Periodic Symmetry-Adapted Encoding: Qubit Reduction in Crystalline Electronic Structure

Abstract:We extend the symmetry-adapted encoding (SAE) framework to periodic electronic structure, enabling qubit-efficient quantum simulation of crystalline materials. By constructing a $\Gamma$-point supercell Hamiltonian from a folded $k$-point calculation and systematically identifying all applicable space-group symmetry generators -- including spin-parity, point-group, and crystal translation symmetries -- we obtain qubit Hamiltonians with fewer qubits than the Jordan--Wigner starting point. We benchmark diamond, silicon, 3C-SiC, MgO, NaCl, CsCl, h-BN, wurtzite AlN, $\alpha$-quartz SiO$_2$, and MgF$_2$ using active spaces chosen to preserve complete near-degenerate frontier manifolds across cubic, hexagonal, trigonal, and tetragonal space groups. Across the suite the periodic SAE removes 4--8 qubits. The B2 CsCl benchmark realises eight independent Boolean generators, i.e. a symmetry group isomorphic to $\mathbb{Z}_2^8$, reducing CAS(6,7) from 14 to 6 qubits. This exceeds the $\mathbb{Z}_2^5$ maximum of molecular SAE, where only two spin parities and at most three independent Boolean point-group generators are available, because the folded crystal supplies three additional half-translation symmetries. Noiseless UCCSD-VQE benchmarks against exact diagonalisation in the active-space sector show that the reduced encodings preserve the target energies to well below chemical accuracy while reducing variational parameter counts by $3$--$8\times$ and CNOT counts by up to $309\times$. The largest circuit savings occur when translation and point-group generators act independently in the active space, demonstrating that periodic symmetry can be converted directly into both qubit and ansatz compression. The method is implemented in the open-source QuantumSymmetry package and requires no manual specification of symmetry generators.

arXiv:2606.05777 [quant-ph]

(or arXiv:2606.05777v1 [quant-ph] for this version)

https://doi.org/10.48550/arXiv.2606.05777 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.