This is a data repository for the paper titled:

Chemistry Beyond the Scale of Exact Diagonalization on a Quantum-Centric Supercomputer.

This repository contains for all molecules considered in the study:

• Energy approximations (and exact when possible) obtained from classical many-body methods: /classical_reference_methods.

• The electronic integrals for all molecules considered in fcidump format link.

• The raw experiment data and results in /experiments:

  • Raw dictionaries of samples together with a python script to lead them: /experiments/{molecule name}/experiment_data.zip.
  • Energies (and variances when available) obtained from the quantum experiments using sqd: /experiments/{molecule name}/sqd_hardware_energetics.
  • Template script to generate the qiskit circuits run on the experiments. The circuits were generated using the python package ffsim.
  • Energies obtained by running SQD on samples obtained from the uniform distribution: /experiments/{molecule name}/sqd_on_uniform_distribution.

• Numerics on noiseless performance of SQD: /numerics

The SQD implementation can be found in the python package qiskit-addon-sqd. It is an open source project version-controlled on GitHub. In the manuscript we used the 0.1.0 version of the package. Note that all versions of the package are available on both GitHub and PyPI.

This tutorial implements the same SQD workflow as in the manuscript.

Note: The bond distances for the N2 molecule are given in fractions of 1.0975135 Angstroms. So to convert the units given here to Angstroms, divide by 1.0975135.