Los Alamos and QC Ware model tricky iron oxide molecules on IBM quantum hardware

[Velvet Connor]

Iron oxide sounds ordinary, but it has long embarrassed standard chemistry simulations. Quantum Computing Report covers a preprint from Los Alamos National Laboratory and QC Ware that uses IBM's ibm_miami processor to model small iron oxide systems as a stepping stone towards bulk FeO

Bulk FeO is a Mott insulator, yet density functional theory, the workhorse of computational chemistry, wrongly predicts it should conduct electricity. The reason is strong correlation between electrons, which classical approximations handle badly

The team used a hybrid method called sample based quantum diagonalization. FeO was mapped onto 16 qubits and the Fe2O2 dimer onto 32, with several copies run in parallel across 96 qubits. The quantum processor sampled bitstrings, and classical computers then diagonalised the problem in the smaller space those samples defined

For FeO, the method reached chemical accuracy within a 184 determinant subspace, where a plain variational approach fell short. The Fe2O2 dimer also reached chemical accuracy after a one shell expansion. The predicted FeO excitation energy came out at about 2,167 wavenumbers, against roughly 300 from earlier classical baselines, which the authors say matches experiment

This is a preprint and the quantum chip mainly acted as a sampler, with the circuit parameters tuned in classical simulation. The source is also slightly inconsistent about which state was recovered. Even so, using quantum hardware to attack a famous failure of classical methods is exactly the right kind of target

Are strongly correlated materials the best early use for quantum computers? And how much of the credit belongs to the quantum part of these hybrid methods?

Los Alamos and QC Ware Model Correlated Iron Oxides Using Sample-Based Quantum Diagonalization on IBM Hardware - Quantum Computing Reportquantumcomputingreport.com