High-performance combinatorial optimization based on classical mechanics

Goto, Hayato; Endo, Kazuhiko; Suzuki, Mariko; Sakai, Yoshisato; Kanao, Taro; Hamakawa, Yohei; Hidaka, Ryo; Yamasaki, Masaya; Tatsumura, Kosuke

doi:10.1126/sciadv.abe7953

Cited by 112 publications

(153 citation statements)

References 61 publications

Supporting

Mentioning

150

Contrasting

Order By: Relevance

“…More recently, the same team proposed a modified SB algorithm (47), with which a fully connected 100,000-node graph with 10-bit weight resolution was solved. They showed that it can improve the solution accuracy and calculation time compared with the previous SB algorithm (46).…”

Section: Discussionmentioning

confidence: 99%

100,000-spin coherent Ising machine

et al. 2021

View full text Add to dashboard Cite

Computers based on physical systems are increasingly anticipated to overcome the impending limitations on digital computer performance. One such computer is a coherent Ising machine (CIM) for solving combinatorial optimization problems. Here, we report a CIM with 100,512 degenerate optical parametric oscillator pulses working as the Ising spins. We show that the CIM delivers fine solutions to maximum cut problems of 100,000-node graphs drastically faster than standard simulated annealing. Moreover, the CIM, when operated near the phase transition point, provides some extremely good solutions and a very broad distribution. This characteristic will be useful for applications that require fast random sampling such as machine learning.

show abstract

Section: Discussionmentioning

confidence: 99%

100,000-spin coherent Ising machine

et al. 2021

View full text Add to dashboard Cite

show abstract

“…For large-scale problem sizes, D-Wave's new Leap Hybrid service [47], [48] is capable of processing up to 20,000 QUBO variables with fully connected graphs, using an approach of problem partitioning. Digital annealing technologies [49]- [52] also pave the way towards time sensitive optimization scenarios with high-parallelism and highperformance dedicated architectures.…”

Section: Discussionmentioning

confidence: 99%

Evaluation of Quantum Annealer Performance via the Massive MIMO Problem

Tabi¹,

Marosits

Kallus

et al. 2021

IEEE Access

View full text Add to dashboard Cite

Quantum annealing offers an appealing route to handle large-scale optimization problems. Existing Quantum Annealing processing units are readily available via cloud platform access for solving Quadratic Unconstrained Binary Optimization (QUBO) problems. In particular, the novel D-Wave Advantage device has been recently released. Its performance is expected to improve upon the previous stateof-the-art D-Wave 2000Q annealer, due to higher number of qubits and the Pegasus topology. Here, we present a comparative study via an ensemble of Maximum Likelihood (ML) Channel Decoder problems for MIMO scenarios in Centralized Radio Access Network (C-RAN) architectures. The main challenge for exact optimization of ML decoders with ever-increasing demand for higher data rates is the exponential increase of the solution space with problem sizes. Since current 5G solutions mainly use approximate methodologies, Kim et al. leveraged Quantum Annealing for large MIMO problems with Phase Shift Keying and Quadrature Amplitude Modulation scenarios. Here, we extend their work and analyze experiments for more complex modulations and larger MIMO antenna array sizes. By implementing the extended QUBO formulae on the novel annealer architecture, we uncover the limits of state-of-the-art quantum optimization for the massive MIMO ML decoder. We report on the improvements and discuss the uncovered limiting factors learned from the 64-QAM extension. We include the enhanced evaluation of raw annealer sampling via implementation of post-processing methods in the comparative analysis between D-Wave 2000Q and D-Wave Advantage system.

show abstract

“…This trend is also observed in the SBM. [27,29] Whether or not CIM-SFC utilizes chaotic dynamics, however, is beyond the scope of this paper; to answer this question, we would need to further analyze how the parameters of CIM-SFC affect its dynamics and gain a deeper understanding of how CIM-SFC finds ground states.…”

Section: Numerical Simulation Of Cim-cac Cim-cfc and Cim-sfcmentioning

confidence: 99%

“…[25] Similarly, a network of dissipation-less quantum oscillators with adiabatic Hamiltonian modulation that can be described using a set of N deterministic equations can also be used to implement a heuristic algorithm on a modern digital platform. [27][28][29] Such heuristic algorithms are called simulated bifurcation machines (SBMs), [27,29] and a variant SBM will be studied in Section 6. Although the original SBM was inspired by dissipation-less adiabatic quantum computation, the version of SBM discussed in this paper (dSBM) is not a true unitary system, as dissipation is artificially added using inelastic walls to improve the performance of the algorithm.…”

mentioning

confidence: 99%

Coherent Ising Machines with Optical Error Correction Circuits

Reifenstein¹,

Kako²,

Khoyratee³

et al. 2021

Adv Quantum Tech

View full text Add to dashboard Cite

A coherent Ising machine (CIM) comprising a network of open-dissipative quantum oscillators with optical error correction circuits is proposed. In the proposed network, the squeezed/anti-squeezed vacuum states of the constituent optical parametric oscillators establish quantum correlations below the threshold through optical mutual coupling and collective symmetry breaking is induced above the threshold as a decision-making process. This initial search process is followed by a chaotic solution search step facilitated by optical error correction feedback. The particular algorithm used by the proposed network is derived from the truncated Wigner stochastic differential equation. As an optical hardware technology, the proposed CIM has several unique features, including programmable all-to-all Ising coupling in the optical domain, directional coupling (J ij ≠ J ji )-induced chaotic behavior, and the ability to operate at low power at room temperature. The proposed CIM is evaluated in terms of its performance and how this scales at different problem sizes. It is shown that the various CIMs discussed in this paper are effective at solving many problem types, although the optimal algorithm depends on the problem case. It is further shown that the proposed optical implementations can be achieved with low energy consumption on thin-film LiNbO 3 platforms.

show abstract

High-performance combinatorial optimization based on classical mechanics

Cited by 112 publications

References 61 publications

100,000-spin coherent Ising machine

100,000-spin coherent Ising machine

Evaluation of Quantum Annealer Performance via the Massive MIMO Problem

Coherent Ising Machines with Optical Error Correction Circuits

Contact Info

Product

Resources

About