cometta / quantum_search_ansatz Goto Github PK

Find the best ansatz for ground state energy using genetic algorithm. Able to run on any Qiskit V1 fake device with and without noise. Capable to search bigger hibert space compared to original implemention on the paper. Tested up to 12 qubits Hamiltonian run on the 16 qubits device.

• install below packages

  sudo apt install cmake
  sudo apt-get install libopenblas-dev
  

• create python 3.11 environment

• pip install -r requirements.txt

• add your problem in the directory hamiltonian/<new problem>.py

• add your problem config file in the directory conf/search/<new problem>.yaml

• edit config.yaml file, change the search value based on your problem

• run the search by python main.py

• output are stored at outputs/<date>/<time>/*

  • circuit_selected.qpy Ansatz found is stored in Qiskit binary serializer format
  • nas_result_sorted.txt Best result is on the top, sorted in the format [subnet] (score, electric ground state value)
  • param_selected.txt best parameters found

```
from qiskit import qpy
ansatz_custom = qpy.load(open(os.path.join('<path>/circuit_selected.qpy'), 'rb'))[0] 
```

• hamiltonian qubits < device qubits supported, transpile_q_layout can be None or List integer

• hamiltonian qubits = device qubits in config file, need to fill in transpile_q_layout

• hamiltonian qubits = device qubits and transpile_q_layout = None known issue, does not supported

@article{du2020quantum,
  title={Quantum circuit architecture search: error mitigation and trainability enhancement for variational quantum solvers},
  author={Du, Yuxuan and Huang, Tao and You, Shan and Hsieh, Min-Hsiu and Tao, Dacheng},
  journal={arXiv preprint arXiv:2010.10217},
  year={2020}
}

fork of https://github.com/yuxuan-du/Quantum_architecture_search