Purpose of this Document

This document records the conceptual discussion between Dr. Michael Rhoades and Claude (Anthropic) regarding Phase IV of the Quantum Computational Creativity research project. It is intended as a reference for when active development begins, preserving the creative and technical intentions formulated during the planning stages. It also serves as supporting context for the IBM Quantum Credits Program grant application.

Overview

Phase IV represents the most ambitious execution of the Quantum Computational Creativity methodology to date. Sixteen circuit schemas, each utilizing nine qubits, will engage 144 of IBM Kingston’s 156 available qubits in a single large-scale hardware execution. This is not simply a scaling of previous phases. Phase IV introduces fundamental changes at the circuit architecture level, changes driven directly by discoveries made during Phase III development that exposed both the limitations of the previous approach and the path beyond them.

A pilot execution on April 1, 2026 confirmed that 144-qubit parallel circuit execution on Kingston is fully viable: 6,816 circuits executed in 7 minutes 47 seconds of quantum time, at a calibrated rate of 0.069 seconds per circuit. 180 minutes of compute time on Kingston were subsequently made available following usage that exceeded the 20-minute threshold, valid for one year from April 2026. This resource provides the foundation for the initial steps of Phase IV execution. A separate application to IBM’s Quantum Credits Program, requesting extended compute time to support full-scale execution, has been submitted and is currently pending review.

All Phase IV circuit designs will be validated in Qiskit Aer simulation before any IBM compute time is committed. This simulation-first discipline, established during Phase III development, ensures that hardware execution is reserved for designs that have demonstrated their value and viability computationally.

The IBM/Quantum Inspire hardware comparison benchmark, documented fully on the quantumcomputationalcreativity.com research pages and reproduced in the hardware section below, provides the empirical foundation for the compute time estimates submitted in the grant application.

Discoveries from Phase III that Inform Phase IV

Three specific challenges emerged during Phase III audio and visual development. Each is inherent to the Phase III circuit architecture and cannot be resolved in post-production. Each must be addressed at the circuit design level in Phase IV.

The Chord Challenge

Within each Phase III schema, all eight qubits evolve together through the same parameter sweep, arriving at significant quantum states simultaneously. When complex samples derived from Bloch sphere trajectories are played back across all eight voices at once, the result collapses into an undifferentiated mass. Individual voices cannot be tracked perceptually. The solution requires circuits in which different qubits arrive at significant states at different times — temporal separation built into the quantum architecture itself.

The Waveform Similarity Challenge

Phase III tomographic waveforms are too similar across qubits and across schemas. Entanglement keeps qubits correlated, which is physically accurate but compositionally limiting. The eight voices within a schema move together rather than independently. Phase IV circuits must be designed to produce genuinely distinct waveform shapes between qubits, with timbral variety arising from asymmetric gate structures rather than from post-production manipulation.

Unified Temporal Phrasing

A third insight functions less as a challenge and more as a design opportunity: if non-uniform temporal distribution is built into the circuit execution through non-uniform parameter sampling, then audio and visual material emerge from the same underlying temporal structure from the beginning. The phrasing is not imposed on the data after the fact. It is inherent in the quantum execution itself. This is the deepest realization of the QCC unified methodology to date: compositional phrasing as an emergent quantum phenomenon, arising from the circuit execution rather than from artistic interpretation applied afterward.

Phase IV Design Principles

Principle 1 – Temporal Separation by Design

Phase IV circuits are designed so that different qubits arrive at significant quantum states at different times. Approaches under development include staggered parameter sweeps, sequential entanglement activation, and topology-driven temporal structure. The entanglement topology becomes the compositional structure. The physics and the music are unified at the circuit level rather than reconciled afterward.

Principle 2 – Timbral Variety by Design

Phase IV circuits move away from uniformly scaled rotation gates toward asymmetric gate structures that produce distinctly different waveform shapes between qubits. Different rotation gate types per qubit, deliberately varied entanglement strengths, non-uniform parameter ranges, and non-uniform parameter sampling all contribute to a palette of genuinely distinct quantum voices within each schema.

Principle 3 – Compositional Requirements as Circuit Constraints

The most significant methodological shift in Phase IV is treating desired sonic and perceptual outcomes as design constraints on the quantum circuits themselves. In Phases II and III, circuits were designed for quantum hardware execution and compositional interpretation came afterward. In Phase IV, the composition informs the circuit architecture from the beginning. The quantum physics and the artistic vision are no longer sequential — they are simultaneous.

Dynamic Entanglement Architecture

The most significant conceptual advancement in Phase IV circuit design is making entanglement topology itself a dynamic variable. In Phases II and III, each schema had a fixed entanglement topology determined at design time and held constant throughout execution. Phase IV proposes circuits in which different qubit pairs become entangled and disentangled at different points during the parameter sweep.

A qubit’s Bloch trajectory is shaped by what it is entangled with and when. A qubit entangled with one partner in early parameter steps, decoupled in the middle range, and re-entangled with a different partner in the final steps will exhibit three distinctly different characters across those regions. This is genuine timbral variation arising from the circuit architecture, not from post-production processing.

Offset Phase RZZ Entanglement Layers

The primary implementation approach uses multiple RZZ entanglement layers with offset phase angles. Rather than a single entanglement layer at uniform strength, the circuit employs several layers whose gate angles are functions of θ with different phase offsets:

Layer 1: RZZ(sin(θ)) between qubits 0-1, 2-3, 4-5, 6-7 — peaks at θ=π/2

Layer 2: RZZ(sin(θ+π/4)) between qubits 1-2, 3-4, 5-6 — peaks offset by π/4

Layer 3: RZZ(sin(θ+π/2)) between qubits 0-4, 1-5, 2-6, 3-7 — peaks offset by π/2

Each layer’s entanglement peaks and troughs at different points in the sweep. Qubits experience constantly shifting entanglement relationships, with some strengthening while others weaken. This directly addresses the chord challenge, the waveform similarity challenge, and unified temporal phrasing simultaneously.

Mid-Circuit Measurements as Topology Change Triggers

A more radical approach uses mid-circuit measurements to trigger topology changes during execution. When a qubit is measured mid-circuit, its wavefunction collapses to a definite classical state, breaking all entanglement relationships at that point. The circuit then continues with the collapsed qubit entering new entanglement relationships with different partners.

The resulting Bloch trajectory shows complex entangled evolution in the first section, a sharp discontinuity at the measurement point, and a new evolution under different entanglement conditions afterward. Those discontinuities are themselves compositionally significant. In audio they translate to timbral shifts. In visual output they produce sudden changes in filament movement character.

Mid-circuit measurements with classical feedforward are supported on IBM’s Heron processors including Kingston, one of the advanced capabilities that makes Kingston the appropriate platform for this architecture. This approach places Phase IV at the leading edge of current quantum circuit research. It emerged organically from the compositional challenges identified in Phase III, and the novel perspectives on dynamic topology change that it opens represent a genuinely original contribution to the field — one that extends well beyond the artistic domain into quantum circuit design and research.

Post-Measurement Hadamard Reinitialization

After a mid-circuit measurement collapses a qubit to a classical state, a Hadamard gate throws it back into superposition, equal probability of |0⟩ and |1⟩. New entangling gates then establish fresh entanglement relationships with different partners. The qubit begins a new quantum evolution within the same circuit execution.

The full sequence — entangle, evolve, measure, reinitialize, re-entangle differently — applied to different qubits at different points in the sweep, produces a quantum counterpoint in which voices drop out and re-enter the texture at staggered intervals with fresh quantum characteristics. The compositional analogy is a fugal entry structure, but one governed by quantum physical laws rather than contrapuntal convention.

Hardware

IBM Kingston – Primary Platform

IBM Kingston is IBM’s most advanced currently available quantum processor, featuring 156 qubits and 340K CLOPS. 180 minutes of compute time became available following usage that exceeded the 20-minute threshold in April 2026, valid for one year from the award date. Phase IV will engage 144 qubits across 16 schemas of 9 qubits each. Kingston supports mid-circuit measurements and classical feedforward, both essential for the dynamic entanglement architecture described above.

A separate IBM Quantum Credits Program grant application requesting extended compute time to support full-scale Phase IV execution has been submitted and is pending review. The 180 minutes currently available provide the foundation for the initial simulation-validated circuit designs that will form the basis of that larger execution if the grant is awarded.

IBM ibm_fez – Secondary Platform

ibm_fez benchmarked at 4.8× faster than Quantum Inspire’s Tuna-9 in direct comparison testing conducted in February 2026, and at virtually identical execution time per circuit to Kingston in subsequent comparison. ibm_fez remains available as a secondary platform for circuit validation and comparison runs. The full hardware comparison benchmark is documented on the quantumcomputationalcreativity.com research pages and reproduced below.

Quantum Inspire Tuna-9 – Legacy Platform

Tuna-9 executed all Phase II transduction work and all Phase III tomography. It remains available as a validation and testing platform, and its characteristic hardware noise signature gives the Phase II and III material its distinctive quantum character — a reminder that hardware identity is embedded in the creative output.

Composition Concept

Phase IV envisions a single unified compositional ecosystem rather than the independent schema compositions of previous phases. Sixteen waveform sets serve as a palette of quantum-derived sonic objects from which a multi-movement or continuous composition is constructed. The dynamic entanglement architecture will produce waveforms with internal variation that no previous phase has achieved — voices that change timbral character mid-evolution as entanglement topology shifts during execution.

The visual methodology established in Phase III extends to Phase IV at dramatically larger scale. 144 individual quantum voices, each spatializable in an Ambisonics field and visualizable as animated filaments, will produce a visual environment of unprecedented complexity and internal variety. The dynamic entanglement architecture introduces a new visual dimension: filament trajectories that shift character mid-animation in direct correspondence with the mid-circuit measurement events that produce the same shifts in the audio domain. The coherence between audio and visual that was discovered organically in Phase III becomes, in Phase IV, an architectural certainty.

Preparation and Next Steps

Phase IV active development begins after the following are complete:

• Phase III visual renders for all eight schemas
• Phase III audio composition, Dance of the Qubits, finalized
• CMJ paper review outcome received and any revisions completed
• Phase IV circuit simulations validated using Qiskit Aer on clasical computer hadware.
• Dynamic entanglement architecture experiments: offset phase RZZ, mid-circuit measurements, Hadamard reinitialization
• IBM Quantum Credits Program grant decision received

I. Project Overview

Program Context

• Phase IV is the fourth phase of the Quantum Computational Creativity (QCC) research and creative praxis program.
• QCC is centered on the direct transduction of quantum hardware execution into unified audiovisual compositions.
• Phase III produced Dance of the Qubits — a nine-minute stereoscopic audiovisual composition derived from full quantum state tomography of eight entanglement schemas executed on Quantum Inspire’s Tuna-9 processor.
• Phase IV engages all 156 qubits of IBM Kingston simultaneously — the specific circuit architecture has been empirically determined through the calibration suite.

Phase IV Goals

• Develop a unified audiovisual composition derived from IBM Kingston hardware execution.
• Introduce dynamic entanglement architecture via classical feedforward and mid-circuit measurements (MCMs).
• Establish a rigorous empirical foundation through a calibration suite before finalizing circuit designs.
• Advance the QCC transduction methodology into new compositional territory.
• Explore direct transduction at the audio sample level — quantum data as audio samples rather than compositional parameters.

Infrastructure Summary

• IBM Kingston — Heron r2 processor, 156 qubits, 340K CLOPS, native gate set: CZ, RZ, SX, X, id, measure, if_else, store.
• Compute nodes DS1 and DS2 — identical Fedora 44 Server headless machines, RTX 4090, Ryzen 9 16-core, 64GB RAM, 1.8TB NVMe.
  • Python 3.12 virtual environment at ~/qcc-env
  • Qiskit 2.4.2, qiskit-aer 0.17.2 (CPU backend), qiskit-ibm-runtime 0.47.0
  • JupyterLab 4.6.1 running as systemd service — DS1 port 8888, DS2 port 8889
  • Notebook directory: /home/mrhoades/qcc/phase_iv/notebooks/calibration
• Workstation — Qiskit 2.1.2, qiskit-aer 0.17.2, qiskit-ibm-runtime 0.47.0, RTX 4060.
• Compute budget — approximately 180 minutes IBM Kingston execution time remaining not including a grant for 50 minutes more.

II. Architectural Decisions

Circuit Architecture — Validated

• The 52-triplet architecture is the validated foundation for Phase IV compositional circuits.
• 156 qubits partitioned into 52 non-overlapping triplets of 3 qubits each.
• Middle qubit of each triplet is measured (Q1, Q4, Q7… Q154). Outcome 0 branches left with RY. Outcome 1 branches right with RY.
• No baseline chain entanglement — all operations are MCM-driven and isolated within triplets.
• This architecture succeeds at 156 qubits within IBM’s compiler tolerance.

What Works at 156 Qubits

• Chain entanglement (155 RZZ) with NO MCMs — succeeds, ~10 minutes execution.
• 52-triplet MCM-only architecture (52 MCMs, no baseline entanglement) — succeeds, ~8 minutes execution.

What Fails at 156 Qubits

• Chain entanglement combined with 40+ MCMs — IBM internal compiler error 6053.
• The overlap between deterministic chain entanglement and conditional MCM re-entanglement on the same qubits is the confirmed cause.
• Untested: chain entanglement with small numbers of MCMs (5-10). This boundary has not been empirically established.

Key Paradigm Shift — Depth vs. Breadth

• Critical discovery: Kingston’s execution time is governed by circuit depth (sequential gate layers), not circuit breadth (qubit count).
• Adding qubits does not increase execution time — the processor executes all qubits in parallel simultaneously.
• T01 through T07 all completed in 8-10 minutes regardless of qubit count (13, 26, or 156).
• Aer simulation does NOT share this property — scales exponentially with qubit count. Hardware-only testing required for circuits beyond 13 qubits.

Direct Transduction at Sample Rate — Emerging Concept

• A 10-minute composition at 48kHz requires 28,800,000 audio samples.
• At 156 qubits with 3 Bloch coordinates, one parameter step yields 468 data points.
• ~61,500 steps would be needed to approach sample-rate density — feasible in principle given Kingston’s flat execution time scaling.
• This concept moves QCC from using quantum data as compositional parameters toward quantum hardware execution as the composition itself.
• Not yet ready for implementation — requires architectural design and data management planning.

Quantum Automaton — Emerging Concept

• Circuits where qubit roles are fluid — qubits shift between measured and target roles across successive MCM rounds.
• Multiple MCM rounds on same qubits create circuits with accumulated quantum history.
• The entanglement topology emerges from successive quantum decisions made by the hardware itself.
• Bifurcations cascade — each measurement outcome opens a different path, subsequent measurements act on that path’s state.
• Requires empirical grounding before implementation.

MCM Placement as Compositional Variable

• MCM placement can be conditioned on accumulated shot statistics — the measurement fires when the quantum state meets a defined threshold.
• This removes the composer’s direct control over when collapse occurs — the quantum state itself determines the moment.
• Discussed but not yet implemented.

3D Waveform / MAX Patch Architecture — Emerging Concept

• Bloch sphere x, y, z coordinates as three simultaneous audio routing dimensions rather than collapsed to a single waveform.
• MAX patch with parallel processing paths: waveform generation, temporal presentation, spatialization, etc.
• X, Y, Z scaled and routed to VBAP spatial coordinates directly — quantum state geometry becomes perceptual space geometry.
• In development — conceptual stage only.

Key Technical Decisions

• CPU Aer chosen over aer-gpu — GPU package stuck at 0.15.1, incompatible with Qiskit 2.4.2.
• Hardware-only testing adopted for circuits beyond 13 qubits — Aer scaling is exponential and impractical.
• Classical feedforward via if_else and store instructions central to Phase IV branching architecture.
• Decoherence treated as a compositional variable — not noise to be corrected.
• Daily IBM Kingston calibration data consulted before hardware submissions.
• Transpilation to native gates is an explicit pipeline step — logical depth and native depth tracked separately.

Qubit Selection as Compositional Strategy

• IBM Kingston publishes per-qubit calibration data daily — T1, T2, gate error rates, measurement error rates.
• T04 revealed physical qubit 5 has measurement error rate of 9.937E-2 — producing anomalous Bloch radius values of 213% of Aer prediction.
• T07 anomalous qubits (Q112, Q120, Q121, Q130, Q148) map directly to calibration-flagged problem qubits — Q121 had T1 of only 5.03 microseconds on day of execution.
• High measurement error introduces physically determined randomness — a distinctive compositional character.
• The 52-triplet architecture functions as a diagnostic instrument — its symmetry makes hardware characteristics directly readable in the output data.
• Every hardware characteristic is compositional fodder — nothing is waste.

Philosophical Framework

• The circuit design defines the intended landscape; hardware physical imperfections shape what actually emerges.
• The quantum computer as instrument: the circuit is a score the hardware interprets. The interpretation carries the character of the instrument.
• QCC methodology is built around dialectical thinking — outliers are compositionally significant, not statistical noise to be eliminated.
• The composer designs circumstances conducive to emergent quanta — what Kingston produces within those circumstances is the emergence.
• As circuit complexity grows, classical simulation becomes exponentially more expensive while hardware execution remains constant — these circuits are genuinely made for quantum computing.
• The growing gap between Aer simulation time and hardware execution time is a direct measure of how deeply into genuinely quantum territory the circuits are reaching.
• Hardware randomness is physically determined, bounded by quantum law, and compositionally irreplaceable.
• Every hardware characteristic is compositional fodder — nothing is waste.

III. Compositionally Relevant Parameters — Reference List

The following parameters have been identified as compositionally relevant in Phase IV. Each sits on a spectrum from fully controllable (designed into the circuit) to fully emergent (contributed by the hardware). This list will expand as the work develops.

Circuit-Level Parameters (Controllable)

• Entanglement topology — the geometric pattern of qubit connections. Determines correlation structure of Bloch sphere trajectories.
• Qubit count — number of qubits engaged per circuit. Does not affect execution time on Kingston due to parallel architecture.
• Mid-circuit measurement count — number of measurement events within a single circuit execution. Nearly free in hardware budget terms.
• Mid-circuit measurement placement — where in the circuit measurements occur. Determines what quantum state is measured and what entanglement history precedes the measurement.
• Branch complexity — the scope of qubit operations triggered by each measurement outcome.
• Branch consistency — whether outcome paths are structurally identical or different.
• Branching conflicts — deliberately targeting the same qubit from multiple measured qubits. Produces compounded entanglement pressure.
• Temporal MCM stacking — multiple measurement events on the same qubits at different circuit depths. Creates circuits with accumulated quantum history.
• Fluid qubit roles — designing circuits where qubits shift between measured and target roles across successive MCM rounds.
• Parameter sweep profile — range and distribution of theta across steps. Shapes temporal evolution.
• Shot count — number of repeated circuit executions per step. Affects statistical convergence.
• Parameter step count — number of theta steps. Primary driver of execution time and data volume.
• Native gate selection — designing circuits using Kingston’s native gates directly minimizes transpilation overhead.

Hardware-Level Parameters (Emergent / Selectable)

• T1 relaxation time — per-qubit resistance to amplitude damping. Published daily by IBM.
• T2 dephasing time — per-qubit coherence time. Published daily by IBM.
• Single-qubit gate error rate — probability of error per single-qubit operation. Median ~3E-4 on Kingston.
• Two-qubit (CZ) gate error rate — probability of error per entangling operation. Median ~2E-3 on Kingston.
• Measurement error rate — probability of readout misclassification per qubit. Highly variable — a high-error qubit introduces physically determined randomness into the transduced data.
• Per-qubit variation across the chip — T06_NoMCM revealed Bloch radius range of 0.2819 to 0.8101 across 156 physical qubits, reflecting Kingston’s natural timbral landscape.
• Crosstalk — unwanted coupling between adjacent qubits.
• Daily calibration drift — all parameters change daily. Each execution day produces a unique hardware signature.
• Physical qubit mapping — which physical qubits correspond to logical qubits. Default transpiler mapping may be overridden for compositional purposes.
• IBM internal compiler ceiling — circuits combining dense baseline entanglement with 40+ MCMs exceed the compiler’s tolerance at 156 qubits (error 6053). MCM-only circuits with no baseline chain entanglement are within tolerance.

IV. Test Log

T01 — Baseline (13 qubits, no MCM)

• Circuit: 13 qubits, nearest-neighbor chain RZZ, Hadamard, theta 0-2pi, 600 steps, 1024 shots, three-basis tomography.
• Aer: 89.84s. Transpilation: 7.74s. Native depth: 53/112/145/136.
• Kingston job ID: d936g78oamcc73dbrkf0. Execution: 8m 12s (492s).
• Aer mean R: 0.5291. HW mean R: 0.5100. Correlation: 96.4%.
• Key: Endpoint qubits show highest purity. 50-instruction decoherence limit is outdated.

T02 — Full 13-Qubit Entanglement

• Circuit: 13 qubits, full connectivity (78 RZZ gates), no MCM.
• Aer: 87.91s. Transpilation: 13.48s. Native depth: 489/~1080-1199.
• Kingston job ID: d93hlka47v0s73820pcg. Execution: 8m 41s (521s).
• Aer mean R: 0.2551. HW mean R: 0.2883. Correlation: 113% — hardware outperformed simulation.
• Key: Gate errors prevent full entanglement realization, preserving more purity. Kingston execution only 29s longer than T01 despite 8x native instruction increase.

T03 — Single Mid-Circuit Measurement

• Circuit: 13 qubits, chain, 1 MCM on Q6, minimal branch.
• Aer: 223.12s. Transpilation: 8.98s. Native depth: 53/113/146/137.
• Kingston job ID: d93m7nq47v0s738261i0. Execution: 8m 46s (526s).
• Aer mean R: 0.4760. HW mean R: 0.4551. Correlation: 95.6%.
• Key: Single MCM adds only 1 native instruction and no measurable Kingston runtime cost.

T04 — Four Mid-Circuit Measurements

• Circuit: 13 qubits, chain, 4 MCMs on Q2/Q5/Q8/Q11, clean separation.
• Aer: 253.78s. Transpilation: 8.98s. Native depth: 56/116/155/142.
• Kingston job ID: d944mv6vtlqs73ftsc3g. Execution: 8m 19s (499s).
• Aer mean R: 0.3225. HW mean R: 0.3046. Correlation: 94.5%.
• Key: Q5 anomaly at 213% — physical qubit 5 measurement error 9.937E-2. ~1 native instruction per MCM confirmed.

T04a — 26-Qubit Aer Scaling Test

• Circuit: 26 qubits, chain, 8 MCMs. 50 steps only.
• Abandoned after 8+ hours — DS1 could not complete 10 steps. Confirmed Aer exponential scaling wall.

T05 — 26 Qubits, 8 Mid-Circuit Measurements

• Circuit: 26 qubits, chain, 8 MCMs on Q3/Q6/Q9/Q12/Q15/Q18/Q21/Q24, clean separation.
• No Aer simulation. Transpilation: 12.82s. Native depth: 237-315.
• Kingston job ID: d954sqcql68s73ca79m0. Execution: 8m 25s (505s).
• HW mean R: 0.2912.
• Key: Execution time identical to 13q tests — confirms Kingston parallel architecture. Estimated execution time: 6m 47s (same as all previous tests).

T06 Series — 156-Qubit Testing

• T06 (51 MCMs + chain): Failed twice — IBM compiler error 6053. Job IDs: d9565qlgc6cc73ffr2ng, d956sk4ql68s73ca9g0g.
• T06_40MCM (40 MCMs + chain): Failed — compiler error 6053. Job ID: d95759sql68s73ca9peg.
• T06_NoMCM (156q, chain, NO MCMs): SUCCEEDED. 10m 23s (623s). Job ID: d957a4lgc6cc73ffsc70.
  • HW mean R: 0.4454. Per-qubit range: 0.2819 (Q61) to 0.8101 (Q32).
  • Confirms: 156-qubit chain without MCMs within compiler tolerance. Chain+MCM combination causes failures.

T07 — 52-Triplet Architecture (156 qubits, 52 MCMs, no baseline entanglement)

• Circuit: 156 qubits, Hadamard only, NO baseline chain entanglement. 52 MCMs on Q1/Q4/Q7…Q154. Each triplet isolated. Simple RY branch operations only. 600 steps, 1024 shots, three-basis tomography.
• Transpilation: 36.71s. Native depth: 676-1144. 208 measure instructions (156 final + 52 mid-circuit).
• Kingston job ID: d95fo4sql68s73cair30. Execution: 8m 14s (494s).
• HW mean R: 0.4376. Measured qubits: consistently 0.05-0.08 (near-zero purity from collapse). Non-measured qubits: consistently 0.61-0.63 (high purity, undisturbed).
• Anomalous measured qubits: Q70 (0.1544), Q94 (0.1554), Q112 (0.4504), Q121 (0.5796), Q130 (0.4733), Q148 (0.3684).
• Anomalous non-measured: Q120 (0.4040), Q146 (0.0497).
• All anomalies map directly to problem qubits in today’s calibration data — Q121 had T1 of only 5.03 microseconds. Q112, Q120, Q121, Q131 had missing or severely degraded gate error values.
• Key: Compiler ceiling confirmed — MCM-only architecture without baseline chain entanglement succeeds at 156 qubits. The 52-triplet architecture functions as a diagnostic instrument of Kingston’s current hardware health state. Symmetric design makes hardware characteristics directly readable in output data. 843,600 data points per execution (600 rows × 1,406 columns).

V. Ongoing Questions and Open Issues

• First compositional circuit design — ready to proceed. Key question: what entanglement to add within or between triplets while staying within compiler tolerance.
• Compiler ceiling for chain+MCM combination at 156 qubits — specific threshold (5-10 MCMs?) not yet empirically established.
• Temporal MCM stacking — multiple measurement rounds on same qubits within a single circuit.
• Fluid qubit roles — qubits shifting between measured and target roles across successive MCM rounds.
• Direct transduction at sample rate — architectural and data management questions remain open.
• 3D waveform / MAX patch architecture — conceptual stage, not yet implemented.
• Deliberate branching conflicts as compositional tool — not yet tested.
• Optimal qubit selection strategy incorporating daily calibration data.
• Whether to install Fedora on Compute-0-0 NVMe for aer-gpu — less relevant now; hardware-only testing is primary mode.

VI. Key Findings — Consolidated

• Kingston executes in parallel — depth not breadth drives execution time. Flat 8-10 minutes across 13, 26, and 156 qubits.
• Aer simulation scales exponentially with qubit count — impractical beyond 13 qubits on DS1.
• The growing gap between Aer simulation time and hardware execution time measures how deeply into genuinely quantum territory the circuits are reaching.
• Single MCM adds only one native instruction and no measurable hardware execution time overhead.
• MCM scaling highly favorable — ~1 native instruction per measurement regardless of count.
• Classical feedforward executes cleanly on Kingston’s Heron r2 processor.
• Chain entanglement + 40+ MCMs at 156 qubits causes IBM compiler error 6053.
• Solution: MCM-only architecture with no baseline chain entanglement succeeds at full chip scale.
• 156-qubit chain without MCMs succeeds — mean R 0.4454, remarkable per-qubit variation 0.2819 to 0.8101 reflecting Kingston’s natural timbral landscape.
• 52-triplet architecture validated — 8m 14s, clean pattern of collapsed measured qubits against high-purity non-measured qubits.
• T07 anomalous qubits map directly to daily calibration data — the architecture functions as a hardware diagnostic instrument.
• High measurement error rate produces anomalous Bloch radius values — a measurement artifact and a distinctive compositional character.
• Daily IBM calibration data enables deliberate qubit selection based on error personality — a compositional tool.
• 156-qubit executions produce 843,600 data points — data management at this scale is a significant open question for the compositional pipeline.
• Hardware randomness is physically determined, bounded by quantum law, and compositionally irreplaceable.
• Every hardware characteristic is compositional fodder — nothing is waste.
• The project is ready to transition from calibration to compositional circuit design.