Karim Eldefrawy

Cryptography, Cybersecurity, Privacy

Co-founder and CTO at Confidencial.io
2017-2021: SRI
2011-2016: HRL Laboratories
2006-2010: PhD@UC Irvine

Scientific curiosity

Scientific knowledge map · Paper #60

Quantum Optimization Heuristics with an Application to Knapsack Problems

Wim van Dam, Karim Eldefrawy, Nicholas Genise, and Natalie Parham

2021 · IEEE International Conference on Quantum Computing and Engineering (QCE)

  • Theory
  • Applied
  • algorithm

What does the paper try to establish?

Can shallow QAOA-style heuristics use a classical greedy solution to stay near feasible high-value regions of a constrained knapsack search space and outperform comparably simple classical heuristics?

What is the proposed answer?

The paper initializes qubits from smoothed greedy marginals and introduces hourglass and copula mixers that preserve those biases while exploring nearby solutions. Statevector simulations on small hard-instance families report better performance than selected shallow classical baselines, with explicit classical-simulability and scaling caveats.

Six dimensions, kept separate

The chart summarizes documented evidence and process. It is not a correctness probability, confidence score, or ranking, and no composite score is calculated.

The visual spider chart requires JavaScript. The complete values and rationales follow in text.

LowMediumHighN/A = not assessed

A smaller value means less documented support for that dimension, not that the paper is false or unimportant.

Epistemic evidence Medium

The algorithms are derived and evaluated systematically on five instance families with transparent baselines and caveats, but evidence is limited to small statevector simulations and does not establish scaling or quantum advantage.

Biased initial states, hourglass mixer, and copula mixer Classical comparison setup Simulation results and sensitivity Conclusions, limits, and future work
Auditability High

The complete checked-in archive copy has page count and SHA-256 and is linked to the official DOI; algorithms and experiment settings are explicit, though code and raw results are absent.

Motivation, contributions, and classical-simulability caveat Official IEEE QCE publication identity
Production provenance Medium

Authors, venue, versions, funding, algorithms, and experimental choices are documented; code revision, raw data, contributor roles, and tool versions are not.

Official IEEE QCE publication identity Five hard-instance distributions Simulation results and sensitivity
External scrutiny Medium

IEEE QCE publication establishes venue review, but review reports and independent reproduction are not represented.

Official IEEE QCE publication identity
Reception Low

The dated exact-DOI OpenAlex record located 4 citations. Under the author-defined rule, 0 through 8 located citations is Low; counts vary by index and date.

Dated citation-count snapshot
Contribution significance Medium

The bias-preserving and copula mixer constructions are reusable ideas for constrained optimization, but the paper itself rules out a quantum-advantage interpretation for one path and leaves scaling open.

Motivation, contributions, and classical-simulability caveat Conclusions, limits, and future work

Assessment: Ai draft author review pending · 2026-07-11 · rubric 0.2. These dimensions describe documented support and process, not truth, correctness, or a universal ranking. No composite score is calculated.

Hierarchical knowledge map

Collapse a branch for a top-level reading, or follow its source links and child nodes to audit the evidence and boundaries underneath it.

paper

Quantum Optimization Heuristics with an Application to Knapsack Problems

Two biased QAOA-style optimization algorithms for constrained knapsack instances, analyzed algebraically and evaluated by small statevector simulations.

Motivation, contributions, and classical-simulability caveat
  1. algorithm Biased xQAOA specified

    A logistic smoothing of the lazy-greedy stopping ratio assigns each item a qubit inclusion probability, creating a constant-depth initial state near the capacity boundary rather than a uniform superposition.

    Biased initial states, hourglass mixer, and copula mixer Knapsack-specific xQAOA design
  2. evidence group Comparison design controlled simulation

    Quantum heuristics are compared with lazy greedy, very greedy, warm-start simulated annealing, and a global one-step annealing variant on strongly correlated, inverse-strong, profit, and two spanner distributions.

    Five hard-instance distributions Classical comparison setup
  3. claim

    Better results than selected shallow heuristics

    simulation result

    Across the sampled ten-item instances, the hourglass variant typically exceeds the four chosen classical heuristics and the copula variant improves further; infeasible outputs count as zero rather than being postselected away.

    Simulation results and sensitivity
  4. claim

    Limited sensitivity on the tested size

    numerical observation

    The tested copula mixer generally prefers maximal anticorrelation, performance varies weakly across the sampled bias strengths, and favorable beta angles cluster in two regions, suggesting some parameter reuse at n equals 10.

    Simulation results and sensitivity
  5. limitation group Boundaries material

    Experiments are statevector simulations at n equals 10 and depth one, use heavily optimized per-instance parameters, and compare only simple classical methods. The authors expect more elaborate classical algorithms to outperform these quantum heuristics and leave larger n, deeper p, and parameter-scaling behavior open.

    Quantum algorithms and parameter optimization Simulation results and sensitivity Conclusions, limits, and future work
  6. scrutiny

    External scrutiny

    venue reviewed

    The paper appeared at IEEE QCE 2021. Public review reports and independent experimental reproduction are not linked.

    Official IEEE QCE publication identity

Source index

Locators state the depth of the current audit. PDF page numbers, where present, are one-based file pages; metadata-, summary-, and abstract-bounded records explicitly identify their limitations.

  1. Motivation, contributions, and classical-simulability caveat Abstract and Sections 1-1.1, PDF pages 1-2
  2. QAOA and penalty-function baseline Section 2, PDF pages 2-3
  3. Biased initial states, hourglass mixer, and copula mixer Section 3, PDF pages 3-7
  4. Knapsack definition and classical baselines Section 4, PDF pages 7-9
  5. Knapsack-specific xQAOA design Section 5, PDF pages 9-11
  6. Five hard-instance distributions Section 6.1, PDF pages 11-12
  7. Classical comparison setup Section 6.2, PDF pages 12-13
  8. Quantum algorithms and parameter optimization Sections 6.3.1-6.3.2, PDF pages 13-15
  9. Simulation results and sensitivity Sections 6.3.3-6.3.4, PDF pages 15-16
  10. Conclusions, limits, and future work Section 7, PDF page 18
  11. Official IEEE QCE publication identity DOI 10.1109/QCE52317.2021.00033
  12. Dated citation-count snapshot OpenAlex reported 4 citations when accessed 2026-07-11