Log — Corrections

Corrections log

Every reported error is logged here publicly — whether or not it turns out to be right — and corrected figures supersede rather than overwrite, so history stays checkable. Correcting quietly is a violation of our own policy.

Logsince 2026-08-20

Date	Where	What was wrong → what it says now
2026-09-24	Catalog: PsiQuantum, Xanadu Aurora, Xanadu Borealis, ORCA PT-2	The technology field said trapped-ion (PsiQuantum, both Xanadu systems) and superconducting (ORCA PT-2), contradicting each record’s own sourced technology detail. All four are photonic and are now labelled photonic. No measured figure changed. A consistency check now runs in CI.
2026-09-24	Course: Quantum Compilation — superconducting vs all-to-all	“About 65 times sooner” → about 67 (200 µs ÷ 3 µs ≈ 67). Also a product 0.9559 × 0.970 = 0.9276 was really ≈ 0.9272; recomputed with 4-digit factors so every step checks. Results (92.8% / 90.0%) unchanged.
2026-09-24	Course: Quantum Compilation — qubit mapping; circuit depth	A Try-this said both halves of a pair “moved home” after two SWAPs; they moved away (the result is 110). A depth example mixed two circuits; it now states H–H has depth 2, and 0 after cancelling.
2026-09-24	Course: Inside Quantum Processors — annealing	Removed an unsourced “about 28 chained physical qubits per variable” figure; replaced with a clearly labelled illustrative number. Couplers-per-qubit now cites our Advantage (15) and Advantage2 (20) records.
2026-09-24	Course: Quantum Algorithms — variational algorithms	A 1% readout bias (0.010) was called “the same size as” ±0.027 shot noise; it is smaller. Now: “not much smaller than”, with the arithmetic shown.
2026-09-24	Several chapters	Removed duplicated passages pasted twice mid-text, and reworded two lines that implied cloning (“CX copies q0’s value”) or called amplitudes “chances”.

Found an error? Write to corrections@qpu137.com — every report is recorded here with its resolution.