This is the implementation of the paper "Quantum Multiple Rotation Averaging" by Wang et al. We introduce IQARS, a novel method for globally synchronizing multiple local rotation matrices. For details, we refer to the project page.

This repository provides a Python implementation of Quantum Multiple Rotation Averaging using QUBO formulations and annealing-based optimization (simulated or quantum).

The code is organized into modular Python files:

• lie.py — SO(3) geometry utilities and derivatives
• data.py — synthetic rotation dataset generation
• qubo.py — QUBO construction
• objective.py — residual evaluation
• solvers.py — simulated / quantum annealing interface
• main.py — experiment runner

git clone https://github.com/shutengW/Quantum-Multiple-Rotation-Averaging.git

cd Quantum-Multiple-Rotation-Averaging

conda env create --name EnvName --file environment.yaml conda activate EnvName

This installs all required dependencies for running QMRA experiments.

To run the synthetic rotation averaging experiment:

python main.py

The script will:

• generate synthetic relative rotations
• construct the QUBO problem
• solve it using annealing
• iteratively update rotations

If you use QMRA in your research, please consider citing our work:

    @inproceedings{Wang2026quantum,
                title={Quantum Multiple Rotation Averaging},
                author={Wang, Shuteng and Kuete Meli, Natacha and Möller, Michael and Golyanik, Vladislav},
                booktitle={International Conference on 3D Vision},
                year={2026}
                }