The Unitary Foundation 2025 Quantum Open Source Software Survey is now open. Share your voice before it closes on October 3, 2025! Take the survey!
Data for benchmarking machine learning models, taken from Train on classical, deploy on quantum: scaling generative quantum machine learning to a thousand qubits. This dataset contains bit strings sampled from a thermal distributions of an Ising Hamiltonian that has a connectivity given by a scale-free network with 1000 nodes.
Description of the dataset
The dataset contains bit strings of length 1000 that correspond to spin configurations of the Ising distribution. The graph describing the two-body interactions of the Ising Hamiltonian corresponds to a scale free network, which is constructed via the Barabasi-Albert algorithm with connectivity parameter 2. The Ising energy has a local bias that is dependent on the degree of the corresponding node, which biases the values of that bit towards zero. The data was generated via a Metropolis Hastings algorithm by sampling a million configurations on eight independent Markov chains, and then selecting 20000 train and test points randomly from equally spaced points on the chain.
Example usage
>>> [ds] = qml.data.load("other", name="scale-free")
>>>
>>> np.shape(ds.train)
(20000, 1000)
>>> np.shape(ds.test)
(20000, 1000)
Joseph Bowles
version 0.1 : initial public release
Joseph Bowles
Quantum Machine Learning researcher at Xanadu