arXiv:2610.08241
Multi-time advantage for certifying quantum coherence
Karol Sajnok, Fabio Costa
Abstract
Non-classical dynamics in open quantum systems can be hidden in temporal correlations that are invisible to standard state- and channel-based tests. We define coherence for multi-time quantum processes and show that access to intermediate times can reveal coherence absent from all corresponding pairwise channels. Incoherent processes are fixed points of dephasing on all input and output spaces, and their coherence is quantified by relative entropy. As a proof of principle, we use an Ising model in which all three pairwise channels are incoherent while the full three-time process remains coherent. We observe this multi-time coherence on a superconducting quantum processor using a coherence witness and full process tomography.
AI Summary
Can intermediate-time interventions reveal coherence when every pairwise dynamical channel is incoherent? This paper defines basis-dependent coherence of a multi-time process operator $W$ by its failure to remain invariant under dephasing of every input and output leg. It quantifies this using $\mathcal C_{\rm rel}(W)=S(\mathcal D(\omega))-S(\omega)$, where $\omega=W/\operatorname{tr}W$, and develops linear coherence witnesses. The framework includes convexity and monotonicity under dephasing-covariant superprocesses. An exactly solvable Ising model couples one system qubit to two persistent environmental qubits. At observation times $0,\tau,2\tau$, with $\tau=\pi/(8g)$, all three pairwise channels are completely dephasing, yet the three-time process has coherence $\tfrac12\ln2$. Resetting the memory between intervals eliminates this coherence. A four-Pauli witness has ideal expectation $-1/2$. The effect requires only classical memory and is related explicitly to spin echo, rather than requiring quantum environmental memory. Appendices give a quantum-memory example and a single-memory construction whose coherence is at least $(N-1)\ln2$ over $N$ intervals. Ancilla-assisted measurements on an IBM superconducting processor yield witness magnitude $0.3486\pm0.0146$ and tomographic process coherence $0.2689$, versus ideal $0.3466$; reconstructed pairwise-channel coherences remain below $0.022$. The experimental channels therefore retain residual coherence rather than exhibiting exact separation. The notion differs from classicality under fixed-basis sequential probes. Pairwise channels in non-Markovian settings depend on the specified earlier closure, and tomographic entropy estimates use eigenvalue clipping without reported uncertainties.
The Qiskit implementation was developed with assistance from ChatGPT (5.6 Sol) and verified by the authors.
Error found: The claimed monotonicity under dephasing-covariant deterministic superprocesses is false: incoherent input replacement can increase normalized-Choi coherence from (ln 2)/2 to ln 2. The single-environment appendix displays the wrong process matrix for its stated interaction times and later cites it at different times, although its reported coherence values remain correct.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 7+5 pages, 4+3 figures, 1 table