pith. sign in

arxiv: 1906.08445 · v1 · pith:UDVE3YJ4new · submitted 2019-06-20 · 🪐 quant-ph

Associative memory on qutrits by means of quantum annealing

V.E. Zobov , I.S. Pichkovskiy This is my paper

Pith reviewed 2026-05-25 20:01 UTC · model grok-4.3

classification 🪐 quant-ph
keywords associative memoryqutritsquantum annealingadiabatic evolutionquantum memoryspins S=1superposition states
0
0 comments X

The pith

Associative memory on qutrits via quantum annealing shows higher capacity than on qubits.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper shows that associative memory can be built on three-level quantum systems (qutrits) by recording patterns as a superposition and recalling them through slow changes to the system's Hamiltonian. An auxiliary term is added to the Hamiltonian to make all stored states equally likely before it is switched off. Simulations for two and three qutrits confirm that the number of patterns that can be stored and retrieved reliably grows when qubits are replaced by qutrits. A reader would care because this offers a concrete route to larger quantum memories without needing more elements. The approach stays within the adiabatic quantum annealing framework already used for optimization tasks.

Core claim

Associative memory is realized on qutrits represented by spins with S = 1. Patterns are recorded into a superposition of quantum states and recalled by adiabatic variation of the Hamiltonian. An auxiliary Hamiltonian equalizes the probabilities of the states in the superposition and is turned off at the end of the evolution. Simulations performed on two and three qutrits demonstrate an increase in memory capacity after replacing qubits with qutrits.

What carries the argument

The auxiliary Hamiltonian that equalizes probabilities of states in the superposition and is turned off at the end without distorting recall.

If this is right

  • Memory capacity increases when qubits are replaced by qutrits.
  • Patterns stored in superposition can be recalled by adiabatic Hamiltonian evolution.
  • The auxiliary term can be removed at the end of the schedule without creating extra attractors.
  • The method works for at least two and three qutrits in numerical tests.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same auxiliary-term technique might allow capacity gains in larger qutrit networks if the equalization remains effective.
  • This construction could be combined with other adiabatic protocols that already use multi-level systems for optimization.
  • Hardware experiments on superconducting qutrits would directly test whether the simulated capacity gain survives decoherence.

Load-bearing premise

The auxiliary Hamiltonian equalizes probabilities of states in the superposition without distorting the adiabatic recall dynamics or introducing spurious attractors when turned off at the end of the evolution.

What would settle it

A simulation on three qutrits in which the number of reliably recalled patterns does not exceed the qubit case or in which disabling the auxiliary term produces recall errors.

Figures

Figures reproduced from arXiv: 1906.08445 by I.S. Pichkovskiy, V.E. Zobov.

Figure 1
Figure 1. Figure 1: Fig.1. Associative memory on two [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Fig.2. The probabilities of [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Fig.3. Dependencies of the probabilities of [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Fig.4. Depending on [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
read the original abstract

We study the functioning of associative memory on three-level quantum elements, qutrites represented by spins with S = 1. The recording of patterns into the superposition of quantum states and their recall are carried out by adiabatic variation of the Hamiltonian with time. To equalize the probabilities of finding the system in different states of superposition, an auxiliary Hamiltonian is proposed, which is turned off at the end of evolution. Simulations were performed on two and three qutrits and an increase in the memory capacity after replacing qubits with qutrits is shown.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript studies associative memory implemented on qutrits (S=1 spins) via quantum annealing. Patterns are encoded in a quantum superposition, recalled by adiabatic Hamiltonian evolution, and an auxiliary Hamiltonian is introduced to equalize superposition probabilities before being ramped to zero at the end of the schedule. Simulations for systems of two and three qutrits are reported to show an increase in memory capacity relative to the corresponding qubit case.

Significance. If the capacity increase is robustly demonstrated and free of artifacts from the auxiliary term, the result would indicate that higher-dimensional local Hilbert spaces can improve the performance of quantum associative memory models, extending the scope of quantum annealing beyond qubits. The auxiliary-Hamiltonian construction itself is a concrete technical proposal that could be reusable in other annealing-based memory or optimization tasks.

major comments (2)
  1. [Abstract] Abstract and simulation description: the central claim of increased capacity rests on numerical results for N=2 and N=3, yet no Hamiltonian parameters, annealing schedule details, number of patterns tested, convergence diagnostics, or error bars are supplied. Without these, the reported capacity gain cannot be reproduced or assessed for statistical significance.
  2. [Auxiliary Hamiltonian] Auxiliary Hamiltonian section: the assumption that the auxiliary term equalizes probabilities without distorting the instantaneous eigenstates or introducing spurious attractors once it is turned off is load-bearing for the capacity claim. No spectral gap analysis, instantaneous eigenstate overlaps, or explicit checks against spurious fixed points are provided, even though for small N the landscape is dense and any hidden distortion would directly affect the reported numbers.
minor comments (1)
  1. Notation for the qutrit states and the explicit form of the pattern Hamiltonian should be stated once in a dedicated subsection to improve readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading of the manuscript and the constructive comments. We address each major comment below and indicate the revisions that will be made to improve reproducibility and validation of the results.

read point-by-point responses

  1. Referee: [Abstract] Abstract and simulation description: the central claim of increased capacity rests on numerical results for N=2 and N=3, yet no Hamiltonian parameters, annealing schedule details, number of patterns tested, convergence diagnostics, or error bars are supplied. Without these, the reported capacity gain cannot be reproduced or assessed for statistical significance.

    Authors: We agree that the simulation details require expansion for full reproducibility. The revised manuscript will include an expanded methods section specifying the Hamiltonian parameters (including the form of the pattern-encoding and auxiliary terms), the precise annealing schedule (linear or otherwise), the exact number of patterns tested for each N, convergence diagnostics used in the adiabatic evolution, and error bars derived from multiple independent runs where applicable. These additions will allow direct assessment of the reported capacity increase. revision: yes

  2. Referee: [Auxiliary Hamiltonian] Auxiliary Hamiltonian section: the assumption that the auxiliary term equalizes probabilities without distorting the instantaneous eigenstates or introducing spurious attractors once it is turned off is load-bearing for the capacity claim. No spectral gap analysis, instantaneous eigenstate overlaps, or explicit checks against spurious fixed points are provided, even though for small N the landscape is dense and any hidden distortion would directly affect the reported numbers.

    Authors: The referee is correct that explicit validation of the auxiliary Hamiltonian is essential. Although the small system sizes permit exact diagonalization, these checks were omitted from the original submission. In the revision we will add (i) the spectral gap throughout the schedule with and without the auxiliary term, (ii) overlaps between the instantaneous eigenstates of the full and reduced Hamiltonians, and (iii) explicit enumeration of fixed points of the final Hamiltonian to rule out spurious attractors. These analyses will confirm that the observed capacity gain is not an artifact of the auxiliary construction. revision: yes

Circularity Check

0 steps flagged

No circularity: capacity increase shown via independent simulations of 2- and 3-qutrit systems

full rationale

The paper's derivation consists of proposing an auxiliary Hamiltonian to equalize superposition probabilities during adiabatic evolution, then reporting numerical simulations on small systems (N=2,3) that exhibit increased pattern capacity relative to the qubit case. No load-bearing step reduces by construction to a fitted parameter, self-definition, or self-citation chain; the reported capacity numbers are outputs of the simulations rather than inputs renamed as predictions. The auxiliary term is introduced explicitly and turned off at the end, with the capacity claim resting on the simulation outcomes themselves rather than on any prior result by the same authors that would render the conclusion tautological.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no explicit free parameters, axioms, or invented entities are stated. The auxiliary Hamiltonian is introduced as a proposal without independent justification beyond its intended effect.

pith-pipeline@v0.9.0 · 5613 in / 1079 out tokens · 21559 ms · 2026-05-25T20:01:22.232996+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

23 extracted references · 23 canonical work pages · 1 internal anchor

  1. [1]

    Each of the projections im can take one of three values: 1, 0, -1

    e-mail: rsa@iph.krasn.ru 2 use a basis nmmm ,...,, 21 of the eigenfunctions of the operators z iS of the projections of the spins on the Z axis. Each of the projections im can take one of three values: 1, 0, -1. Let it is necessary to keep in memory p elements represented by p quantum vectors   n21 ,...,, mmm , where p...,,2,1 . Choose the project...

    work page

  2. [2]

    Biamonte, P

    J. Biamonte, P. Wittek, N. Pancotti, P. Rebentrost, N. Wiebe, and S. Lloyd, Nature 549, 195 (2017)

    work page 2017

  3. [3]

    Dunjko and H

    V. Dunjko and H. J. Briegel, Reports on Progress in Physics 81, 074001 (2018)

    work page 2018

  4. [4]

    (Prentice Hall Europe, Copyright, 1999)

    Robert Callan, The essence of neural networks. (Prentice Hall Europe, Copyright, 1999)

    work page 1999

  5. [5]

    Neigovzen, J.L

    R. Neigovzen, J.L. Neves, R. Sollacher, and S.J. Glaser, Phys. Rev. A 79, 042321 (2009)

    work page 2009

  6. [6]

    Seddiqi and T.S

    H. Seddiqi and T.S. Humble, Front. Phys. 2, 79 (2014)

    work page 2014

  7. [7]

    Santra, O

    S. Santra, O. Shehab, and R. Balu, Phys. Rev. A 96, 062330 (2017)

    work page 2017

  8. [8]

    Dlaska, L.M

    C. Dlaska, L.M. Sieberer, and W. Lechner, Phys. Rev. A 99, 032342 (2019)

    work page 2019

  9. [9]

    L. M. Sieberer and W. Lechner, Phys. Rev. A 97, 052329 (2018)

    work page 2018

  10. [10]

    R. Das, A. Mitra, V. Kumar, and A. Kumar, Int. J. Quantum Inf. 1, 387 (2003)

    work page 2003

  11. [11]

    Klimov, R

    A.B. Klimov, R. Guzmán, J.C. Retamal, and C. Saavedra, Phys. Rev. A 67, 062313 (2003)

    work page 2003

  12. [12]

    Tamir, Phys

    B. Tamir, Phys. Rev. A 77, 022326 (2007)

    work page 2007

  13. [13]

    Zobov and D.I

    V.E. Zobov and D.I. Pekhterev, JETP Letters 89, 260 (2009)

    work page 2009

  14. [14]

    Zobov and V.P

    V.E. Zobov and V.P. Shauro, JETP 113, 181 (2011)

    work page 2011

  15. [15]

    Choi, N.Y

    S. Choi, N.Y. Yao, M.D. Lukin, Phys. Rev. Lett. 119, 183603 (2017)

    work page 2017

  16. [16]

    O'Keeffe, L

    M.F. O'Keeffe, L. Horesh, D.A. Braje, and I.L. Chuang, New J. Phys. 21, 023015 (2019)

    work page 2019

  17. [17]

    Steffen, W

    M. Steffen, W. van Dam, T. Hogg, G. Breyta, and I. Chuang , Phys. Rev. Lett. 90, 067903 (2003)

    work page 2003

  18. [18]

    Mandra, Z

    S. Mandra, Z. Zhu, and H.G. Katzgraber, Phys. Rev. Lett. 118, 070502 (2017)

    work page 2017

  19. [19]

    Lifshitz, Quantum Mechanics: Non-relativistic Theory, 3rd ed

    L.D Landau and E.M. Lifshitz, Quantum Mechanics: Non-relativistic Theory, 3rd ed. (Per- gamon Press, Oxford, 1977)

    work page 1977

  20. [20]

    Different Strategies for Optimization Using the Quantum Adiabatic Algorithm

    E. Crosson, E. Farhi, C. Yen-Yu Lin, Han-Hsuan Lin, and P. Shor, arXiv:1401.7320

    work page internal anchor Pith review Pith/arXiv arXiv

  21. [21]

    Zobov and A.S

    V.E. Zobov and A.S. Ermilov, JETP 114, 923 (2012)

    work page 2012

  22. [22]

    Zobov and I.S

    V.E. Zobov and I.S. Pichkovskiy, Proc. SPIE 11022, International Conference on Micro- and Nano-Electronics 2018, 110222V (15 March 2019); doi: 10.1117/12.2521253

    work page doi:10.1117/12.2521253 2018

  23. [23]

    ASSOCIATIVE MEMORY ON QUTRITS BY MEANS OF QUANTUM ANNEALING

    See Supplemental Material. 9 Supplemental material to article “ASSOCIATIVE MEMORY ON QUTRITS BY MEANS OF QUANTUM ANNEALING” V.E. Zobov, I.S. Pichkovskiy Let us consider the dynamics of the system of two spins 1S with a Hamiltonian (3) and (9) in the process of transition at Tt  to a threefold degenerate ground state (2). At each time moment t for Hamilt...

    work page