Independent forensic re-analysis (red_team-gated; see QDAY_NONECC_FORENSIC_REANALYSIS.md) found the QPE histogram is a monotone, mirror-symmetric DC noise pile-up: the eight largest bins are exactly the eight bins nearest phase zero. The “SNR 25.4×” figure is bin-probability ÷ a uniform 1/128 floor — the wrong null for QPE; against a leakage-aware null there is no excess at α. The “α = 1/137 DETECTED” bin label was applied post-hoc (its higher mirror bin was labeled differently), and the “P(bin k)=P(bin 128−k)” symmetry cited here as proof of coherence is in fact the signature of an unresolved pile-up — evidence against a detection. What is true: a 15-qubit QPE circuit executed on real IBM hardware (ibm_fez); no α eigenphase was resolved. Everything below is retained as honest record. Do not cite any figure on this page as a result.
Quantum Hardware Research · Build 38
α = 1/137 “Detection”
— Retracted on Re-Analysis
This page documents a claim we made and then retracted. We originally reported a “first quantum hardware detection” of α = 1/137 as a distinct eigenphase from a 15-qubit QPE circuit on IBM Quantum hardware (ibm_fez, 156 qubits). On independent re-analysis the claim does not hold: the distribution is a monotone, mirror-symmetric DC-noise pile-up — the bright bins are simply the bins nearest phase 0 — and the “SNR 25.4×” figure was bin-probability divided by a uniform 1/128 floor, the wrong null for QPE. The mirror symmetry that was cited as “quantum coherence” is in fact the pile-up signature. The circuit and its public job ID are real; no α eigenphase was resolved. We keep this page as an honest record of the error and its correction.
Backend
ibm_fez
Qubits
15
α detection
RETRACTED
IP Claims
#27·#28·#29 (corrected)
IBM Quantum forensic anchor
Job d6ts6i0v5rlc73f38ftg
Submitted 2026-03-19T03:35:04 -06:00 · 53 seconds quantum runtime · 8,192 shots · status DONE
Independently verifiable on quantum.cloud.ibm.com → My Jobs. Full IBM job ledger →
KEY FINDINGS
Experimental Results
FINDING 01
α = 1/137 detection — RETRACTED on re-analysis
A 15-qubit QPE circuit ran on real IBM hardware (ibm_fez) — the run and its public job ID are genuine. ⚠ The prior claim of two 'distinct eigenphase peaks' (m=1 Coxeter at bin 0, α=1/137 at bin 1, SNR 25.4×) is fully retracted. Re-analysis: the bright bins are simply the bins nearest phase 0 in a monotone, mirror-symmetric DC-noise pile-up; 'SNR 25.4×' was bin-probability ÷ a uniform 1/128 floor — the wrong null for QPE; the α-bin label was post-hoc. No α eigenphase was resolved. This page is retained as an honest record of the claim and its correction.
HARDWARE PARAMETERS
QPE SPECTRUM
Phase Distribution — ibm_fez Hardware
E8 QUANTUM WALK
Coxeter Exponents — Hardware Recovery
The E8 Dynkin diagram (7 edges, Coxeter number h=30) was implemented as an 8-qubit quantum walk on ibm_fez. All 8 theoretical Coxeter exponents were recovered. Eigenvalue sum = 16.0 (exact, matching the E8 rank). The α coupling angle θ_α = π·α was detected as a distinct phase perturbation.
Backend
ibm_fez
Shots
4,096
Qubits
8
Walk steps
8
Coxeter h
30
Eigenval. sum
16.0 (exact)
E8 COXETER EXPONENTS {m} = Σ 16
m1
1
m2
7
m3
11
m4
13
m5
17
m6
19
m7
23
m8
29
ABSTRACT
arXiv draft — WITHDRAWN (not submitted)
⚠ The draft abstract that previously appeared here claimed a “first quantum hardware detection” of α=1/137 as a distinct E8-walk eigenphase. It was never submitted and is withdrawn. Independent re-analysis showed the reported “two distinct peaks” and “SNR 25×” are an artifact of a monotone, mirror-symmetric DC-noise pile-up scored against the wrong (uniform 1/128) null; the cited time-reversal symmetry P(bin k)=P(bin 128−k) is the pile-up signature, not coherence. No α eigenphase was resolved. The QPE circuit did run on real ibm_fez hardware (public job ID); that is the only defensible claim. The paper’s reproducibility thesis depended on the retracted measurement and cannot be submitted on it.
COLLABORATE