pith. sign in

arxiv: 2606.07727 · v1 · pith:C3IYQVKInew · submitted 2026-06-05 · 🪐 quant-ph · cs.CL· math.OC· q-fin.PM

Benchmarking Quantum Algorithmic Resilience for CVaR Portfolio Optimization: The Expressibility-Coherence Trade-off

Prashik N. Somkuwar , K. Srinivasan , G. Raghavan This is my paper

Pith reviewed 2026-06-27 21:32 UTC · model grok-4.3

classification 🪐 quant-ph cs.CLmath.OCq-fin.PM
keywords CVaR portfolio optimizationNISQ benchmarkingWS-QAOAHE-VQNNquantum financeexpressibility-coherence trade-offvariational quantum algorithmsheavy-hex topology
0
0 comments X

The pith

NISQ hardware forces a nonviable choice between expressibility and coherence in CVaR portfolio optimization.

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

The paper benchmarks WS-QAOA against HE-VQNN for a hybrid mean-variance plus CVaR portfolio objective on real IBM hardware. A classical-quantum hybrid proxy matrix is introduced to avoid the auxiliary qubit normally required for CVaR, allowing up to 16 NIFTY 50 assets to be mapped onto a heavy-hex processor. Systematic measurement of the SWAP tax shows that WS-QAOA delivers exact theoretical mappings yet incurs exponential nonlocal gate counts that destroy coherence, while HE-VQNN stays within coherence limits yet cannot represent the dense tail-risk correlations. The work concludes that the absence of all-to-all connectivity on present devices makes dense financial optimization problems fundamentally nonviable.

Core claim

When the CVaR objective is mapped via a hybrid proxy matrix that bypasses auxiliary qubits, WS-QAOA supplies exact theoretical encoding but experiences catastrophic decoherence from exponential nonlocal gate overhead on limited-connectivity hardware, whereas HE-VQNN maintains coherence yet lacks the mathematical expressibility needed to capture dense asset tail-risk correlations.

What carries the argument

The expressibility-coherence trade-off generated by routing SWAP gates when embedding a CVaR objective on heavy-hex topology.

If this is right

  • Dense correlation problems such as tail-risk CVaR cannot be solved at scale on current NISQ processors.
  • Algorithm selection for financial optimization must trade exact mathematical fidelity against hardware survival time.
  • The dominant cost in these mappings is the SWAP overhead required by limited qubit connectivity rather than the variational parameters themselves.
  • Hybrid classical-quantum proxies can remove certain qubit-count bottlenecks but leave the connectivity limitation intact.

Where Pith is reading between the lines

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

  • Hardware with native all-to-all links would remove the forced choice between the two algorithmic families.
  • Targeted error mitigation focused on nonlocal gates could extend the usable depth of WS-QAOA for finance tasks.
  • Simpler risk measures that avoid auxiliary variables may prove more practical on near-term devices than full CVaR.

Load-bearing premise

The hybrid proxy matrix accurately bypasses the CVaR auxiliary qubit without distorting the expressibility or coherence comparison.

What would settle it

Running the identical 16-asset instances on a device with all-to-all connectivity and finding that both algorithms achieve comparable solution quality without the observed trade-off.

Figures

Figures reproduced from arXiv: 2606.07727 by G. Raghavan, K. Srinivasan, Prashik N. Somkuwar.

Figure 1
Figure 1. Figure 1: 3D visualization of the hybrid Markowitz CVaR [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 3
Figure 3. Figure 3: Hybrid Quantum Classical architecture for HE [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Simulator level abstract quantum circuit diagram [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Optimizer convergence history for WS-QAOA [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Comparison of transpiled hardware overhead [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Absolute energy comparison between evalu [PITH_FULL_IMAGE:figures/full_fig_p007_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Measured state distributions on ibm_fez. (a) HE-VQNN preserves hardware fidelity but fails to isolate the optimal state due to poor mathematical ansatz expressibility. (b) WS-QAOA maps the geometric problem accurately, but its massive required physical routing degrades the amplitude distribution entirely into static noise. (a) Hardware ISA depth scaling. (b) Energy Approximation Error [PITH_FULL_IMAGE:fig… view at source ↗
Figure 9
Figure 9. Figure 9: Topological constraints and scaling analysis across varying asset pool sizes (4 to 16 assets). WS-QAOA suffers [PITH_FULL_IMAGE:figures/full_fig_p008_9.png] view at source ↗
read the original abstract

Quantum combinatorial optimization offers theoretical advantages for complex financial modeling, but physical implementation on Noisy Intermediate Scale Quantum (NISQ) devices is severely constrained by hardware topology. This study presents a hardware benchmarking analysis between a Hardware Efficient Variational Quantum Neural Network (HE-VQNN) and the Warm Start Quantum Approximate Optimization Algorithm (WS-QAOA) for a hybrid Mean Variance and Conditional Value at Risk (CVaR) portfolio objective. By implementing a novel classical quantum hybrid proxy matrix to bypass the CVaR auxiliary qubit bottleneck, we map up to 16 assets from the NIFTY 50 index onto an IBM heavy hex processor. We systematically quantify algorithmic resilience to the "SWAP tax" incurred during routing. Empirical results reveal a critical operational trade-off: WS-QAOA provides exact theoretical mapping but suffers catastrophic hardware decoherence due to exponential nonlocal gate overhead. Conversely, HE-VQNN preserves hardware coherence but lacks the mathematical expressibility to capture dense tail risk asset correlations. This study exposes the limitations of dense financial optimization on current architectures forces an nonviable choice between algorithmic inexpressibility and hardware decoherence. This is indicative of a deeper limitation as to what can and cannot be done with NISQ computers lacking in all-to-all connectivity.

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 / 2 minor

Summary. The paper benchmarks WS-QAOA against HE-VQNN for a hybrid mean-variance plus CVaR portfolio optimization problem on up to 16 NIFTY-50 assets mapped to IBM heavy-hex hardware. It introduces a novel classical-quantum hybrid proxy matrix claimed to bypass the auxiliary-qubit overhead of exact CVaR, then reports an empirical trade-off: WS-QAOA supplies an exact theoretical mapping but incurs catastrophic decoherence from exponential nonlocal gate overhead during routing, while HE-VQNN preserves coherence yet lacks sufficient expressibility to capture dense tail-risk correlations. The conclusion is that current NISQ devices without all-to-all connectivity force an untenable choice between algorithmic inexpressibility and hardware decoherence.

Significance. If the hybrid proxy matrix is shown to reproduce the exact CVaR objective (including tail quantiles on dense correlation matrices) without systematic bias, the work would supply concrete hardware evidence that dense financial combinatorial problems remain out of reach on present NISQ topologies. The use of real-device runs for 16 assets and explicit quantification of SWAP-tax effects would be a useful addition to the NISQ-finance literature; however, the absence of any reported error bounds or validation metrics for the proxy leaves the central trade-off claim unverified.

major comments (2)
  1. [Section 3] Section 3 (Hybrid Proxy Matrix construction): the manuscript presents the proxy as faithfully encoding the CVaR objective without auxiliary qubits, yet supplies neither analytic error bounds on the approximation for dense 16-asset correlation matrices nor numerical comparisons against the standard auxiliary-qubit CVaR formulation. Because the expressibility-coherence trade-off conclusion rests directly on this mapping being unbiased, the lack of validation is load-bearing.
  2. [Section 5] Section 5 / Results (hardware runs): the abstract asserts systematic quantification of algorithmic resilience and empirical observation of catastrophic decoherence versus inexpressibility, but the provided text contains no quantitative metrics, error bars, asset-selection criteria, or statistical tests. Without these, the reported trade-off cannot be assessed and the claim that WS-QAOA suffers exponential overhead while HE-VQNN is inexpressive remains unverified.
minor comments (2)
  1. [Abstract] Abstract, final sentence: 'forces an nonviable choice' should read 'forces a nonviable choice'.
  2. [Introduction / Methods] Notation: the acronym 'CVaR' is introduced without an explicit equation reference to the precise tail-risk functional being optimized; adding the standard definition (e.g., Eq. (X)) would improve clarity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their thorough review and constructive criticism. We address each major comment below and outline the revisions we will make to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Section 3] Section 3 (Hybrid Proxy Matrix construction): the manuscript presents the proxy as faithfully encoding the CVaR objective without auxiliary qubits, yet supplies neither analytic error bounds on the approximation for dense 16-asset correlation matrices nor numerical comparisons against the standard auxiliary-qubit CVaR formulation. Because the expressibility-coherence trade-off conclusion rests directly on this mapping being unbiased, the lack of validation is load-bearing.

    Authors: We agree that validation of the hybrid proxy matrix is necessary to substantiate the unbiased nature of the mapping. The revised manuscript will include a new subsection deriving analytic error bounds under the assumption of multivariate normal returns (standard in portfolio theory) and providing numerical benchmarks against the auxiliary-qubit CVaR formulation for asset counts up to 8. For the 16-asset NIFTY-50 instances, we will report the maximum observed deviation in the tail quantile estimates across the tested correlation matrices and discuss conditions under which bias remains negligible. revision: yes

  2. Referee: [Section 5] Section 5 / Results (hardware runs): the abstract asserts systematic quantification of algorithmic resilience and empirical observation of catastrophic decoherence versus inexpressibility, but the provided text contains no quantitative metrics, error bars, asset-selection criteria, or statistical tests. Without these, the reported trade-off cannot be assessed and the claim that WS-QAOA suffers exponential overhead while HE-VQNN is inexpressive remains unverified.

    Authors: The current results section relies on figures to convey the observed trade-off, but we concur that additional quantitative detail is required. The revision will expand Section 5 to specify the asset-selection criteria (top 16 NIFTY-50 constituents by trading volume with pairwise correlations above a 0.3 threshold), include error bars computed from 10 independent hardware runs per circuit, report the exact additional SWAP-gate counts for each algorithm, and add paired statistical tests (Wilcoxon signed-rank) on the final CVaR values to quantify the significance of the expressibility versus decoherence differences. revision: yes

Circularity Check

0 steps flagged

No significant circularity in empirical benchmarking study

full rationale

The paper reports hardware benchmarking results comparing WS-QAOA and HE-VQNN on a CVaR portfolio task for NIFTY assets, using a novel proxy matrix solely as an implementation tool to enable the runs. No equations, parameters, or claims reduce by construction to their own inputs; the expressibility-coherence trade-off is presented as an observed outcome on IBM hardware rather than a fitted or self-defined quantity. No self-citation chains, uniqueness theorems, or ansatzes are invoked as load-bearing premises. The analysis is self-contained against external hardware execution and does not rely on renaming known results or smuggling assumptions via prior author work.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available; no equations, methods sections, or data are provided to identify free parameters, axioms, or invented entities.

pith-pipeline@v0.9.1-grok · 5777 in / 1159 out tokens · 15903 ms · 2026-06-27T21:32:54.553688+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

34 extracted references · 3 canonical work pages · 1 internal anchor

  1. [1]

    Quantum Compu- tation and Quantum Information,

    M. A. Nielsen and I. L. Chuang, "Quantum Compu- tation and Quantum Information,"Cambridge Uni- versity Press, 2010

    2010

  2. [2]

    Portfolio Selection,

    H. Markowitz, "Portfolio Selection,"The Journal of Finance, vol. 7, no. 1, pp. 77–91, 1952

    1952

  3. [3]

    Optimization of Conditional Value at Risk,

    R. T. Rockafellar and S. Uryasev, "Optimization of Conditional Value at Risk,"Journal of Risk, vol. 2, pp. 21–42, 2000

    2000

  4. [4]

    A Quantum Approximate Optimization Algorithm

    E. Farhi, J. Goldstone, and S. Gutmann, "A Quan- tum Approximate Optimization Algorithm,"arXiv preprint arXiv:1411.4028, 2014

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  5. [5]

    Ising formulations of many NP prob- lems,

    A. Lucas, "Ising formulations of many NP prob- lems,"Frontiers in Physics, vol. 2, p. 5, 2014

    2014

  6. [6]

    Hardware efficient variational quantum eigensolver for small molecules and quan- tum magnets,

    A. Kandalaet al., "Hardware efficient variational quantum eigensolver for small molecules and quan- tum magnets,"Nature, vol. 549, pp. 242–246, 2017

    2017

  7. [7]

    Improving Variational Quan- tum Optimization using CVaR,

    P. K. Barkoutsoset al., "Improving Variational Quan- tum Optimization using CVaR,"Quantum, vol. 4, p. 256, 2020

    2020

  8. [8]

    Warm starting quantum optimiza- tion,

    D. J. Eggeret al., "Warm starting quantum optimiza- tion,"Quantum, vol. 5, p. 479, 2021

    2021

  9. [9]

    Quantum machine learning,

    J. Biamonteet al., "Quantum machine learning,"Na- ture, vol. 549, pp. 195–202, 2017

    2017

  10. [10]

    Variational quantum algorithms,

    M. Cerezoet al., "Variational quantum algorithms," Nature Reviews Physics, vol. 2, pp. 625–644, 2021

    2021

  11. [11]

    Multivariate stochastic approximation using a simultaneous perturbation gradient approx- imation,

    J. C. Spall, "Multivariate stochastic approximation using a simultaneous perturbation gradient approx- imation,"IEEE Transactions on Automatic Control, vol. 40, no. 9, pp. 332–341, 1992

    1992

  12. [12]

    Qiskit: An Open source Frame- work for Quantum Computing,

    Qiskit contributors, "Qiskit: An Open source Frame- work for Quantum Computing,"Zenodo, 2024

    2024

  13. [13]

    Quantum comput- ing for finance: Overview and prospects,

    R. Orús, S. Mugel, and E. Lizaso, "Quantum comput- ing for finance: Overview and prospects,"Reviews in Physics, vol. 4, p. 100028, 2019

    2019

  14. [14]

    A survey of quantum computing for finance,

    D. Hermanet al., "A survey of quantum computing for finance,"IEEE Transactions on Quantum Engi- neering, vol. 3, pp. 1–24, 2022

    2022

  15. [15]

    Solving the optimal trading tra- jectory problem using a quantum annealer,

    G. Rosenberget al., "Solving the optimal trading tra- jectory problem using a quantum annealer,"IEEE Journal of Selected Topics in Signal Processing, vol. 10, no. 6, pp. 1053–1060, 2016

    2016

  16. [16]

    Dynamic portfolio optimization with real datasets using quantum processors and quantum inspired tensor networks,

    S. Mugelet al., "Dynamic portfolio optimization with real datasets using quantum processors and quantum inspired tensor networks,"Physical Review Research, vol. 3, no. 1, p. 013006, 2022

    2022

  17. [17]

    Quantum approximate opti- mization of nonplanar graph problems on a planar superconducting processor,

    M. P. Harriganet al., "Quantum approximate opti- mization of nonplanar graph problems on a planar superconducting processor,"Nature Physics, vol. 17, pp. 332–336, 2021

    2021

  18. [18]

    Scaling of the quantum ap- proximate optimization algorithm on superconduct- ing qubit based hardware,

    J. Weidenfelleret al., "Scaling of the quantum ap- proximate optimization algorithm on superconduct- ing qubit based hardware,"Quantum, vol. 6, p. 870, 2022

    2022

  19. [19]

    Barren plateaus in quantum neural network training landscapes,

    J. R. McCleanet al., "Barren plateaus in quantum neural network training landscapes,"Nature Commu- nications, vol. 9, p. 4812, 2018

    2018

  20. [20]

    Connecting ansatz expressibility to gradient magnitudes and barren plateaus,

    Z. Holmeset al., "Connecting ansatz expressibility to gradient magnitudes and barren plateaus,"PRX Quantum, vol. 3, p. 010313, 2022

    2022

  21. [21]

    Ex- pressibility and entangling capability of parameter- ized quantum circuits for hybrid quantum classical algorithms,

    S. Sim, P. D. Johnson, and A. Aspuru-Guzik, "Ex- pressibility and entangling capability of parameter- ized quantum circuits for hybrid quantum classical algorithms,"Advanced Quantum Technologies, vol. 2, no. 12, p. 1900070, 2019

    2019

  22. [22]

    Topological and subsystem codes on lattice surgery networks with IBM quan- tum experience,

    C. Chamberlandet al., "Topological and subsystem codes on lattice surgery networks with IBM quan- tum experience,"Physical Review X, vol. 10, no. 1, p. 011022, 2020. 9

    2020

  23. [23]

    Error mitigation for short depth quantum circuits,

    K. Temme, S. Bravyi, and J. M. Gambetta, "Error mitigation for short depth quantum circuits,"Physi- cal Review Letters, vol. 119, no. 18, p. 180509, 2017

    2017

  24. [24]

    Efficient variational quan- tum simulator incorporating active error minimiza- tion,

    Y . Li and S. C. Benjamin, "Efficient variational quan- tum simulator incorporating active error minimiza- tion,"Physical Review X, vol. 7, no. 2, p. 021050, 2017

    2017

  25. [25]

    Hodson, Brendan Ruck, Hugh Ong, David Garvin, and Stefan Dulman

    M. Hodsonet al., "Portfolio optimization with a quantum algorithm,"arXiv preprint arXiv:1911.05296, 2019

    work page arXiv 1911

  26. [26]

    Quantum Computing in the NISQ era and beyond,

    J. Preskill, "Quantum Computing in the NISQ era and beyond,"Quantum, vol. 2, p. 79, 2018

    2018

  27. [27]

    Credit risk analysis using quantum computers,

    D. J. Eggeret al., "Credit risk analysis using quantum computers,"IEEE Transactions on Computers, vol. 70, no. 12, pp. 2136-2145, 2020

    2020

  28. [28]

    Quantum risk analysis,

    S. Woerner and D. J. Egger, "Quantum risk analysis," npj Quantum Information, vol. 5, no. 1, p. 15, 2019

    2019

  29. [29]

    Quantum machine learning for fi- nance,

    A. Pistoiaet al., "Quantum machine learning for fi- nance,"arXiv preprint arXiv:2109.04298, 2021

    work page arXiv 2021

  30. [30]

    Quantum optimization for portfolio allocation,

    A. Abbaset al., "Quantum optimization for portfolio allocation,"Quantum, vol. 7, p. 956, 2023

    2023

  31. [31]

    Practical quan- tum error mitigation for near future applications,

    S. Endo, S. C. Benjamin, and Y . Li, "Practical quan- tum error mitigation for near future applications," Physical Review X, vol. 8, no. 3, p. 031027, 2018

    2018

  32. [32]

    Machine learning of noise resilient quantum circuits,

    P. Cincioet al., "Machine learning of noise resilient quantum circuits,"PRX Quantum, vol. 2, p. 010324, 2021

    2021

  33. [33]

    Benchmarking the perfor- mance of portfolio optimization with QAOA,

    S. Brandhoferet al., "Benchmarking the perfor- mance of portfolio optimization with QAOA,"Quan- tum Information Processing, vol. 22, p. 25, 2022

    2022

  34. [34]

    Mitigating quantum errors via ran- domized compiling,

    S. Bravyiet al., "Mitigating quantum errors via ran- domized compiling,"Nature, vol. 608, no. 7924, pp. 676-681, 2022. 10

    2022