Toward Live Noise Fingerprinting in Quantum Software Engineering
By: Avner Bensoussan , Elena Chachkarova , Karine Even-Mendoza and more
Noise is a major bottleneck in today's quantum computing, stemming from decoherence, gate imperfections and other hardware limitations. Accurate noise fingerprints are essential, yet undocumented noise model differences between Quantum Ecosystems undermine core functionality, such as compilation, development and debugging, offering limited transferability and support for quantum software engineering (QSE) tasks. We propose a new research direction: live empirical noise fingerprinting as a lightweight QSE-oriented "noise fingerprinting". Though explored in physics as device-level diagnostics, we reposition them as a QSE paradigm: we propose leveraging classical shadow tomography to enable a new generation of techniques. As a first step, we introduce SimShadow, which prepares reference states, applies shadow-tomography-inspired estimation and constructs deviation fingerprints. Initial experiments uncover systematic discrepancies between platforms (e.g. Frobenius distances up to 7.39) at up to 2.5x10^6 lower cost than traditional methods. SimShadow opens new directions for noise-aware compilation, transpilation, cross-platform validation, error mitigation, and formal methods in QSE.
Similar Papers
Noisy Quantum Learning Theory
Quantum Physics
Makes quantum computers work better with errors.
Quantum Noise Tomography with Physics-Informed Neural Networks
Quantum Physics
Makes quantum computers work better by understanding noise.
Assumption-free fidelity bounds for hardware noise characterization
Quantum Physics
Makes quantum computers work better despite errors.