Quantum correlations and exceptional points in superconducting processors: collective measurements, generative learning, and dynamical control
Year: 2026
Type: dissertation
Abstract: Quantum systems are inherently linear at the level of their evolution equations, yet the properties that make them useful for computation, communication, and sensing are profoundly nonlinear.Entanglement cannot be captured by any single linear observable; generating a state with prescribed quantum correlations requires navigating a highly non-convex optimization landscape; and the dynamics of open quantum systems can develop singularities in parameter space that have no counterpart in ordinary Hermitian physics.This dissertation addresses all three aspects of this nonlinearity on current superconducting quantum hardware: how to measure it efficiently, how to generate it deliberately, and how to exploit it as a resource rather than treat it as an obstacle.The characterization problem is addressed in Chapter 2 through a family of multicopy measurement protocols that operate on pairs or small sets of identically prepared state copies rather than on a single copy at a time.This approach reduces the number of measurement settings required to estimate entanglement from the exponential cost of full quantum state tomography, reducing the exponent base from 4 n to 2 n in the number of qubits, with a verified 67% reduction in overhead for two-qubit systems on IBM Quantum hardware.A central result of this chapter is the identification of partial-transpose moments as chirality-chirality correlators: the difference between partial-transpose and purity moments encodes the scalar spin chirality, a quantity familiar from condensed-matter physics as the order parameter of chiral spin liquids.This connection provides a physical interpretation for an otherwise abstract algebraic operation, and it extends naturally to bound entanglement detection through the realignment non-Hermiticity gap.Using the same controlled-SWAP circuit infrastructure, a multi-channel spectral classifier detects bound entanglement across all three known families of 3 × 3 positive-partial-transpose entangled states, including states invisible to any single realignment criterion, achieving 99.6% cross-validated recall at zero false positives in simulation and 94% recall on ibm_fez hardware.This is the first family-agnostic demonstration of bound entanglement detection on a gate-based superconducting processor.Gröbner basis decision trees provide an additional route to negativity estimation, converting spectral moments into entanglement values adaptively and achieving a five-to sevenfold reduction in measurement settings relative to full tomography.
Author: Tulewicz, Patrycja
Institution: Adam Mickiewicz University in Poznań
Language: English
Subfield: Atomic and Molecular Physics, and Optics
Field: Physics and Astronomy
Domain: Physical Sciences
Keywords: quantum entanglement, entanglement measurement, superconducting quantum processors, partial transposition, realignment criterion, spin chirality, chiral spin liquid, non-Hermitian exceptional points, open quantum systems, non-convex optimization, quantum state tomography, spectral classification
Open Access status: green
OpenAlex ID: W7214508592
