In a world where the accuracy of quantum-chemical calculations determines the future of catalysts and biological processes, a team from IBM, RIKEN and NVIDIA has made a notable step forward. Using a 72-qubit quantum processor and a powerful supercomputer, the researchers substantially improved the accuracy of energy calculations for the iron-sulfur cluster Fe4S4.
The object of study is the model complex [Fe4S4(SCH3)4]²⁻, a key element of ferredoxins and nitrogenase. These clusters participate in electron transfer in living organisms, and their electronic states are extremely complex due to the strong correlation of many electrons. The active space included 54 electrons and 36 orbitals, which makes the calculations a serious challenge for classical methods.
The scientists applied a hybrid method, Sample-Based Quantum Diagonalization (SQD). The ibm_kingston quantum processor collected samples of electronic configurations, while the ROQUO supercomputer with NVIDIA GPUs performed the subsequent processing and diagonalization. The main innovation was the use of broken spin symmetry as the initial state (BS-UHF), which made it possible to account for the magnetic interactions between iron atoms right away.
The result is impressive: the best direct energy turned out to be only 4,3 millihartree above the value obtained by the DMRG method with a bond dimension of 4000. This is roughly 79 times closer to the reference than in previous experiments. Extrapolation to infinite space gave a value that practically coincides with DMRG within the margin of error.
The achievement shows how important a high-quality starting point is for quantum-classical hybrid algorithms. Even with a limited number of qubits, the right choice of initial state and powerful classical post-processing make it possible to approach chemical accuracy in complex systems.
The work opens the way to more accurate modeling of enzymes and new materials. The next steps are to transfer the methodology to calculations of reaction energies and redox potentials, where accuracy is especially critical for practical applications.
