Table 9 An example of the types of quantum data features which may be included in a dedicated large-scale dataset for QML.
Item | Description |
|---|---|
Quantum states | Description of states in computational basis, usually represented as vector or matrix (for ρ). May include initial and evolved (intermediate or final) states |
Measurement operators | Measurement operators used to generate measurements, description of POVM. |
Measurement distribution | Distribution of measurement outcome of measurement operators, either the individual measurement outcomes or some average (the QDataSet is an average over noise realisations). |
Hamiltonians | Description of Hamiltonians, which may include system, drift, environment etc Hamiltonians. Hamiltonians should also include relevant control functions (if applicable). |
Gates and operators | Descriptions of gate sequences (circuits) in terms of unitaries (or other operators). The representation of circuits will vary depending on the datasets and use case, but ideally quantum circuits should be represented in a way easily translatable across common quantum programming languages and integrable into common machine learning platforms (e.g. TensorFlow, PyTorch). |
Noise | Description of noise, either via measurement statistics, known features of noise, device specifications. |
Controls | Specification and description of the controls available to act on the quantum system. |
