We report a set of pre-registered experiments on the three 156-qubit IBM Heron r2 processors reachable on a single open-plan account (ibm_kingston, ibm_marrakesh, ibm_fez), carried out on 14–15 October 2026. The object of study is not computational speedup but channel reliability: whether fixed symbolic messages can be delivered intact through a present-day noisy processor when the processor's per-wire readout response is measured fresh in the same sitting, and whether two such processors can exchange messages with per-position verification. Messages are 17-symbol sequences drawn from a 13-symbol test alphabet whose preparations collapse to nine distinguishable readout classes. A receiver combining per-wire selection from a fresh characterization, grammar-constrained decoding, and a corpus-frequency tie-break delivered 36/36 messages on ibm_kingston in two independent campaigns, 36/36 on ibm_marrakesh, and 35/36 on ibm_fez, against stale-plan and uncharacterized controls ranging from 2/36 to 30/36 depending on backend and control type. The grammar-constrained receiver without the corpus prior scored 30/36, 30/36, and 29/36 on the three backends; its losses are a single systematic mechanism — a deterministic tie on one battery message — identified from saved counts and closed by the prior. In a separate experiment, ibm_kingston and ibm_marrakesh exchanged self-introduction streams authored by each backend from its own fresh characterization, through a six-stage symmetric relay, with 204/204 position-flights received exactly and zero substitutions. In a counts-only blind test, an independent decoder with no shared codebook recovered the discreteness and grouping structure of two authored streams (pooled pairwise grouping agreement 0.971 against a pre-registered 0.90 strong-pass bar) while overestimating the alphabet size (reported 8, true value 6). A census of repeated characterizations shows that the ranking of a device's clearest wires turns over within hours (same-day top-17 wire overlap of 1–2 of 17 on ibm_marrakesh), while per-wire response quality moves by only ~0.01, so characterizations are within-sitting instruments. No speedup, no machine networking, and no machine understanding is claimed. The full receiver implementation is withheld as a trade secret; verification is offered through the published job IDs, retained raw counts, and tiered third-party reproduction.
We report a set of pre-registered experiments on the three 156-qubit IBM Heron r2 processors reachable on a single open-plan account (ibm_kingston, ibm_marrakesh, ibm_fez), carried out on 14–15 October 2026. The object of study is not computational speedup but channel reliability: whether fixed symbolic messages can be delivered intact through a present-day noisy processor when the processor's per-wire readout response is measured fresh in the same sitting, and whether two such processors can exchange messages with per-position verification. Messages are 17-symbol sequences drawn from a 13-symbol test alphabet whose preparations collapse to nine distinguishable readout classes. A receiver combining per-wire selection from a fresh characterization, grammar-constrained decoding, and a corpus-frequency tie-break delivered 36/36 messages on ibm_kingston in two independent campaigns, 36/36 on ibm_marrakesh, and 35/36 on ibm_fez, against stale-plan and uncharacterized controls ranging from 2/36 to 30/36 depending on backend and control type. The grammar-constrained receiver without the corpus prior scored 30/36, 30/36, and 29/36 on the three backends; its losses are a single systematic mechanism — a deterministic tie on one battery message — identified from saved counts and closed by the prior. In a separate experiment, ibm_kingston and ibm_marrakesh exchanged self-introduction streams authored by each backend from its own fresh characterization, through a six-stage symmetric relay, with 204/204 position-flights received exactly and zero substitutions. In a counts-only blind test, an independent decoder with no shared codebook recovered the discreteness and grouping structure of two authored streams (pooled pairwise grouping agreement 0.971 against a pre-registered 0.90 strong-pass bar) while overestimating the alphabet size (reported 8, true value 6). A census of repeated characterizations shows that the ranking of a device's clearest wires turns over within hours (same-day top-17 wire overlap of 1–2 of 17 on ibm_marrakesh), while per-wire response quality moves by only ~0.01, so characterizations are within-sitting instruments. No speedup, no machine networking, and no machine understanding is claimed. The full receiver implementation is withheld as a trade secret; verification is offered through the published job IDs, retained raw counts, and tiered third-party reproduction.
On 4-5 July 2026, an identical shock protocol was executed on the same 156-qubit IBM Heron R2 architecture (ibm_fez) but with the QITE lambda feedback replaced by uniformly random rotation angles per cycle. This random-angle null control preserves every structural variable — RY+RZ+CZ layer, Heavy-Hex CZ coupling, 4096 shots/cycle, seed 20260602, the same 150 cycles, and the same six shocks at the same cycles (20/40/60/80/100/120), the same target qubits (q78, q69, q67, q47, q37, q16) with the same amplitudes (l1,l2) — while removing only the decentralized per-qubit lambda feedback dynamics. Backend selection was driven by queue availability; both are 156-qubit Heron R2 devices with identical architecture, and calibration differences cannot account for the qualitative burst-count contrast reported below. The result is a clean negative: zero collective bursts across all six shocks. The four non-parametric statistical tests (identical methodology to v1) all confirm the absence of shock-driven collective response: (1) permutation test yields p = 1.0 for shock alignment against the noise-saturated per-cycle transitions; (2) pairwise Pearson correlation between per-transition delta-p1 vectors drops from r_mean = +0.65 in the multi-shock run to r_mean = +0.017 in the null; (3) cross-session hypergeometric overlap between null "top-active" qubits and the v1 multi-shock top-active set is not significant (Jaccard = 0.11, p = 0.24); (4) Mann-Whitney U on shock-cycle max|delta p1| versus baseline shows ratio 0.989 with p = 0.89 — statistically indistinguishable from noise. Baseline global coherence exp(-8*Var(p1)) sits at 0.413 +/- 0.022 across all 150 cycles under random angles, versus 0.99 +/- 0.01 under QITE feedback in v1 — two distinct dynamical regimes. That the shock protocol itself executed correctly on the null run is verified locally at the target qubit: for each of the six shocks, the measured p1[q_target] at the shock cycle agrees with the theoretical prediction from theta_shock = 2*atan2(l2, l1) to within <5% (q78 shock: predicted 0.016, measured 0.018; q69 shock: predicted 0.985, measured 0.939; and so on for all six). One cycle after each shock, the target qubit returns to the random baseline distribution — with no feedback dynamics, there is no restoring trajectory. The shocks are physically executed; what is absent is the collective response. Causal conclusion: the QITE per-qubit lambda feedback dynamics are necessary for the observed collective burst phenomenology. The circuit structure (RY+RZ+CZ on Heron R2 Heavy-Hex topology) and shock protocol alone do not produce bursts. Six identical shocks on QITE-stabilized baseline produce 10 collective bursts with 6/10 shock-aligned (p = 0.008); the same six shocks on random baseline produce zero. This 10-vs-0 hardware contrast, verified by four independent non-parametric tests, establishes feedback necessity as a falsifiable causal claim rather than a phenomenological observation. The v2 bundle contains the complete null-control primary data (150-cycle JSONL, consolidated JSON, p1 numpy matrix), all IBM Runtime job IDs with per-cycle timestamps, Tee'd console logs of both the initial run and the resume run (a mid-run job-poll stall on 4 July required a resume from cycle 7 with deterministic per-cycle RNG seeds), five analysis figures including direct null-vs-multi comparison plots, the full side-by-side statistical comparison JSON, the v1 multi-shock p1 matrix and statistical test outputs for reanalysis, an 8-page supplement PDF (Section 5) integrating with the v1 paper, and SHA-256 checksums of every file in MANIFEST_v2.json. Backend (v2 null control): IBM Heron R2 (ibm_fez), 156 qubits, 4096 shots/cycle.Companion patents: DPMA Az. 10 2025 003 906.9 (QITE), 10 2025 004 891.2 (QISE).Predecessor (v1): Westrick, Reproducible Collective Burst Modes on 156 Qubits: Decentralized Per-Qubit QITE Lambda Field on IBM Heron R2, 1 June 2026. On 1 June 2026, a decentralized per-qubit lambda field was implemented on the IBM Heron R2 processor (ibm_kingston, 156 qubits) and probed with a multi-shock excitation protocol. Unlike the predecessor work (Westrick, Controllable Quantum Collective States on 156 Qubits: Experimental Results of the QITE Formalism on IBM Heron R2, 12 April 2026), which tested a globally controlled QITE quantum collective, each qubit here carries its own lambda state and receives, in the feedback loop, only its own measurement statistic p1_i. The only inter-qubit coupling propagates physically through the CZ gates of the Heavy-Hex topology. Six shocks were applied at six different positions on the chip: central (q*=78), nearest neighbor (d=1), intermediate (d=3, d=5), deep (d=8), and edge (d=12). Each individual shock triggered a chip-wide burst within 1-8 cycles (peak |delta p1| between 0.308 and 0.366; 10 to 22 simultaneously active qubits). Four additional spontaneous bursts were detected between shocks, for a total of 10 bursts in 150 cycles. The 6/6 reproducibility was verified by four independent non-parametric statistical tests, all positive: (1) permutation test p = 0.008 against chance alignment; (2) pairwise burst-vector Pearson correlation r_mean = +0.65 with all 15 pairs above noise; (3) cross-session reproducibility 9/10 qubit overlap between two independent IBM Quantum sessions, hypergeometric p = 0.008; (4) Mann-Whitney U test of burst amplitude versus Phase-A baseline, ratio 2.7, p < 0.0001. This bundle contains the complete primary data (raw JSON logs, NumPy matrices), all 27 visualizations, IBM job IDs of every submitted job, console logs of all runs, the pre-registration documents with locked detection thresholds, and SHA-256 checksums of every file (MANIFEST.json) for independent verification. Backend: IBM Heron R2 (ibm_kingston), 156 qubits, 4096 shots/cycle.Patent filings: DPMA Az. 10 2025 003 906.9 (QITE), 10 2025 004 891.2 (QISE).Predecessor: Westrick, Controllable Quantum Collective States on 156 Qubits, 12 April 2026.
H. Billah and M.M.H. Sarker contributed equally to this work. Interpreting a fixed quantum circuit's output as evidence of a hardware effect requires ruling out two things usually conflated with it: finite-sample fluctuation, and structure already present in the circuit's own noiseless output. Less commonly checked is a third: whether the reference circuit compared against is the one actually executed, once compiler-level qubit routing is accounted for. We execute a fixed 35-qubit, 243-gate diagnostic circuit on IBM Heron-family hardware (three runs, two backends) and show this third check is essential: the routed circuit submitted to hardware has different noiseless predictions than the logical circuit intended, for several qubits. Against the correct, routed-circuit reference, the aggregate deviation is significant in two of three runs, and four registers (0, 9, 14, 27) show a significant output-probability deviation in every run — though we distinguish a high-confidence subset, {0, 9, 27}, with tightly time-matched calibration, from qubit 14, included only on an exploratory basis given its calibration was retrieved roughly two weeks post-execution. Same-backend replicates correlate more strongly by the Pearson measure, though a rank-based (Spearman) test shows this gap does not survive at the level of register ordering, indicating substantial leverage from a small number of registers rather than a broadly robust pattern; these three executions also share a byte-identical payload and are not independent replications in the strictest sense. The mechanism is only partially resolved: independent single-qubit readout error and raw gate count are ruled out as complete explanations for most of these registers, but the simplest version of the leading hypothesis for the single most-affected register does not reproduce the observed cross-backend pattern even with live, backend-specific timing data. A second, structurally different circuit validated only in simulation calibrates the statistical procedure's behavior under one synthetic noise model; it does not itself establish generality to real hardware, and the specific register set found here should not be expected to generalize beyond this study. Note on scope: this is a methodology and case-study contribution centered on executed-circuit references for fixed-circuit diagnostics, not a characterization of IBM Heron hardware in general or a randomized-benchmarking experiment in the formal sense. Supporting data, the routed OpenQASM circuits, raw measurement counts, and analysis code for this study are available on request from the corresponding author. Affiliation: Sonicium Quantum Lab Ltd., Dhaka, Bangladesh.
On 1 June 2026, a decentralized per-qubit lambda field was implemented on the IBM Heron R2 processor (ibm_kingston, 156 qubits) and probed with a multi-shock excitation protocol. Unlike the predecessor work (Westrick, QITE Hardware Results 156Q, 8-12 April 2026), which tested a globally controlled QITE quantum collective, each qubit here carries its own lambda state and receives, in the feedback loop, only its own measurement statistic p1_i. The only inter-qubit coupling propagates physically through the CZ gates of the Heavy-Hex topology. Six shocks were applied at six different positions on the chip: central (q*=78), nearest neighbor (d=1), intermediate (d=3, d=5), deep (d=8), and edge (d=12). Each individual shock triggered a chip-wide burst within 1-8 cycles (peak |delta p1| between 0.308 and 0.366; 10 to 22 simultaneously active qubits). Four additional spontaneous bursts were detected between shocks, for a total of 10 bursts in 150 cycles. The 6/6 reproducibility was verified by four independent non-parametric statistical tests, all positive: (1) permutation test p = 0.008 against chance alignment; (2) pairwise burst-vector Pearson correlation r_mean = +0.65 with all 15 pairs above noise; (3) cross-session reproducibility 9/10 qubit overlap between two independent IBM Quantum sessions, hypergeometric p = 0.008; (4) Mann-Whitney U test of burst amplitude versus Phase-A baseline, ratio 2.7, p < 0.0001. This bundle contains the complete primary data (raw JSON logs, NumPy matrices), all 27 visualizations, IBM job IDs of every submitted job, console logs of all runs, the pre-registration documents with locked detection thresholds, and SHA-256 checksums of every file (MANIFEST.json) for independent verification. Backend: IBM Heron R2 (ibm_kingston), 156 qubits, 4096 shots/cycle. Patent filings: DPMA Az. 10 2025 003 906.9 (QITE), 10 2025 004 891.2 (QISE). Predecessor: Westrick, QITE Hardware Results 156Q (8-12 April 2026).
We report the first hardware-executed quantum-classical coordination pipeline on NISQ hardware for a societal coordination problem task-to-member matching in a 500-agent community economy. The three-stage architecture (AI decomposition → 19 qubit quantum semantic scoring on IBM Heron R2 classical QUBO assignment) executes end-to-end on three Heron R2 processors (ibm_fez,ibm_marrakesh, ibm_kingston). We observe a three-tier regret stratification: (i) a noiseless ceiling at 1.48% (five independent simulator controls + ibm_marrakesh), (ii) a classical-kernel tier at 0.86% via the Shin–Teo–Jeong (2024) dequantization kernel K_c, and (iii) a hardware tier at 0.61% reproduced independently on ibm_fez and ibm_kingston with identical θ. Since K_c's RKHS provably contains the quantum kernel's RKHS at logical depth, the hardware tier is a reproducible operating regime outside the dequantized function class. A companion multi-cycle society-model experiment shows pipeline weight choice is a time-horizon-dependent policy decision; the QUBO matcher also acts as a performative Granovetter machine (80% more weak ties than greedy, p = 9.1 × 10⁻⁵, d ≈ 27 at n = 10 runs).
The Frozen Token Reaches Quantum: a complete, pre-registered map of a deterministic single-token encoder's monitoring laws on a real quantum processor — fingerprint, privacy, integrity, qubit health, drift, and a molecular-state twin Randolph James Ferlic, M.D. and Kimberly Kate Ferlic — Fieldstone Analytics, LLC, Austin, TX, USA Preprint · Zenodo DOI: 10.5281/zenodo.23670079 · CC-BY 4.0 · Community: spiral-domain-encoder-campaign · This record characterizes previously filed and published methods applied to a new signal source (a quantum processor's measurement outputs) and discloses no new patentable subject matter; the one candidate new application (client-side quantum-job attestation) was deliberately not pursued as intellectual property — see the notice below. Abstract A frozen, deterministic single-token encoder reduces a window of inputs, by a fixed computation, to one 8-bit token and a bounded ≤256-cell lookup, with the nearest-centroid distance serving as a label-free novelty score. A prior program established, across many classical sensing modalities, a predictive theory of this encoder — a competence-zone separability law, a per-entity self-twin law, a compose-versus-recommission law, a cost-not-accuracy law, a deployment-sizing law, a privacy-inversion bound, a safety-interlock characterization, and a geometry-gated drift-adaptation capability. This record asks whether those laws extend to a new kind of entity: a real cloud quantum processor, probed only through the data a present-day gate-model machine returns — measurement counts (single-shot in-phase/quadrature readout is no longer exposed). Established by a pre-registration with frozen falsifiable bands and a simulate-then-validate-on-hardware discipline, the answer is that the monitoring laws reach a quantum processor essentially in full, from counts alone, with no new algorithm. A frozen 1-byte token fingerprints which of three IBM Heron-r2 156-qubit backends ran a job (balanced accuracy 1.00; chance 0.333), proves the signature stable rather than shot-noise (same-device test-retest 0.65 ≪ 1.00), exposes a privacy-inversion (device re-identification 1.00 under an abstracted backend), acts as a result-integrity dead-switch (wrong-device 1.00; gross tamper ≥1σ 0.98–1.00, including a real circuit-level tamper at 1.00 with the novelty spiking ~95× and no cheap null-space evasion; honest subtle-tamper blind spot 0.51), recovers a device's qubit-level calibration ranking from a four-second probe (Spearman 0.88; eight of ten worst qubits; generalizing to a second device at 0.86), and builds a molecular-state twin (H₂ on real hardware: bond-distortion Spearman 1.00; energy ordering 0.771; distortion-twin robust across a finer 12-geometry ladder and two Heron devices, 0.96–0.99). The keystone laws transfer: the separability screen forecasts reach when matched to the monitor's geometry (matched 1.00; generic 0.69, a characterized blind spot); cost-not-accuracy holds (byte = 200-tree forest, Δ0.00); sizing holds (≈4 commissioning sub-batches for fingerprinting, ≈8 for integrity); compose-versus-recommission holds exactly and closes a clean division of labor with the drift law. A hybrid classical→quantum→LLM pipeline monitor localizes which stage faulted at 100% for three bytes. Temporal drift (QF-2) is positive and tracks the device's calibration state rather than the clock — over seven snapshots across nine days the token's label-free novelty tracks IBM's own reported readout error at Spearman +0.89, more tightly than elapsed time (+0.68), because the device recalibrates ~daily so calibration drift is recal-cyclic (the novelty climbs and then falls back as the device is recalibrated, and the token follows it up and back down) and it tracks readout-layer drift specifically, not gate drift (+0.64), C10 reaching the quantum service-life axis, honestly scoped as a bounded n=7 result. Honest negatives are reported with equal weight (no untrained quantum-feature-map advantage, no label-derivative "synergy," no cheap two-qubit-gate-error readout). The transferable lesson: the cheapest possible deterministic monitor — one frozen byte — reaches the most advanced computing hardware we have, as a fingerprint, a privacy probe, an integrity dead-switch, and a health and drift gauge, from the measurement counts the machine already returns, without a new algorithm or a gradient. Highlights · The monitoring laws reach a real quantum processor from measurement counts alone — no new algorithm. A frozen 1-byte token fingerprints which of three IBM Heron-r2 backends ran a job (1.00; chance 0.333) and the test-retest control (same device 0.65 ≪ cross-device 1.00) proves it is a stable device signature, not shot noise. · A privacy inversion on cloud quantum. An "anonymized" backend is re-identifiable from job counts (QF-4, re-identification 1.00) — extending the program's confidentiality-is-not-anonymity bound to a quantum processor; dual-use (user-side audit / provider side-channel). · A result-integrity dead-switch with a declared blind spot. From counts, label-free: trips on a wrong/substituted device (1.00) and on gross result tampering (≥1σ, 0.98–1.00, clean dose-response) — including a real circuit-level tamper (an X-flip of half the register before measurement) at 1.00 with the novelty spiking ~95×, with no cheap null-space evasion (an adaptive quiet-direction override is caught immediately, 1.00) and no false trip on a genuine same-device re-run (0.50) — and honestly fails on subtle device-consistent (on-manifold) tampering (0.51) — defense-in-depth, not a cryptographic verifier. · Qubit-level health from a four-second probe. The token recovers the vendor's own readout-error calibration ranking (Spearman 0.88, 156 qubits) and flags eight of its ten worst qubits — a cheap, fresh, label-free health map that generalizes to a second Heron device (Spearman 0.86 at matched shots). · A molecular-state digital twin on real hardware. Commissioned on H₂'s equilibrium geometry, the token's label-free novelty tracks bond distortion (Spearman 1.00) and preserves the dissociation-energy ordering (0.771), on the device's noisy prepared state — the filed self-twin on a quantum-chemistry target; the distortion-twin is robust across a finer 12-geometry ladder and two Heron devices (0.96–0.99), its finer energy recovery shot-limited rather than device-limited. · The keystone laws transfer. The separability screen forecasts monitor reach iff matched to the monitor's geometry (matched 1.00; a generic two-class screen 0.69, blind to additive tamper); cost-not-accuracy holds (byte equals a 200-tree forest, Δ0.00); sizing holds (≈4/≈8 commissioning sub-batches); and compose-versus-recommission holds exactly — a device-A monitor flags a healthy device B at 100% and must be recommissioned, closing a clean division of labor with the drift law (small within-device drift is the drift law's regime; large cross-device shift is recommission's). · A three-byte hybrid-pipeline monitor. An array of 1-byte self-twins localizes which stage of a classical→quantum→LLM pipeline faulted at 100%, provided each stage is probed with the matched geometry (workload self-twin for covariate faults; a fixed reference probe for the quantum stage). · Honest negatives, and no new intellectual property. An untrained quantum feature-map carries no classical-data advantage, stacking label-derivative codebooks yields no super-additive "synergy," and a cheap echo probe cannot rank small two-qubit gate errors from counts. Every result is the filed, published self-twin applied to a new signal source; the one candidate new application (client-side quantum-job attestation) was weighed as intellectual property and deliberately not pursued, the chosen protection being this publication. · Trust in a cloud the user cannot see, at near-zero cost. The self-twin triad answers the practical cloud-quantum questions — did my job run on the device I was allocated (QF-1), is the abstracted backend unlinkable (QF-4; no), was my result substituted or grossly corrupted (QF-5), which qubits were healthy at run time (QF-6) — from the measurement histogram the user already receives, with no provider cooperation, trusted server, or extra circuits. The provider-side alternatives (randomized benchmarking, tomography, cryptographic verification) spend many designed circuits, privileged access, or interaction; this is the "read the receipt you already have" layer, and it is defense-in-depth beneath verification, not a replacement for it. · One primitive, more than twenty orders of magnitude. The identical mechanism now monitors a vibrating bearing, an electrocardiogram, a radio emitter, an industrial audio machine, and the measurement readout of a 156-qubit superconducting processor — a single cheap, deterministic, auditable primitive reaching the exotic end of the computing-substrate spectrum, the strongest evidence yet for the program's cost-not-accuracy premise (see the breadth figure). · Temporal drift (QF-2) is positive and vendor-corroborated — the token tracks the device's calibration *state*, not the clock. Over seven snapshots across nine days and six recalibration cycles, the token's label-free novelty tracks IBM's own historical reported readout error at Spearman +0.89 — more tightly than elapsed time (+0.68) — and the two-qubit gate error at only +0.64. Because the device recalibrates ~daily, calibration drift is recal-cyclic (the novelty rises, then is recalibrated back toward baseline), and the token follows the calibration state up and back down: the two late snapshots have low novelty and low reported readout error even though their gate error is the worst of the series. C10 on the quantum service-life axis, honestly bounded at n=7. The transferable lesson A single, auditable, sub-milliwatt monitoring primitive — one frozen deterministic byte — now d
Experimental measurement data of the QITE formalism (Quantum-Inspired Thermal Equilibrium) on IBM quantum processors of the Heron R2 generation with 156 physical qubits. Altogether 75+ jobs on two different backends (ibm_kingston and ibm_fez), over 830,000 shots, carried out on 8, 10, and 11 April 2026. Seven experiment series: (1) Static phases (R=0.5 / 1.0 / 2.0): three clearly separated Hamming-weight distributions (16.3σ significance), positive ZZ correlations under QITE (+0.318 at R=2.0) versus negative ones in the null experiment (−0.061). Controlled comparison experiment with null hypothesis (identical circuit structure with random parameters). (2) Real-time feedback loop (8 cycles): two targeted perturbations (R=3.0 and R=0.3); the system returns autonomously to equilibrium within 2 cycles. First demonstration of dynamic controllability of a QITE quantum collective on a real quantum computer. (3) Cross-backend reproducibility: identical results on ibm_kingston and ibm_fez — deviations within a few percent. (4) Zero-noise extrapolation (ZNE): circuit folding (1×, 3×, 5×) with Richardson extrapolation. Result: coherence 0.474 → 0.474 extrapolated (improvement: −0.0005). The optimized circuit (depth 13, 120 CZ gates) already operates at the hardware noise floor — ZNE cannot improve it further because the circuit is already near-optimal. (5) Scaling comparison 27 vs. 156 qubits: the 27-qubit version on the same backend (ibm_kingston) shows ZZ correlation of 0.788 against 0.318 at 156 qubits (Δ = +0.470). Circuit depth drops from 13 to 6, CZ gates from 120 to 13. Confirms: the QITE signal scales with gate quality. (6) Extended feedback loop (20 cycles): four extreme perturbations — R=3.0 (excess), R=0.3 (deficit), R=5.0 (5-fold excess), R=0.1 (10-fold deficit). Maximum coherence C=0.785 (highest value ever measured). The system reorganizes itself after each perturbation within 2–4 cycles. Long-term stability demonstrated across 20 cycles and 436 seconds of total runtime. (7) 3-minute stability demonstration (20 cycles, 262.7 seconds): dedicated long-term test with two targeted shocks (R=4.0 and R=0.2). Result: ones_fraction remains between 0.491 and 0.499 over 4.4 minutes (maximum deviation from equilibrium: 0.009). Drift across the entire runtime: +0.003. Complete recovery after both shocks. 18/18 non-shock cycles stable. First demonstration of a self-correcting quantum field on a real quantum computer over several minutes. Circuit optimization relative to 8 April: reduction of circuit depth from 295 to 13 (factor 22×), two-qubit gates from 1,946 to 120, by exclusive use of native Heron R2 gates (CZ, SX, RZ, X) and direct lambda encoding via RY(θ) with θ = 2·arctan(λ₂/λ₁). The system demonstrably produces controllable, correlated, self-organizing quantum states on real hardware — no simulation, no theory, but physical measurement data of a concrete technical system. All 59 IBM job IDs are contained in the raw data and verifiable on quantum.ibm.com. German patent applications Az. 10 2025 003 906.9 (QITE) and Az. 10 2025 004 891.2 (QISE). Final document (12 April 2026) — 21 sections, 9 experiment series, 372 individual quantum measurements. Extensions compared to the previous version: Section 19: 9-minute experiment (209 cycles, 156 qubits, 4096 shots) Phase A: null test (23 cycles, R=1.0) — baseline control measurement Phase B: 8 shocks (R=0.1 to R=8.0) incl. double shock — complete recovery after each shock Section 20: overall summary of all 9 experiment series (8–12 April 2026) Section 21: raw-data appendix with SHA-256 checksums of all 9 JSON files Core results: 372 quantum measurements on IBM Heron R2 (ibm_kingston, ibm_fez) Dynamic range R=0.1 to R=8.0 (factor 80) with complete self-correction The null test proves: recovery is an active process, not a hardware artefact No drift, no collapse across 209 consecutive measurements Optimized circuit: depth 13, 120 CZ gates, at the hardware noise floor (ZNE-confirmed)