BornMachineTomo

Data for Quantum Tomography with Born Machine, in support of our work arXiv:1712.03213

File Organization

• Definitions of the classes ./CS6.py
• Efficiency experiments on typical states ./trial8-Efficiency/
  • main.py conducts the experiments
    • Using measurement outcomes in ./MeasOutcomes/
  • Resulting fidelity sequences are in the folders [typ]/[N]/
  • sat_persite.py postprocess the fidelity sequeces to analyze what if we set the per-site fidelity as criterion.
• Efficiency experiments on random states ./trial14-RandTarget/
  • main.py conducts the experiments
    • Using measurement outcomes in ./MeasOutcomes/Random/
  • Resulting fidelity sequences are in the folders [Dmax]/[N]/
• Demonstration of Fidelity Estimation ./trial13-FidEstL249
  • prep.py prepares measurement outcomes from the virtual target states stored in ./trial8-Efficiency/[type]/[N]/R[seed]/L249/ and ./trial9-randomTarget/random/[N]/[seed]/R[seed]/L249/ and stores the outcomes in vir_measout/
  • main.py conducts the experiments
    • Using measurement outcomes in vir_measout/
  • Resulting fidelity sequences are in the folders [type]/[N]/
• Robustness Experiments ./errRobust/
  • prep.py prepares measurement outcomes from the noised target states $\sigma_\epsilon = (1-\epsilon)\sigma + \frac{\epsilon}{q^N}\mathrm{I}$ and stores them in ./errRobust/MeasOutcomes/
  • main.py conducts the experiments
    • Using measurement outcomes in ./errRobust/MeasOutcomes/[type]/[N]/[noise]/
  • Resulting fidelity sequences are in the folders [type][N]/[noise]/
  • elist.npy includes the values of noisy level we considered.
• ./MeasOutcomes/ includes raw outcomes of simulated measurements.
  • prep.py measures the typical states and stores the results in:
  • [type]/[N]/R[seed]Set.pickle, which pickles the outcomes from the state of [type] and length [N] in the random case initiated by [seed], the state being stored as [type]/[N]/stdmps.pickle
  • Random/
    • prep_Rand.py measures the random states and stores the results in:
    • [Dmax]/[N]/[seed]/R[seed]Set.pickle, which pickles the outcomes from the random state (by [seed]) whose Dmax is [Dmax] and length is [N] in the random case initiated by [seed], the state being stored as [type]/[N]/[seed]/stdmps.pickle
• ./WorkSpace.ipynb is the Jupyter Notebook where the results are analyzed and plotted

Note

• Due to the big volume of the outcomes, we only uploaded some of the training sets (measurement outcomes) we used in the numerical experiments mentioned in our manuscript, yet the scripts prep*.py for each experiment suffice the generation of all the training data.
• Of course one could iteratively measure and train, which is a more faithful simulation of the process of our scheme. We store these measurement outcomes in advance simply because in this way:
  1. They can be reused.
  2. Comparison between different hyper parameter configurations become reasonable.