How much information about a device is needed to use its common noise field as a quantum memory?
This repository studies a register coupled to one Gaussian dephasing field through fixed, spatially unequal coefficients. When those coefficients are supplied, a large protected subspace can be chosen from the known spectrum. When each coefficient is independently uncertain within a fixed interval, one code must instead work for every allowed profile. That information constraint changes the achievable storage time even though the bath still contains only one noise field.
For
The register is passive during the storage interval. Arbitrary joint encoding and recovery are allowed at its endpoints, including classical records and independent shared randomness. Every quantum system retained across the interval counts among the noisy carriers. Calibration supplies a classical description of the fixed coupling profile before the payload arrives; it does not reveal the subsequent noise realization. The model specifies these resources and the worst-profile half-diamond error criterion.
Fixed-error lifetime under interval uncertainty. If every
for all sufficiently large
Exact calibration-rate frontier for two coupling values. For
Here
Supplied-profile comparison. For known bounded couplings, spectral occupancy gives
at half-diamond error at most
The scientific consequence is a distinction between a small number of noise fields and access to their protected subspaces. Independent uncertainty in an extensive spatial profile can prevent scalable passive storage even when the corresponding supplied-profile device supports it.
| Question | Start here |
|---|---|
| How can I build up the ideas from Watrous's textbook? | Learning route, source bridge, and 30 milestones |
| What does the result mean physically, and which limit matters? | Interpretation and boundaries |
| What resources and errors are being counted? | Model |
| Why does uncalibrated lifetime stop growing with register size? | Continuous coupling uncertainty |
| Why does one calibration bit per carrier buy one asymptotic logical qubit per carrier? | Two-value calibration frontier |
| How does supplied or finite-precision calibration restore long storage? | Memory with supplied couplings |
| Which ideas are inherited, and how do nearby results differ? | Related work and primary-source record |
| What can be reproduced computationally? | Verification guide |
The science map gives proof dependencies. references.bib supplies the bibliography, and llms.txt maps reader questions to the authoritative files.
The learning route uses one educational anchor: John Watrous, The Theory of Quantum Information (2018). Its 11 background milestones and 19 narrative milestones connect channel and information foundations to the physical storage problem, the constructions, and the limits of the results.
The results price retained classical calibration information, not the experiments needed to acquire it. They allow ideal endpoint operations and do not establish efficient code compilation or decoding. The continuous-profile information cost is bounded but not exactly characterized. Allowing ideal operations during storage changes the problem: an explicit toggling sequence removes unknown sign correlations and permits positive-rate storage without calibration. These boundaries are developed in Interpretation.
Two deterministic Python programs check finite physical sign averages, complete-positivity comparisons, exact code eigenspaces, information accounting, Gaussian error-floor probabilities, and an actual seven-block recovery under correlated noise. They support the calculations and examples; the all-size statements are proved in the scientific pages. See Verification for commands, pinned dependencies, results, and numerical tolerances.
This repository serves as a record of the work and a guide for the author’s self-directed learning. For discussion or potential collaboration, please contact Ruge Lin at gogoko699@gmail.com.
The repository uses the MIT license.