Self-contained relaxation-based dynamical Ising machines

Mikhail Erementchouk,Aditya Shukla,Pinaki Mazumder

2024-12-05

Abstract:Dynamical Ising machines are based on continuous dynamical systems evolving from a generic initial state to a state strongly related to the ground state of the classical Ising model on a graph. Reaching the ground state is equivalent to finding the maximum (weighted) cut of the graph, which presents the Ising machines as an alternative way to solving and investigating NP-complete problems. Among the dynamical models, relaxation-based models are distinguished by their relations with guarantees of performance achieved in time scaling polynomially with the problem size. However, the terminal states of such machines are essentially non-binary, necessitating special post-processing relying on disparate computing. We show that an Ising machine implementing a special continuous dynamical system (called the V${}_2$ model) solves the rounding problem dynamically. We prove that the V${}_2$ model, starting from an arbitrary non-binary state, terminates in a state that trivially rounds to a binary state with the cut at least as big as obtained by optimal rounding of the initial state. Besides showing that relaxation-based dynamical Ising machines can be made self-contained, this result presents a non-Boolean realization of solving a non-trivial information processing task on Ising machines. Moreover, we prove that if the initial state of the V${}_2$-machine is a random limited amplitude perturbation of a binary state, the machine progresses to a state with at least as high cut as that of the initial binary state. Since the probability of improving the cut is finite, this shows that the V${}_2$-machine with random agitations converges to a maximum cut state almost surely.

Emerging Technologies,Dynamical Systems

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the optimal rounding problem of dynamic Ising machines when solving the maximum - cut problem. Specifically, the author proposes an Ising machine model (called the V² model) based on a special continuous dynamical system. This model can start from any non - binary state and finally reach a state that can be simply rounded to a binary state, and the cut produced by this binary state is not less than the cut obtained by optimally rounding the initial state. This result not only proves that the relaxation - based dynamic Ising machines can be self - contained, but also shows that the dynamic system can directly perform complex signal - processing tasks.

### Background and Problem Description of the Paper

**Background**:
- **Ising Model**: The classical Ising model is a spin system defined on a graph, and its ground state (the spin distribution with the lowest energy) can be used to solve quadratic unconstrained binary optimization problems, especially the maximum - cut problem.
- **Maximum - Cut Problem**: Given a graph \(G=(V, E)\), the maximum - cut problem is to find a partition \(V = V_+ \cup V_-\) such that the total weight of the edges connecting \(V_+\) and \(V_-\) is maximized. This is an NP - complete problem.
- **Dynamic Ising Machines**: These machines are models based on continuous dynamical systems and can evolve from a general initial state to a state closely related to the ground state of the Ising model. Existing dynamic Ising machines usually require external processing to achieve optimal rounding, which makes them not self - contained.

**Problem**:
- **Optimal Rounding Problem**: How to design a dynamic Ising machine so that it can evolve from any non - binary state to a state that can be simply rounded to a binary state without relying on external processing, and the cut produced by this binary state is not less than the cut obtained by optimally rounding the initial state.

### Main Contributions

1. **Proposal of the V² Model**:
- A new dynamic Ising machine model (the V² model) is proposed, which can start from any non - binary state and finally reach a state that can be simply rounded to a binary state.
- It is proved that the terminal state of the V² model can be obtained as a binary state by simple rounding, and the cut produced by this binary state is not less than the cut obtained by optimally rounding the initial state.

2. **Self - Contained Dynamic Ising Machines**:
- It is proved that the dynamic Ising machines based on the V² model can be self - contained, that is, optimal rounding can be completed without external processing.
- This result not only solves the optimal rounding problem, but also shows that the dynamic system can directly perform complex signal - processing tasks.

3. **Theoretical and Experimental Verification**:
- The effectiveness and performance of the V² model are verified through theoretical analysis and numerical simulation.

### Conclusion

By proposing the V² model, this paper solves the optimal rounding problem of dynamic Ising machines when solving the maximum - cut problem, realizes self - contained dynamic Ising machines, and provides a new and effective method for solving NP - complete problems.