Quantum Computing through Computational Imaging: Representations, Measurement, and Inference
Abstract
An algebraic correspondence between an imaging model and a quantum state does not specify a computational method: input access, measurement and decoding determine what can be returned. This exposition develops four distinctions for computational-imaging and engineering readers. It separates amplitude encodings of input values from encodings of complete candidate labelings, explaining why amplitude storage does not provide full-array readout. Complex-field interference motivates unknown-phase inference: complementary measurements resolve ambiguities that local Fisher information alone does not detect. Reshaping joint amplitudes into a matrix connects singular-value decomposition to pure-state entanglement, while distinguishing Schmidt rank from conditions for efficient classical simulation. In a four-pixel segmentation, an objective becomes a cost operator and a sampling circuit. One layer of the quantum approximate optimization algorithm (QAOA) lowers mean energy from $1.675$ to $0.686$ relative to uniform sampling, yet makes a nonoptimal labeling more probable than the optimum 0.484 versus 0.365. Archived IBM and IQM measurements illustrate a separate six-variable local boundary decision. Low-rank state reconstruction and selected quantum algorithms connect these distinctions to an end-to-end resource account. The derivations and reproducible numerical references assume complex linear algebra and probability, without prior quantum mechanics. The accompanying Seeing Amplitudes browser laboratory supports interactive exploration; the article remains self-contained and connects each representation to the estimate or decision it permits and the resources that output requires.
Related
- Realization Divergence in Multi-Provider Quantum Computing: A Pre-Registered Study of What Survives Measurement
- When Measurement Constraints Favor Quantum Computational Sensing for Stealthy Power-Grid Attack Detection
- Do Quantum Measurement Advantages Survive to Astrophysical Inference? Seven Benchmarks in Optical Interferometry and Imaging
- GroupFunctions.jl: computing individual entries of the irreducible representations of the unitary group U(d)
- Hypervectors from Optical Disorder: Programmable Encoding and Optical Inference for Hyperdimensional Computing
BibTeX
@misc{kourousias2026quantum,
title = {{Quantum Computing through Computational Imaging: Representations, Measurement, and Inference}},
author = {George Kourousias and Sergio Carrato and Roberto Pugliese and Panagiotis Dimitrakis and Alberto Carini},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.40393},
eprint = {2610.40393},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.40393},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Quantum Computing through Computational Imaging: Representations, Measurement, and Inference
AU - George Kourousias
AU - Sergio Carrato
AU - Roberto Pugliese
AU - Panagiotis Dimitrakis
AU - Alberto Carini
PY - 2026
DA - 2026/10/29
PB - arXiv
DO - 10.48550/arxiv.2610.40393
UR - https://arxiv.org/abs/2610.40393
AB - An algebraic correspondence between an imaging model and a quantum state does not specify a computational method: input access, measurement and decoding determine what can be returned. This exposition develops four distinctions for computational-imaging and engineering readers. It separates amplitude encodings of input values from encodings of complete candidate labelings, explaining why amplitude storage does not provide full-array readout. Complex-field interference motivates unknown-phase inference: complementary measurements resolve ambiguities that local Fisher information alone does not detect. Reshaping joint amplitudes into a matrix connects singular-value decomposition to pure-state entanglement, while distinguishing Schmidt rank from conditions for efficient classical simulation. In a four-pixel segmentation, an objective becomes a cost operator and a sampling circuit. One layer of the quantum approximate optimization algorithm (QAOA) lowers mean energy from $1.675$ to $0.686$ relative to uniform sampling, yet makes a nonoptimal labeling more probable than the optimum 0.484 versus 0.365. Archived IBM and IQM measurements illustrate a separate six-variable local boundary decision. Low-rank state reconstruction and selected quantum algorithms connect these distinctions to an end-to-end resource account. The derivations and reproducible numerical references assume complex linear algebra and probability, without prior quantum mechanics. The accompanying Seeing Amplitudes browser laboratory supports interactive exploration; the article remains self-contained and connects each representation to the estimate or decision it permits and the resources that output requires.
ER -