I'm a researcher in computational chemistry, and I'm exploring how near-term quantum computers could be used to simulate complex molecular interactions that are intractable for classical computers. I've been studying variational quantum algorithms like VQE, but the practical implementation seems daunting given current hardware limitations like noise and qubit connectivity. For others working at this intersection of quantum computing and applied science, what are the most promising real-world problem classes that could see a quantum advantage in the next five to ten years, and what software frameworks or cloud-based quantum processors are you using to prototype these algorithms today?
Real-world problem classes with potential quantum advantage in the 5–10 year horizon include a few that map reasonably well to near-term quantum hardware. First, small, strongly correlated molecules where active-space methods are essential—think transition-metal complexes or bond-dissociation problems (H2O, FeS models, etc.). Embedding approaches like DMET or quantum subspace methods can let a quantum solver tackle the hard, multi-reference part while a classical solver handles the rest. Second, reaction energetics and potential-energy surfaces along a reaction coordinate requiring accurate multi-state treatment; this naturally invites excited-state chemistry via quantum subspace expansions or VQE variants. Third, lattice-like models of catalytic active sites or correlated materials (Hubbard-type/Anderson models) to capture collective electron phenomena that are hard for CCSD(T)-level methods. Lastly, quantum dynamics tasks such as real-time propagation or spectral functions where linear-response or time-dependent variants could potentially show a speedup with fault-tolerant devices, though that’s further out. In practice, expect the near-term wins to come from combining quantum solvers for a small active space with robust embedding, not full-system, black-box chemistry on noisy devices.