New method automatically finds errors in quantum states

Researchers from Toyota Central R&D Labs., Inc., and Basic Research Laboratories, NTT, Inc., detailed a new framework published September 7, 2026, for automatically identifying sources of error in quantum states. The method combines simulation and parameter optimization with quantum state tomography data to quantify how experimental factors degrade quantum information. By modeling and adjusting parameters within a state preparation simulator, the team reduced the trace distance between simulated and experimental data from 0.177 to 0.024, indicating that their modeled error sources explain 86% of the errors. The proposed framework offers a practical path to automated error attribution and calibration directly from experimental results. The framework, termed model-based quantum error quantification (MBQEQ), estimates error sources from experimentally reconstructed density matrices obtained through quantum state tomography (QST). This approach moves beyond “black box” optimizations common in quantum device calibration by providing human-interpretable physical error models, a step toward refining quantum systems. MBQEQ integrates a simulator, an evaluator, and an optimizer to translate QST data into quantifiable error parameters, allowing for a more direct understanding of experimental imperfections. The team found that accidental coincidences and measurement-basis phase errors were significant contributors to performance degradation. To validate the method, researchers intentionally introduced accidental coincidence…