Pith. sign in

REVIEW 3 major objections 5 minor 23 references

A sparse, truncated state vector can recover the most probable output of peaked quantum circuits while storing far fewer than 2^n amplitudes.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

2026-07-10 17:28 UTC pith:N2VJGGTI

load-bearing objection Solid open-source systems paper: sparse+truncated state-vector engine with dual backends that recovers the mode of a 44-qubit peaked circuit under modest k; claims are scoped and the main failure mode is already stated. the 3 major comments →

arxiv 2607.07816 v1 pith:N2VJGGTI submitted 2026-07-08 quant-ph cs.ET

A Sparse and Truncated State Vector Simulator for Peaked Circuits

Diogo R. Ferreira This is my paper
classification quant-ph cs.ET
keywords peaked circuitssparse state vectortruncated simulationtop-k truncationprobability-mass truncationgate fusionGPU accelerationquantum circuit simulation
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Peaked circuits are built so that one bit-string dominates the final measurement distribution. The paper shows that this structure can be exploited by keeping only a sparse list of the largest amplitudes and discarding the rest after each block of gates. Two truncation rules are offered: keep the top-k amplitudes, or keep enough amplitudes to retain a chosen fraction of the total probability mass, then renormalize. Combined with gate reordering that postpones the introduction of new qubits and with fusion of consecutive gates into larger unitaries, the method recovers the correct peak on a 44-qubit “sharp-peak” example while using fewer than thirty-two terms. Runtime grows linearly with the number of retained terms, and a GPU backend yields roughly a ten-fold speed-up until device memory is exhausted. The same truncation can fail when intermediate probability mass is still spread over an enormous number of basis states, so the approach is presented as a practical, approximate tool rather than a universal simulator.

Core claim

For peaked circuits whose probability mass remains sufficiently concentrated on a small set of basis states throughout evolution, a sparse state vector truncated by either a hard top-k limit or a retained-probability-mass threshold, after gate reordering and fusion, correctly identifies the most probable output bit-string while using far fewer than 2^n amplitudes.

What carries the argument

Sparse truncated state-vector evolution: amplitudes and basis indices are stored only for the currently nonzero terms; after each fused multi-qubit unitary a top-k or p-mass truncation retains the largest-magnitude components and renormalizes, with all arithmetic expressed as vectorized array operations that run on CPU or GPU.

Load-bearing premise

Intermediate states of a peaked circuit stay concentrated enough that discarding small amplitudes after each fused block never eliminates the eventual mode.

What would settle it

Run the open-source simulator on a known peaked circuit that deliberately spreads a large fraction of intermediate probability mass over many basis states; if the truncated evolution systematically returns the wrong mode while the untruncated evolution returns the correct one, the concentration premise fails.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper presents an open-source classical simulator that represents the quantum state as a sparse list of nonzero amplitudes and basis indices, then truncates after each fused gate block either by a hard top-k limit on the number of terms or by a retained probability-mass fraction p, followed by renormalization. Operations are expressed as bulk vectorized array transformations (including segmented sums for amplitude collisions) with matching CPU and GPU backends that keep 128-bit complex amplitudes and 64-bit indices resident on device. A preprocessing strategy of dependency-respecting gate reordering plus fusion of single- and two-qubit blocks is used to delay growth of the support. On the 44-qubit BlueQubit “sharp peak” circuit the mode is recovered with k < 2^5; runtime is shown to scale linearly with k (Fig. 2) and the growth of the support under both truncation rules is plotted (Figs. 3–5).

Significance. If intermediate states of peaked circuits remain sufficiently concentrated, the approach supplies a practical, memory-bounded classical tool that can identify the dominant output bit-string for system sizes beyond the reach of dense state-vector simulators, while remaining complementary to exact sparse simulators and tensor-network methods. The open-source artifact (https://github.com/diogoff/qstvec/), the fully vectorized formulation that ports cleanly to GPU, the explicit linear-scaling measurements, and the candid statement of the failure mode when mass is diffusely spread constitute concrete, inspectable contributions that strengthen the toolbox for classical simulation of peaked circuits.

major comments (3)
  1. [Section VI] Section VI (and the brief mention of “a few others” in the conclusion): the empirical support for mode recovery is essentially a single 44-qubit instance (plus unspecified additional circuits). Because the central claim is scoped to circuits whose probability mass “stays sufficiently concentrated,” a broader, publicly documented suite—e.g., the BlueQubit peaked-circuit set—with success/failure rates versus k and p is required to make the scope of applicability falsifiable rather than anecdotal.
  2. [Section VI.B] Section VI.B / Figs. 2–5: all scaling and growth curves are reported for one circuit only. Without at least one head-to-head comparison (runtime and peak-recovery accuracy) against a dense simulator, an exact sparse tool such as qblaze, and an MPS baseline on the same instances (or on smaller instances that all methods can finish), the practical advantage claimed for the truncated sparse approach remains unquantified.
  3. [Sections III and VI] Sections III and VI: the manuscript invokes the known relation that average fidelity tracks retained probability mass, yet never reports measured state fidelity, total-variation distance, or empirical mode-recovery probability as functions of k or p on the circuits studied. Adding these curves would turn the accuracy claims from qualitative to quantitative and would directly test the load-bearing concentration assumption.
minor comments (5)
  1. [Section IV.A] Section IV.A: the description of how new basis indices are generated by bitwise operations and how collisions are resolved by a segmented sum is clear at a high level, but a short pseudocode fragment or reference to the concrete library primitive used would improve reproducibility.
  2. [Section VI.A] Section VI.A: the gate-reordering heuristic (“keep the set of currently active qubits as small as possible”) is only sketched; an explicit statement of the commuting/dependency rules or a pointer to the implementation would allow others to reproduce the same block structure.
  3. [Figures 2–5] Figures 2–5: axis labels and legends are readable, yet the captions do not state the hardware (CPU model, GPU model, memory) used for the timings; adding this information would make the order-of-magnitude GPU speedup claim more precise.
  4. [Introduction / Related work] References: several recent sparse-simulator and peaked-circuit papers are cited, but a short related-work paragraph contrasting the truncation policy and vectorization strategy with GraFeyn, SparQSim and the sparse-tensor method of Miller et al. would help readers locate the contribution.
  5. [Throughout] Typographical: “V ector” appears with a space in two subsection headings (IV.A, IV.B); “stillexact” lacks a space (Section II); “ann-qubit” should be “an n-qubit”.

Circularity Check

0 steps flagged

No circularity: implementation-and-benchmark paper whose truncation parameters are free user knobs and whose success is checked against known circuit modes.

full rationale

The paper presents a sparse state-vector simulator with top-k and p-mass truncation, gate reordering/fusion, and CPU/GPU backends, then reports empirical scaling and success on BlueQubit peaked-circuit examples (notably the 44-qubit sharp-peak instance with k < 2^5). Truncation thresholds are explicit free parameters, not fitted to force a desired peak; the most-probable bit-string is recovered by direct simulation and compared to the known mode of the example circuits. There is no derivation that reduces a claimed prediction to a quantity defined by that same prediction, no uniqueness theorem imported from the authors, and no self-citation that carries the central claim. The acknowledged limitation (intermediate mass may spread so that truncation discards the eventual mode) is stated as a scope condition, not hidden by circular reasoning. Consequently the derivation chain is self-contained and non-circular.

Axiom & Free-Parameter Ledger

2 free parameters · 3 axioms · 0 invented entities

The central claim rests on standard unitary quantum mechanics plus the domain hypothesis that peaked circuits keep enough probability mass on few basis states for truncation to preserve the mode. The only free knobs are the user-chosen truncation budgets k and p; no new physical entities are postulated.

free parameters (2)
  • k (top-k term limit)
    Hard upper bound on retained amplitudes; chosen by the user to fit memory/runtime and directly controls whether the peak is kept.
  • p (probability-mass fraction)
    Minimum retained probability mass after each block; user-chosen accuracy/resource trade-off that can still allow exponential growth if set high.
axioms (3)
  • standard math Unitary evolution of pure states and Born-rule probabilities (standard quantum mechanics)
    Used throughout Sections I–IV to define amplitude updates and truncation by squared magnitude.
  • domain assumption For the peaked circuits of interest, intermediate and final probability mass remains sufficiently concentrated that discarding small amplitudes after fused blocks still leaves the eventual mode intact
    Explicitly stated as the operating regime in Section VI; the paper notes that the assumption fails for some deep/entangling peaked circuits.
  • standard math Vectorized column-wise multiplies followed by segmented sums correctly implement multi-qubit unitaries on a sparse list of (basis, amplitude) pairs
    Section IV; correctness follows from the definition of the Kronecker action of a gate on selected qubits.

pith-pipeline@v1.1.0-grok45 · 14111 in / 2459 out tokens · 39010 ms · 2026-07-10T17:28:31.559798+00:00 · methodology

0 comments
read the original abstract

In a class of quantum circuits known as peaked circuits, the goal is to predict the most probable bit string at the output of the circuit. Since these circuits are designed to have a sharp peak in their output distribution, in principle it should be possible to simulate them using a truncated state vector with a limited number of terms, or a fraction of the total probability mass. This approximate simulation can be carried out on a classical computer with a sparse representation that stores only the nonzero amplitudes of the state vector, in contrast to the dense representations that are common in most quantum simulators. For efficiency, all operations on the state vector should be vectorized to the furthest possible extent and, if available, hardware acceleration can also be used. This work describes how these requirements were met in an open-source implementation, and discusses its performance and limitations.

Figures

Figures reproduced from arXiv: 2607.07816 by Diogo R. Ferreira.

Figure 1
Figure 1. Figure 1: Excerpt of BlueQubit’s sharp peak circuit. which connects all qubits in a ring structure. This ring con￾nectivity across 44 qubits and 580 instructions yields a large, deep, and highly entangling circuit, which is unfavorable for dense simulators and also for tensor network methods such as MPS (matrix product states), because accurately representing the state requires large bond dimensions [19]. A. Circuit… view at source ↗
Figure 3
Figure 3. Figure 3: Number of terms during circuit simulation with top- [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 2
Figure 2. Figure 2: Circuit simulation time when applying top- [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 5
Figure 5. Figure 5: Number of terms vs. p-mass threshold. fraction of the total probability mass. Both methods prioritize the largest-amplitude components, under the assumption that these will suffice to identify the most probable output of a peaked circuit, provided that the probability mass remains concentrated on a few basis states. Although the approach works for the peaked circuit con￾sidered here and a few others, a car… view at source ↗

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 · 8 internal anchors

  1. [1]

    Open Quantum Assembly Language

    A. W. Cross, L. S. Bishop, J. A. Smolin, and J. M. Gambetta, “Open quantum assembly language,”arXiv:1707.03429, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  2. [2]

    On verifiable quantum advantage with peaked circuit sampling

    S. Aaronson and Y . Zhang, “On verifiable quantum advantage with peaked circuit sampling,”arXiv:2404.14493, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  3. [3]

    How to Write a Simulator for Quantum Circuits from Scratch: A Tutorial

    M. J. McGuffin, J.-M. Robert, and K. Ikeda, “How to write a simulator for quantum circuits from scratch: A tutorial,”arXiv:2506.08142, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  4. [4]

    Quantum computing with Qiskit

    A. Javadi-Abhariet al., “Quantum computing with Qiskit,” arXiv:2405.08810, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  5. [5]

    Qulacs: a fast and versatile quantum circuit simulator for research purpose,

    Y . Suzukiet al., “Qulacs: a fast and versatile quantum circuit simulator for research purpose,”Quantum, vol. 5, p. 559, Oct. 2021

    work page 2021

  6. [6]

    Zenodo, Aug

    Cirq Developers,Cirq. Zenodo, Aug. 2025. [Online]. Available: https://doi.org/10.5281/zenodo.11398048

    work page doi:10.5281/zenodo.11398048 2025

  7. [7]

    PennyLane: Automatic differentiation of hybrid quantum-classical computations

    V . Bergholmet al., “PennyLane: Automatic differentiation of hybrid quantum-classical computations,”arXiv:1811.04968, 2022

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  8. [8]

    GraFeyn: Efficient parallel sparse simulation of quantum circuits,

    S. Westricket al., “GraFeyn: Efficient parallel sparse simulation of quantum circuits,” in5th IEEE International Conference on Quantum Computing and Engineering (QCE24), vol. 01, 2024, pp. 1132–1142

    work page 2024

  9. [9]

    qblaze: An efficient and scalable sparse quantum simulator,

    H. Venev, T. Udomsrirungruang, D. Dimitrov, T. Gehr, and M. Vechev, “qblaze: An efficient and scalable sparse quantum simulator,”Proc. ACM Program. Lang., vol. 9, no. OOPSLA2, Oct. 2025

    work page 2025

  10. [10]

    SparQSim: An efficient quantum algorithm simulator leveraging state sparsity,

    T.-P. Sunet al., “SparQSim: An efficient quantum algorithm simulator leveraging state sparsity,”Chinese Physics B, 2026

    work page 2026

  11. [11]

    Leveraging state sparsity for more efficient quantum simulations,

    S. Jaques and T. H ¨aner, “Leveraging state sparsity for more efficient quantum simulations,”ACM Transactions on Quantum Computing, vol. 3, no. 3, Sep. 2022

    work page 2022

  12. [12]

    A com- prehensive review of quantum circuit optimization: Current trends and future directions,

    K. Karuppasamy, V . Puram, S. Johnson, and J. P. Thomas, “A com- prehensive review of quantum circuit optimization: Current trends and future directions,”Quantum Reports, vol. 7, no. 1, 2025

    work page 2025

  13. [14]

    Classical simulability of Clifford+T circuits with Clifford-augmented matrix product states

    B. Vanhecke, M. V . Damme, J. Haegeman, L. Vanderstraeten, and F. Verstraete, “Tangent-space methods for truncating uniform MPS,” arXiv:2412.17209, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  14. [15]

    Approxi- mate simulation of complex quantum circuits using sparse tensors,

    B. N. Miller, P. K. Elgee, J. R. Pruitt, and K. C. Cox, “Approxi- mate simulation of complex quantum circuits using sparse tensors,” arXiv:2602.04011, 2026

    work page arXiv 2026

  15. [16]

    Segmented operations using matrix multiplications,

    A. Sobczyk, G. Sorrentino, and A. Zouzias, “Segmented operations using matrix multiplications,”arXiv:2506.23906, 2025

    work page arXiv 2025

  16. [17]

    Classical simulation of peaked shallow quantum circuits,

    S. Bravyi, D. Gosset, and Y . Liu, “Classical simulation of peaked shallow quantum circuits,” inProc. 56th Annual ACM Symposium on Theory of Computing (STOC 2024), 2024, p. 561–572

    work page 2024

  17. [18]

    Heuristic quantum advantage with peaked circuits,

    H. Gharibyan, M. Z. Mullath, N. E. Sherman, V . P. Su, H. Tepanyan, and Y . Zhang, “Heuristic quantum advantage with peaked circuits,” arXiv:2510.25838, 2025

    work page arXiv 2025

  18. [19]

    Matrix product states and projected entangled pair states: Concepts, symmetries, theorems,

    J. I. Cirac, D. P ´erez-Garc´ıa, N. Schuch, and F. Verstraete, “Matrix product states and projected entangled pair states: Concepts, symmetries, theorems,”Rev. Mod. Phys., vol. 93, p. 045003, Dec. 2021

    work page 2021

  19. [20]

    Lazy qubit reordering for accelerating parallel state-vector-based quantum circuit simulation,

    Y . Teranishi, S. Hiraoka, W. Mizukami, M. Okita, and F. Ino, “Lazy qubit reordering for accelerating parallel state-vector-based quantum circuit simulation,”ACM Transactions on Quantum Computing, vol. 6, no. 4, Aug. 2025

    work page 2025

  20. [21]

    Fast and memory-efficient classical simulation of quantum machine learning via forward and backward gate fusion

    Y . Kawase, “Fast and memory-efficient classical simulation of quan- tum machine learning via forward and backward gate fusion,” arXiv:2603.02804, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  21. [22]

    GPU-Accelerated Quantum Simulation: Empirical Backend Selection, Gate Fusion, and Adaptive Precision

    P. Kumaresan, P. Muruganantham, L. Rajendran, and S. Sivasubramani, “GPU-accelerated quantum simulation: Empirical backend selection, gate fusion, and adaptive precision,”arXiv:2604.03816, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  22. [23]

    A review on quantum circuit optimization using ZX-calculus,

    T. Fischbach, P. Talbot, and P. Bouvry, “A review on quantum circuit optimization using ZX-calculus,”arXiv:2509.20663, 2026

    work page arXiv 2026

  23. [24]

    Quantum circuit optimization by graph coloring,

    H. Lee, K. C. Jeong, and P. Kim, “Quantum circuit optimization by graph coloring,”Quantum, vol. 10, p. 1996, Feb. 2026

    work page 1996