AI-Driven Dimension Scalable Quantum State Tomography on Edge Devices

Quantum state tomography (QST) is a fundamental tool for characterizing quantum systems, but traditional approaches are computationally intensive for real-time applications. This project presents an Artificial Intelligence (AI)-assisted framework for dimension-adaptive quantum state tomography using edge computing hardware, specifically the Field Programmable Gate Arrays (FPGAs). A machine learning model trained to perform quantum state reconstruction from measurement data is deployed on the FPGA, leveraging its parallel processing capabilities for low-latency inference. The model undergoes high-level synthesis to convert its high-level Python implementation into hardware-compatible logic, enabling seamless deployment on the FPGA fabric. Trained on “m” qubits, this system can reconstruct quantum states for “n” qubit systems where “m > n”, without requiring the training model's dimension to match that of the target system. By utilizing the fidelity's non-decreasing property, we relate the average reconstruction fidelity of “m”-qubit systems to lower-dimensional “n”-qubit systems. With a model trained on four qubits, we reconstruct quantum states for randomly selected one, two, and three-qubit systems. This edge-accelerated approach may significantly reduce computational overhead and energy consumption, demonstrating the feasibility of integrating machine learning-based QST into resource-constrained, real-time quantum diagnostic systems. The results are expected to validate the effectiveness of combining AI with reconfigurable hardware for scalable and efficient dimension aware quantum state characterization.