has a plurality of threshold values, and its operation is expressed by equation (4).

F=OT, ≦ΣW i X i < T 2F=I
T2≦ΣW i X i < T 5F=OT,
≦ΣW i X 1 < T <F=I T4≦Σ
W iX t < T s (4) If such a threshold logic circuit having a plurality of threshold values is employed, a complex logic function can be realized with one threshold logic circuit. For example, the exclusive OR expressed by equation (3) is also (
5) It can be expressed by the threshold numerical value shown in the equation.

F=0 0≦X + + X 2 < 0.5
F=10.5 ≦X+ +X2 < 1.5F=0
1.5 ≦X, +X2 ・ ・ ・ (5)
As shown in the example of equation (5), the number of logic functions that can be realized can be increased by using a threshold logic circuit having a plurality of threshold values. Theoretically, it is possible to realize any logical function by using a threshold logical function that can define a threshold on the infinite side. Therefore, by using a threshold logic circuit that can define a plurality of threshold values, a complex function can be realized with a simple configuration, and learning can be performed easily and quickly, thereby providing an extremely effective neuron circuit.

FIG. 2a shows an example of a two-junction flux-coupled quantum interference circuit that performs the circuit operation of equation (4). The two-junction flux-coupled quantum interference circuit shown in FIG. 2a is a well-known device in the field of Josephson elements, such as T, R, Gheewala,
” Josephson-Logic Device
s and Circuit'', IEEE Trans
, Electron Devices, vol.

ED-27, no, 10. I) Ri, 1857-1
869 is described in detail. The two-junction flux-coupled quantum interference circuit shown in Figure 2a consists of two Josephson junctions 20
1.202 and two inductors 203 and 204, a bias current Ig is supplied via a bias line 211 to a superconducting closed circuit 230. A control current line 210 is placed near the superconducting closed circuit 230, and the magnetic flux generated by the control current IC flowing through the control current line 210 interlinks with the superconducting closed circuit 230. control the maximum superconducting current (critical current) flowing in the FIG. 2b shows an example of the relationship between the control current Ic and the critical current of the two-junction flux-coupled quantum interference circuit shown in FIG. 2a.

The critical current has periodicity with respect to the control current, with the period being the control current that generates magnetic flux corresponding to the magnetic flux quantum.

Here, if a bias current Ig smaller than the critical current is passed through the two-junction flux-coupled quantum interference circuit, the circuit remains in a superconducting state, and if a bias current Ig larger than the critical current is passed, it transitions to a voltage state. For example, if the bias current Ig is set between the maximum and minimum values of the critical current as shown in Figure 2b, the two-junction flux-coupled quantum interference circuit will change the superconducting state and voltage state as the control current Ic changes. repeat. By selecting the bias current Ig, the range ratio between the superconducting state and the voltage state can be changed. 2 shown in Figure 2a.
If the control current 1c of the junction quantum interference circuit is the weighted sum of human input signals ΣW i X i, it is clear that this circuit performs the operation shown in equation (4). It can be adopted as an approximation of the threshold characteristic of equation (4). The threshold characteristic shown in FIG. 3 is expressed, for example, by equation (6).

F = 1 / (1+ e x p (ΣW i
X i T 2)) 1/(1+exp(ΣW i
X i T 3))+1/(1+exp(ΣW i
X i T 4))-1/(1+exp(ΣWiX
i −Ts))・・・・・・(6) Here, T<T2<TI<74<TS.

If equation (4) is approximated by equation (6), these functions can be differentiated, so the backpropagation method used as a learning method for neuron elements can be used.

FIG. 6 shows a configuration example of a threshold logic circuit using the magnetic flux coupled quantum interference circuit shown in FIG. 2a. In the threshold logic circuit of FIG. 6, the weight section 101 is a Josephson element 300.
A bias current is supplied to a variable current source 310 through the parallel connection of a load resistor 301 and a load resistor 301, and a bias current is supplied from a current source 220 to a Josephson element 250 in the threshold section 100. Here, Josephson element 250
For example, the two-junction magnetic flux coupling quantum interference shown in FIG. 2a is employed. The circuit operation of the weighting section 101 shown in FIG. 6 has already been disclosed in Japanese Patent Application No. 1961-15-1961 entitled "Superconducting Threshold Logic Circuit" by Harada. Since the threshold section 250 shown in FIG. 6 has a plurality of threshold values, it is clear that the threshold logic circuit shown in FIG. 6 performs the operation expressed by equation (4).

In the embodiments of the invention described above, a two-junction flux-coupled quantum interference circuit was used as the Josephson element in the threshold section 100.
It is clear that similar functions can be achieved using magnetic flux coupling quantum interference circuits with three or more junctions or current injection quantum interference circuits.

(Effects of the Present Invention) As described above, by using the present invention, a threshold logic circuit with a fast learning speed can be configured using a high speed Josephson switching circuit with a simple configuration.

Therefore, according to the present invention, it is possible to realize a high-speed computer suitable for executing recognition judgment using a threshold logic circuit.

Therefore, the present invention is indispensable for realizing a high-speed computer that performs this sophisticated recognition judgment.

[Brief explanation of drawings]

Figure 1 is a circuit diagram of an embodiment of the present invention, Figure 2a is a two-junction flux-coupled quantum interference circuit used in the present invention, and Figure 2b is a circuit diagram of a two-junction flux-coupled quantum interference circuit used in the present invention.
3 is an approximation diagram of the multi-threshold characteristic according to the present invention, FIG. 4 is a conventional threshold logic circuit diagram, and FIG. 5 is a diagram of the conventional threshold characteristic. The approximate diagram, FIG. 6, is a diagram showing a configuration example of a superconducting multi-threshold logic circuit according to the present invention. 100: L opening part, 101: Weight part, 102:
Input line, 103: Output line, 104: Control line,
200: L threshold, 201.202: Josephson junction 203.204: Inductor, control current line, 211: Bias line, current source,
230. Superconducting closed circuit, Josephson element, Josephson element, load resistance, 3IO: variable current source. Figure 1; T4〈ΣX1Wi≦T5 Figure 2a Figure 3 Figure 4 Figure 5 Figure 6

Claims (1)

[Claims] 1) A threshold logic circuit that compares a weighted sum of input signals with a predetermined threshold value and outputs an output signal according to the comparison result, wherein the weights are input from the outside. A multi-threshold logic circuit characterized in that it has means for changing and can perform comparisons using a plurality of threshold values. 2) A multi-threshold logic circuit according to claim 1, characterized in that it has a configuration in which a weighted sum signal of the input signals is input to a quantum interference circuit using a Josephson junction. Multithreshold logic circuit. 3) The superconducting multi-threshold logic circuit according to claim 2, wherein the quantum interference circuit is a flux-coupled quantum interference circuit. 4) A multi-threshold logic circuit according to claim 1, in which the threshold characteristic is expressed by a function F=1/(1+exp(ΣWiXi-T_2))-1/
(1+exp(ΣWiXi-T_3))+l/(1+e
xp(ΣWiXi-T_4))-1/(1+exp(Σ
WiXi-T_5))(T_1<T_2<T_3<T_
4<T_5 is a threshold value), and learning is performed using the function.