Scientific Deep Dives & Research Notes
Open science requires publishing positive demonstrations, negative findings, and mathematical audits with complete transparency across all 14 physical campaigns.
Q-Psi After 564,840 Physical Quantum Shots
A comprehensive retrospective of 14 audited physical-QPU campaigns on IBM Heron: bounded query advantages, qubit reuse, Data Foundry feasibility, negative baseline comparators, and strict scientific boundaries.
Reusing Physical Qubits with Compiler Liveness Analysis on IBM Heron
Analysis of physical qubit reuse passes on IBM Heron: mapping 8 to 16 logical qubits onto 2 to 3 physical slots while maintaining non-inferior observable fidelity across 6/6 tested cases.
127 Quantum Seconds, 65,138 Candidate States and 93 Verified Coding Records
The audited results of Quantum Data Foundry V2: 332,768 physical shots, 65k sampled candidate states, 93 verified positive repair records, 100% QPU provenance, and the classical verification throughput bottleneck.
Q-Psi vs Qiskit and TKET: What the MQT Benchmark Actually Showed
Full disclosure of the independent MQT benchmark: Qiskit held a small aggregate advantage (-0.0067), while Q-Psi's apparent advantage over TKET was driven by a single routing-sensitive workload.
Negative Results Matter: Where Q-Psi Did Not Beat the Baseline
Examining our negative and inconclusive studies: why Qiskit multi-seed layout outperformed Q-Psi, why fractional pulse lowering gave mixed signals, and why Directed Execution was inconclusive.
Why Lower Circuit Depth Did Not Guarantee Better Physical Fidelity
How dynamic circuits compressed two-qubit gate depth from 73 to 2 on 13 hops, but reset and measurement noise prevented physical fidelity crossover over unitary routing.