Resistively Coupled Ising Machine System and Method

The BRIM system addresses issues of voltage leakage and non-convex energy terrains in Ising machines by using ZIV diodes and spin-locked annealing, enhancing convergence to global minima and improving solution quality in complex computational problems.

JP2024542169AInactive Publication Date: 2024-11-13UNIVERSITY OF ROCHESTER
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
JP2024527174
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Priority Date
2021-11-08
Filing Date
2022-11-07
Publication Date
2024-11-13
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing Ising machines face challenges such as capacitor leakage leading to node voltages dropping to 0V, low equilibrium voltage levels making readings unreliable, and non-convex energy terrains causing convergence to local minima rather than global minima, especially in complex problems with numerous local minima.

Method used

The implementation of a bistable resistively coupled Ising machine (BRIM) with local feedback circuits using ZIV diodes to stabilize node voltages and spin-locked annealing to randomly reverse node spins, ensuring convergence to global minima.

Benefits of technology

The BRIM system effectively stabilizes node voltages and increases the probability of reaching global minima, improving solution quality and efficiency in solving complex computational problems like the maximum cut problem.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024542169000001_ABST
    Figure 2024542169000001_ABST
Patent Text Reader

Abstract

The Ising machine system includes a plurality of capacitors, a plurality of resistors having first and second terminals, each capacitor electrically connected via a respective positive and negative terminal thereof to a first or second terminal of at least one of the plurality of resistors, a power supply having positive and negative rails, a switching network configured to initialize each of the plurality of capacitors with a positive or negative voltage from the power supply and then disconnect all of the plurality of capacitors from the power supply to float the voltages of the capacitors, and at least one voltage measurement device configured to measure a voltage between a positive terminal and a negative terminal of one of the plurality of capacitors. A method of executing the Ising machine system is also disclosed.
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to U.S. Provisional Patent Application No. 63 / 276,849, filed November 8, 2021, which is incorporated by reference in its entirety. [Background technology]

[0002] In recent years, purpose-built designs have been increasingly employed due to their effectiveness in certain types of tasks, such as encryption or network operations. In the course of a related but different line of work, researchers have attempted to leverage the evolution of dynamical systems to effectively execute entire algorithms. In many recent examples, this evolution has the effect of optimizing a particular formula, called an Ising model, which we explain in more detail below. By configuring a dynamical system, the user sets the machine to optimize a particular Ising formula. The effect is thus that reading out the state of such a system at the end of the evolution will yield a solution (usually a very good solution) to the problem that it was mapped to.

[0003] The mechanisms that enable these systems to pursue the optimal state vary by design, but differ from von Neumann machines in that they have no explicit program to follow. Instead, these systems can be thought of as dedicated, physics-based computing machines that optimize a particular objective function (in the form of the Ising equations). As such, they are commonly referred to as Ising machines. Ising machines have been implemented in a variety of ways involving very different (and often complex) physical principles. It remains to be seen whether more complex structures (such as quantum annealers) offer any fundamental advantages at very large scales. In the near term, however, the public electronic versions offer a much more attractive value proposition.

[0004] The Ising model, named after the physicist Ernst Ising, is a mathematical model of ferromagnetism in statistical mechanics. The model consists of discrete variables representing the magnetic dipole moments (σ) of atomic spins, which can be in one of two states (+1 or -1). The spins are arranged in a lattice, where each spin can interact with its neighbors. The energy of coincident neighboring spins is lower than the energy of mismatched spins. Such systems tend to have the lowest energy and are described by the Hamiltonian function in Equation 1 below.

number

[0005] Consider a graph G = (V,E,W), where V represents vertices, E represents edges connecting vertices, and W is the weight of the corresponding edge, which may be negative or non-integer. A cut is a process of dividing V into two sets (V + and V - ), where the cut value or solution is defined as the combination of edge weights that span the two sets of vertices. In particular, if the solution is maximal, it is called a max-cut. A graphical demonstration of this procedure is shown in Figure 1.

[0006] The red line 101 has a vertex V + ={1,4} and V - ={2,3,5}. After some rearrangements, a readable representation of the solution is obtained on the right. To emphasize the cut action, dashed line 102 divides the same set (E i =E j ), and solid lines 103 represent edges connecting vertices in the set (E i ≠Ej ), it is worth pointing out that the cut action does not change the total weight. The reason for this is that the total weight is a fixed value for a particular graph, and is mathematically shown in Equation 2 below.

number

[0007] Here, V + The vertices in have a spin value of σ=+1, and V - Assuming that the vertices in take the value σ=−1, if we multiply each pair of vertices by their corresponding weight, we sum them together to obtain Equation 3 below.

number

[0008] Combining equation 2 and equation 3 and eliminating common terms, we obtain an expression similar to the Ising model, as shown in equation 4 below.

number

[0009] Since the first term is a constant, when H' is minimized, the second term is maximized. This is exactly what the max-cut algorithm searches for. Therefore, for a graph problem (W ij ) for a specific Ising machine configuration (J ij ), when the machine finds the ground state of the Hamiltonian, it is equivalent to a maximum cut. Summary of the Invention

[0010] In one aspect, an Ising machine system includes a plurality of capacitors having respective positive and negative terminals; a plurality of resistors having first and second terminals, each capacitor electrically connected via each of its positive and negative terminals to a first or second terminal of at least one of the plurality of resistors; a power supply having positive and negative rails; a switching network configured to initialize each of the plurality of capacitors with a positive or negative voltage from the power supply and then disconnect all of the plurality of capacitors from the power supply to float the voltages of the capacitors; and at least one voltage measurement device configured to measure a voltage between the positive and negative terminals of one of the plurality of capacitors.

[0011] In one embodiment, the system further comprises a feedback circuit configured to charge at least one capacitor of the plurality of capacitors when the voltage level of the capacitor is below the voltage of the power supply and to discharge the capacitor when the voltage level of the capacitor is about to exceed the voltage of the power supply. In one embodiment, the feedback circuit comprises two back-to-back inverters. In one embodiment, the two back-to-back inverters are implemented in CMOS technology. In one embodiment, the switching network comprises a set of switches electrically connected to a positive or negative terminal of each capacitor of the plurality of capacitors and configured to form a momentary electrical contact between the positive or negative terminal of each capacitor and a positive or negative rail of the power supply. In one embodiment, the switching network is further configured to randomly select a positive or negative terminal of one of the capacitors of the plurality of capacitors after disconnecting all of the plurality of capacitors from the power supply, and momentarily connect and then disconnect the randomly selected terminal to the positive or negative rail of the power supply.

[0012] In one aspect, an Ising machine system comprises a non-transitory computer readable medium having instructions stored thereon that, when executed by a processor, perform steps including emulating a circuit network, the emulated circuit network comprising a plurality of capacitors, each having a positive terminal and a negative terminal, a plurality of resistors having first and second terminals, each capacitor electrically connected via each of its positive and negative terminals to a first or second terminal of at least one of the plurality of resistors, and a power supply having positive and negative rails; initializing each of the plurality of capacitors with a positive or negative voltage from the rails of the power supply; disconnecting all of the plurality of capacitors from the power supply to float the voltages of the capacitors; and measuring and recording a voltage between the positive and negative terminals of at least one of the plurality of capacitors over a period of time.

[0013] In one embodiment, each step further includes randomly selecting a positive or negative terminal of one of the capacitors of the plurality of capacitors after disconnecting all of the plurality of capacitors from the power supply, and momentarily connecting and then disconnecting the randomly selected terminal to the positive or negative rail of the power supply.

[0014] In one aspect, a method of executing an Ising machine includes providing a plurality of resistors and capacitors and a power source; connecting each of the plurality of capacitors to at least one other of the plurality of capacitors through two of the plurality of resistors; initializing each of the plurality of capacitors with a positive or negative voltage from the power source; disconnecting the power source from the plurality of capacitors to float the voltages of the capacitors; and measuring and recording a voltage between a positive terminal and a negative terminal of at least one of the plurality of capacitors over a period of time.

[0015] In one embodiment, the method further includes measuring and recording voltages between all of the positive and negative terminals of the plurality of capacitors over a period of time. In one embodiment, the method further includes designating capacitors having a positive voltage after the period of time as a first set and designating capacitors having a negative voltage after the period of time as a second set, the first set and the second set corresponding to nodes in the max-cut analysis.

[0016] The above and other objects and features will become apparent from the following description and by reference to the accompanying drawings, which are included to provide an understanding of the invention and are incorporated in and constitute a part of this specification, and in which like reference numerals represent like elements. [Brief description of the drawings]

[0017] [Figure 1] 1 is an exemplary graph illustrating a max-cut problem. [Diagram 2] 1 is an exemplary computing device. [Figure 3A] FIG. 1 is a circuit diagram of an exemplary Ising machine. [Figure 3B] 1 is a graph of voltage versus time. [Figure 4A] FIG. 2 is a diagram of an exemplary feedback circuit. [Figure 4B] FIG. 2 is a current / voltage diagram of an example feedback circuit. [Diagram 5] FIG. 2 is a diagram of an example Ising machine node. [Figure 6] 13 is a graph showing the percentage of trials in which the maximum cut was obtained for 20 different graphs. [Figure 7] 1 is a graph showing the average cut value obtained for 20 different graphs and the difference from the maximum cut. [Figure 8A] 13 is a graph of cut value versus time with and without spin modification applied. [Figure 8B]1 is a graph of voltage versus time at multiple nodes at the beginning of a trial. [Figure 8C] 13 is a graph of voltage versus time at multiple nodes at the end of a trial. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] It should be understood that the drawings and description of the present invention have been simplified to illustrate elements relevant to a clear understanding of the present invention, but for clarity, many other elements found in related systems and methods have been omitted. Those skilled in the art may recognize that other elements and / or steps are desirable and / or necessary in implementing the present invention. However, because such elements and steps are well known in the art, and because they do not facilitate a better understanding of the present invention, descriptions of such elements and steps are not provided herein. The disclosure herein covers all such variations and modifications to such elements and methods known to those skilled in the art.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Although any methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention, exemplary methods and materials are described.

[0020] As used herein, each of the following terms has the meaning associated with it in this section.

[0021] The articles "a" and "an" are used herein to refer to one or to more than one (i.e., to at least one) of the grammatical object of the article. By way of example, "an element" means one element or more than one element.

[0022] As used herein when referring to a measurable value, such as an amount, a temporal duration, and the like, "about" is intended to encompass variations of ±20%, ±10%, ±5%, ±1%, and ±0.1% from the stated value, as such variations are appropriate.

[0023] Throughout this disclosure, various aspects of the invention can be expressed in a range format. It should be understood that the description in range format is merely for convenience and brevity and should not be construed as an inflexible limitation on the scope of the invention. Thus, a description of a range should be considered to have all possible subranges specifically disclosed, as well as individual numerical values ​​within that range. For example, a description of a range such as 1-6 should be considered to have specifically disclosed subranges such as 1-3, 1-4, 1-5, 2-4, 2-6, 3-6, and individual numerical values ​​within that range, such as 1, 2, 2.7, 3, 4, 5, 5.3, 6, and all whole and partial increments therebetween. This applies regardless of the breadth of the range.

[0024] In some aspects of the present invention, software executing the instructions provided herein is stored on a non-transitory computer readable medium, which when executed on a processor performs some or all of the steps of the present invention.

[0025] Aspects of the invention relate to algorithms executed in computer software. Although certain embodiments may be described as being written in a particular programming language or executed on a particular operating system or computing platform, it is understood that the systems and methods of the invention are not limited to any particular computing language, platform, or combination thereof. Software executing the algorithms described herein may be written, compiled, or interpreted in any programming language known in the art, including, but not limited to, C, C++, C#, Objective-C, Java, JavaScript, MATLAB, Python, PHP, Perl, Ruby, or VisualBasic. Furthermore, it is understood that elements of the invention may be executed on any acceptable computing platform, including, but not limited to, a server, a cloud instance, a workstation, a thin client, a mobile device, an embedded microcontroller, a television, or any other suitable computing device known in the art.

[0026] Some of the present invention are described as software running on a computing device. Although the software described herein may be disclosed as running on one particular computing device (e.g., a dedicated server or workstation), it is understood in the art that software is portable in nature, and that most software that runs on a dedicated server may also run, for the purposes of the present invention, on any of a wide range of devices, including desktop or mobile devices, laptops, tablets, smartphones, watches, wearable electronics or other wireless digital / cellular telephones, televisions, cloud instances, embedded microcontrollers, thin client devices, or any other suitable computing device known in the art.

[0027] Similarly, portions of the invention are described as communicating over various wireless or wired computer networks. For purposes of the present invention, the terms "network," "networked," and "networking" are understood to encompass wired Ethernet, fiber optic connections, wireless connections including any of the various 802.11 standards, cellular WAN infrastructures such as 3G, 4G / LTE, or 5G networks, Bluetooth, Bluetooth Low Energy (BLE), or Zigbee communications links, or any other manner in which one electronic device can communicate with another electronic device. In some embodiments, elements of the network portion of the invention may be implemented via a Virtual Private Network (VPN).

[0028] 2 and the following discussion are intended to provide a brief, general description of a suitable computing environment in which the invention may be implemented. Although the invention has been described above in the general context of program modules executed in conjunction with application programs running on an operating system on a computer, those skilled in the art will recognize that the invention may also be implemented in combination with other program modules.

[0029] Generally, program modules include routines, programs, components, data structures, and other types of structures that perform particular tasks or implement particular abstract data types. Moreover, those skilled in the art will appreciate that the invention may be practiced with other computer system configurations including handheld devices, microprocessor systems, microprocessor-based or programmable consumer electronics, minicomputers, mainframe computers, and the like. The invention may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote memory storage devices.

[0030] 2 illustrates an exemplary computer architecture of a computer 200 for implementing various embodiments of the present invention. The computer architecture illustrated in FIG. 2 illustrates a conventional personal computer including a central processing unit 250 ("CPU"), a system memory 205 including random access memory 210 ("RAM") and read-only memory ("ROM") 215, and a system bus 235 coupling the system memory 205 to the CPU 250. A basic input / output system containing basic routines that help to transfer information between elements within the computer, such as during start-up, is stored in the ROM 215. The computer 200 further includes an operating system 225, applications / programs 230, and a storage device 220 for storing data.

[0031] Storage device 220 is connected to CPU 250 through a storage controller (not shown) that is connected to bus 235. Storage device 220 and its associated computer-readable media provide non-volatile storage for computer 200. Descriptions of computer-readable media contained herein refer to storage devices such as hard disks or CD-ROM drives, but those skilled in the art should appreciate that computer-readable media may be any available media that can be accessed by computer 200.

[0032] By way of example, and not limitation, computer readable media may include computer storage media. Computer storage media includes volatile and nonvolatile, removable and fixed media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules, or other data. Computer storage media includes, but is not limited to, RAM, ROM, EPROM, EEPROM, flash memory or other solid-state memory technology, CD-ROM, DVD or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and that can be accessed by a computer.

[0033] According to various embodiments of the present invention, computer 200 may operate in a networked environment using logical connections to remote computers through a network 240, such as a TCP / IP network such as the Internet or an intranet. Computer 200 may connect to network 240 through a network interface unit 245 connected to bus 235. It should be appreciated that network interface unit 245 may also be utilized to connect to other types of networks and remote computer systems.

[0034] The computer 200 may also include an input / output controller 255 for receiving and processing input from a number of input / output devices 260, including a keyboard, mouse, touch screen, camera, microphone, controller, joystick, or other type of input device. Similarly, the input / output controller 255 may provide output to a display screen, printer, speaker, or other type of output device. The computer 200 may be connected to the input / output devices 260 via a wired connection, such as, but not limited to, fiber optics, Ethernet, or copper wire, or via wireless means, including, but not limited to, Wi-Fi, Bluetooth, Near Field Communication (NFC), infrared, or other suitable wired or wireless connection.

[0035] As briefly mentioned above, a number of program modules and data files, including an operating system 225 suitable for controlling the operation of a networked computer, may be stored in the storage device 220 and / or RAM 210 of the computer 200. The storage device 220 and RAM 210 may also store one or more applications / programs 230. In particular, the storage device 220 and RAM 210 may store applications / programs 230 for providing a variety of functions to a user. For example, the applications / programs 230 may comprise many types of programs, such as word processing applications, spreadsheet applications, desktop publishing applications, database applications, gaming applications, Internet browsing applications, email applications, messaging applications, and the like. According to one embodiment of the present invention, the applications / programs 230 include a multi-function software application that provides word processing functionality, slide presentation functionality, spreadsheet functionality, database functionality, and the like.

[0036] In some embodiments, computer 200 may include various sensors 265 for monitoring the environment surrounding and internal to computer 200. These sensors 265 may include a global positioning system (GPS) sensor, a photosensitive sensor, a gyroscope, a magnetometer, a thermometer, a proximity sensor, an accelerometer, a microphone, a biometric sensor, a barometer, a humidity sensor, a radiation sensor, or any other suitable sensor.

[0037] Disclosed herein is a proposed Ising machine. The disclosed Ising machine may be represented in some embodiments by a set of electronic components, including, for example, a plurality of resistors and capacitors arranged in a network. In some embodiments, the device of the present disclosure may further comprise a local feedback circuit, including, for example, one or more p-channel field effect transistors (FETs) and / or one or more n-channel FETs. In some embodiments, the feedback circuit may include one or more logic gates, such as one or more inverters. In some embodiments, the feedback circuit may include two back-to-back inverters. In some embodiments, the logic gates, including but not limited to inverters, may be implemented in CMOS technology. In some embodiments, the device of the present disclosure may further comprise one or more switches or relays configured to temporarily connect one side of a capacitor of the device to a positive voltage source, a negative voltage source, or a ground reference. In some embodiments, one or more of the switches or relays may be controlled by a controller, such as a microcontroller or a computing device. In some embodiments, the system disclosed herein may comprise one or more sensors, such as a voltage sensor or a current sensor configured to measure a voltage between two points in a circuit network, or a current sensor configured to measure a current along one path in the network. In some embodiments, some or all of the functionality or individual hardware components of the disclosed devices may be configured in a software simulation running on a computing device.

[0038] From Equation 1, two quantities are required to describe the Ising model: the coupling strength between nodes (J ij ) and the node state (σ i ,σ j) is needed. We also need some mechanism to store state and to control the strength of the connections between the nodes to allow them to interact. These functions can be easily achieved using capacitors and adjustable resistors, respectively.

[0039] Considering that we use a capacitor to represent the node, the first of the plates is defined as positive, where a positive voltage is converted to a spin of +1, while a negative voltage value is converted to a spin of -1. The coupling between the two nodes is achieved by a pair of resistors with equal resistance. The lower the resistance, the stronger the coupling. That is, R ij ∝1 / J ij Positive coupling is achieved by connecting the same polarity of two capacitors, while negative coupling is achieved by connecting the opposite polarity, which are referred to herein as parallel and anti-parallel coupling, respectively.

[0040] The above insight leads to the Ising model being transformed into a resistor-coupled capacitor network. As a result, the max-cut problem becomes a circuit problem. Referring to Figure 1 as an example, first, the relationship between weight and coupling resistance is R ij =1 / W ij where R is the scaling resistor and R ij The sign of indicates the polarity of the coupling. Five capacitors are then used to represent the five nodes, and their voltages are initialized to random signs corresponding to the random spin configurations. The overall circuit is shown in FIG. 3A. The red line 301 indicates the positive terminal, and the blue line 302 is the negative terminal. Capacitors with the same label (e.g., C2, 303) represent the same node. Only coupling resistors cross the lines; that is, lines 302 and 304 overlap in the figure as an example, but each overlap is not necessarily an electrically connected junction. Where a resistor overlaps a line (e.g., where resistor 305 overlaps with line 304) or where a capacitor overlaps a line (e.g., where capacitor 303 overlaps with line 304) represents an electrical junction.

[0041] When the circuit reaches equilibrium, the voltage on each capacitor will branch to a positive or negative level, as shown in the graph in Figure 3B, which is a graph showing the transient steady state of the system. The positive and negative levels are equivalent to two groups of spins. The circuit is then in equilibrium with V + ={2} and V - = {1,3,4,5}. Applying equation 4 gives a cut value of 0.34, which is indeed the maximum cut.

[0042] Although the system described above works as expected, it also has some practical problems. First, due to capacitor leakage, all nodes will eventually drop to 0V. Also, the voltage level at equilibrium is very low, which can make it difficult to read reliably. Another problem is that the energy landscape is usually not convex, so the machine is not guaranteed to converge to a global minimum. In other words, there are many local minima and the machine can easily be drawn into energy basins depending on the initial conditions. Solutions to overcome these problems are described below.

[0043] ZIV Diode A local feedback circuit can be introduced to stabilize the steady-state voltage at a desired level. This circuit must be able to charge the capacitor according to its polarity when the voltage is within the range of the supply voltage, and discharge the capacitor when the voltage is about to exceed the supply voltage. In addition, the circuit must provide low current at low voltage so as not to overwhelm the signal coupled from other nodes. The circuit shown in FIG. 4A can meet these two conditions. FIG. 4A is a schematic diagram of an exemplary ZIV diode. FIG. 4B is a graph of the IV characteristics of the circuit shown in FIG. 4A.

[0044] V p and V m is the differential,

number

[0045] Spin-locked annealing As the complexity of the graph increases and the number of local minima in the Ising formulation increases (e.g., a 32-node graph becomes 2 20, a machine following gradient descent is more likely to reach a local energy minimum. One possible way to escape a local minimum is to flip the polarity of some spins. However, this scheme requires several monitoring blocks to detect and change the polarity of each node, which significantly increases the hardware complexity. In the present disclosure, one or more nodes are instead randomly selected and the node or nodes are clipped (i.e., shorted) to one of the supply rails for a short period of time. In some embodiments, the node or nodes may be clipped to the supply rails for a period of 1 ps to 100 ps, ​​or 1 ps to 50 ps, ​​or 1 ps to 20 ps, ​​or 5 ps to 20 ps, ​​or less than 10 ps, ​​or about 10 ps, ​​or 50 μs to 10 ms, or 50 μs to 500 μs, or 100 μs to 500 μs, or 200 μs to 500 μs, or 1 ms to 500 μs. The selection of the node or nodes may be random or pseudo-random, or in some embodiments may be sequential or predetermined, for example, based on the number and placement of nodes in the graph from which the Ising machine was generated.

[0046] Due to this, the spin of the node will either remain in its current state or switch to the opposite state within a certain time interval. The occurrence of spin locking occurs more frequently at first but gradually decreases. As the energy of the system approaches a global minimum, a single spin lock makes it less likely to escape. This scheme is referred to herein as "spin locking annealing" and two pairs of switches are used to implement it. Note that these switches can also be used to initialize the node voltages before the state of the machine is evolved in search of the energy minimum. A high-level schematic of the BRIM node along with the ZIV diode and its coupling to other nodes is given in Figure 5.

[0047] Experimental Example The present invention will be described in more detail with reference to the following experimental examples.These examples are provided solely for illustrative purposes and are not intended to be limiting unless otherwise specified.Therefore, the present invention should not be construed as being limited in any way to the following examples, but rather as embracing any and all variations that become evident as a result of the teachings provided herein.

[0048] Without further elaboration, it is believed that one of ordinary skill in the art can, using the preceding description and the following illustrations, make and use the systems and methods of the present invention. Accordingly, the following working examples illustrate exemplary embodiments of the present invention, and are not to be construed as limiting in any way the remainder of the disclosure.

[0049] To evaluate the performance of the proposed circuit, a 32-node BRIM was designed on a 45nm Generic Process Design Kit (GPDK) and simulated in Virtuoso Analog Design Environment (ADE), both from Cadence. The test bench contains 20 binary weighted graphs, i.e., W takes values ​​of ±1. First, spin-locked annealing was disabled and each graph was simulated 20 times with different initial conditions. Then, spin-locked annealing was enabled. For a fair comparison, the same node voltage initial conditions were used, but different spin-locking sequences were applied in each run.

[0050] The simulation results are summarized in Figure 6. The graph in Figure 6 shows the percentage of runs that reached the global minimum. For each graph G1-G20, a total of 40 runs were performed, 20 of which were with spin-locked annealing and 20 without spin-locked annealing. It can be seen that the machine is indeed able to find the optimal solution (i.e., the global energy minimum). However, without spin-locked annealing, the percentage of runs that reach the global minimum is, on average, only 29.25%, while with spin-locked annealing, the machine is able to find the optimal solution in 79.75% of runs. There were six graphs that were always able to reach the best solution (i.e., the maximum cut value). For the solutions that were not the best solution, the average cut and the distance to the maximum cut were calculated for each graph and the results are plotted in Figure 7. The bars in the bar graph show the average cut value and the error bars represent the distance between the average cut value and the maximum cut. Note that six cases reaching 100% maximum cut (i.e., G2, G4, G5, G9, G17, and G19 with spin-locked annealing) from Figure 6 have been omitted. With spin-locked annealing, the quality of the solutions improved significantly, with an average cut error of 1.75 compared to 5.89 without spin-locked annealing.

[0051] It is also worthwhile to examine the transient behavior of the system. Graph G5 was selected from the test bench mentioned above, and its evolution of cut value was recorded over time, both with and without spin-locked annealing. Figure 8A shows the evolution of the cut value of graph G5 over time, both without spin-locked annealing (green curve) and with spin-locked annealing (blue curve). The red line represents the maximum cut value, and its position on the y-axis has been intentionally shifted upwards, thereby not hiding the flat part of the blue curve where the maximum cut value is reached. As can be seen in Figure 8A, when spin-locked annealing is disabled (green curve), the machine progressed rapidly towards lower energy levels. This equates to an increase in the cut value, but unfortunately, it got stuck in a local minimum and then settled there.

[0052] Figure 8B shows the transient voltages of each node at the beginning of the simulation (normalized time from 0 to 0.2), where the interactions between the nodes can be clearly seen. On the other hand, when spin-locked annealing is enabled (blue curve), the machine becomes busy exploring numerous regions of attraction in the energy landscape. Once it reaches near the global minimum, further spin-locking of one node increases the energy level only temporarily, but it is quickly pulled back to the global minimum by the remaining nodes. This behavior can be seen in Figure 8C, which shows the transient voltages near the end of the annealing time.

[0053] conclusion The disclosed bistable resistively coupled Ising machine indeed demonstrates its ability to solve challenging computational problems such as the maximum cut problem. By using only simple building blocks, the disclosed topology is more scalable and cheaper compared to other desk-sized (or even room-sized) machines. With the help of spin-locked annealing, the performance of the disclosed machine is further improved: not only does the probability of reaching the maximum cut increase by 50.5%, but the quality of the solution also improves significantly even when the global minimum is not reached, with an average cut error of 1.75.

[0054] The disclosures of each and every patent, patent application, and publication cited herein are incorporated herein by reference in their entirety. Although the present invention has been disclosed with reference to specific embodiments, it will be apparent to others skilled in the art that other embodiments and variations of the present invention may be devised without departing from the true spirit and scope of the present invention. It is intended that the appended claims be construed to include all such embodiments and equivalent variations.

[0055] References The following publications are incorporated herein by reference in their entireties:

[0056] C. Johnson, DHAllen, J. Brown, S. Vanderwiel, R. Hoover, H. Achilles, C. Cher, GMay, H. Franke, J. Xenedis, and C. Basso IEEE International Solid-State Circuits Conference-(ISSCC),2010,pp.104-105.

[0057] B.Erbagci,NECAkkaya,C.Teegarden、and K.Mai,“A 275 Gbps AES Encryption Accelerator Using ROM-Based S-Boxes in 65nm,”in 2015 IEEE Custom Integrated Circuits Conference(CICC),2015,pp.1-4.

[0058] Inagaki T., Haribara Y., Igarashi K., Sonobe T., Tamate S., Honjo T., Marandi A., PLMcMahon T., Umeki T., Enb utsu, O. Tadanaga, H. Takenouchi, K. Aihara, K.-i. Kawarabayashi, K. Inoue, S. Utsunomiya, and H. Takesue, “A coherent ice machine for 2000-node optimization problems,”Science,vol.354,no.6312,pp.603-606,2016.

[0059] T.Wang and J.Roychowdhury,“IOM:Oscillator-Based Icing Machines for Solving Combinatorial Optimization Problems,”2019.

[0060] S.Kirkpatrick,CDGelatt,and MPVecchi,“Optimization by simulated annealing,”science,vol.220,no.4598,pp.671-680,1983.

[0061] E.Guglielmi,F.Toso,F.Zanetto,G.Sciortino,A.Mesri,M.Sampietro、and G.Ferrari,“Design of high-value tunable pseudo-resistors,”IEEE Journal of Solid-State Circuits,vol.55,no.8,pp.2094-2105,2020.

Claims

1. An Ising machine system, a plurality of capacitors, each capacitor having a positive terminal and a negative terminal; a plurality of resistors having first and second terminals, each capacitor being electrically connected via a respective positive and negative terminal thereof to the first or second terminal of at least one resistor of the plurality of resistors; a power supply having positive and negative rails; a switching network configured to initialize each of the plurality of capacitors with a positive or negative voltage from the power supply and then disconnect all of the plurality of capacitors from the power supply to float the voltages of the capacitors; at least one voltage measurement device configured to measure a voltage between the positive terminal and the negative terminal of a capacitor of the plurality of capacitors; The Ising machine system comprising:

2. 2. The Ising machine system of claim 1, further comprising a feedback circuit configured to charge at least one capacitor of the plurality of capacitors when a voltage level of the capacitor is less than a voltage of the power supply and to discharge the capacitor when the voltage level of the capacitor is about to exceed the voltage of the power supply.

3. The Ising machine system of claim 2 , wherein the feedback circuit comprises two back-to-back inverters.

4. The Ising machine of claim 3 , wherein the two back-to-back inverters are implemented in CMOS technology.

5. 5. The Ising machine system of claim 1, wherein the switching network comprises a set of switches electrically connected to a positive or negative terminal of each capacitor of the plurality of capacitors and configured to form a momentary electrical contact between the positive or negative terminal of each capacitor and the positive or negative rail of the power supply.

6. the switching network randomly selecting a positive or negative terminal of one of the capacitors of the plurality of capacitors after disconnecting all of the plurality of capacitors from the power source; 6. The Ising machine system of claim 1, further configured to momentarily connect and then disconnect the randomly selected terminal to the positive rail or the negative rail of the power supply.

7. An Ising machine system, A non-transitory computer-readable medium having instructions stored thereon, the instructions, when executed by a processor, 2. Emulating a circuit network, the emulated circuit network comprising: a plurality of capacitors, each having a positive terminal and a negative terminal; a plurality of resistors having first and second terminals, each capacitor being electrically connected via a respective positive and negative terminal thereof to the first or second terminal of at least one resistor of the plurality of resistors; a power supply having positive and negative rails; the emulating comprises: initializing each of the plurality of capacitors with a positive or negative voltage from the rails of the power supply; disconnecting all of the plurality of capacitors from the power supply to float the voltages of the capacitors; measuring and recording a voltage between the positive and negative terminals of at least one of the plurality of capacitors over a period of time; The non-transitory computer-readable medium performing the steps including: The Ising machine system comprising:

8. Each of the steps is randomly selecting a positive or negative terminal of one of the capacitors of the plurality of capacitors after disconnecting all of the plurality of capacitors from the power source; 8. The Ising machine of claim 7, further comprising: momentarily connecting and then disconnecting the randomly selected terminal to the positive rail or the negative rail of the power supply.

9. 1. A method of running an Ising machine, comprising: providing a plurality of resistors and capacitors and a power source; connecting each of the plurality of capacitors to at least one other of the plurality of capacitors through two of the plurality of resistors; initializing each of the plurality of capacitors with a positive or negative voltage from the power supply; disconnecting the power supply from the plurality of capacitors to float the voltages of the capacitors; measuring and recording a voltage between a positive terminal and a negative terminal of at least one of the plurality of capacitors over a period of time; The method comprising:

10. 10. The method of claim 9, further comprising measuring and recording the voltage between the positive and negative terminals of all of the plurality of capacitors over the period of time.

11. 11. The method of claim 10, further comprising: designating as a first set the capacitors having a positive voltage after the period of time and designating as a second set the capacitors having a negative voltage after the period of time, the first set and the second set corresponding to nodes in a max-cut analysis.