Quantum Computing Finds Hidden Molecular States

A new algorithm that blends quantum computing with classical high-throughput computing has successfully identified elusive molecular resonance states, a breakthrough that could accelerate research in fields from photocatalysis to quantum control. These resonances, which describe how molecules break apart or decay, have been notoriously difficult to calculate with classical s alone due to their complex nature and demanding computational requirements. The research, led by scientists at the University of Wisconsin-Madison, demonstrates how hybrid approaches can leverage the strengths of both quantum and classical systems to tackle problems that have long d computational chemistry.

The key finding is that the algorithm, called qDRIVE (quantum deflation resonance identification variational eigensolver), can accurately determine resonance energies and wavefunctions for a benchmark molecular system. In simulations, qDRIVE identified a bound state and two resonance states in a model of molecular predissociation, with errors as low as 0.000024% compared to exact diagonalization in noiseless conditions. Even when simulating noisy near-term quantum processors like IBM Torino, maintained reasonable accuracy, with errors around 0.91% for some states using a two-qubit approach. This shows that the algorithm can work with existing quantum hardware while promising better as quantum processors improve.

Ology cleverly combines two computational strategies. First, it uses the complex absorbing potential (CAP) formalism, which transforms the problem of finding resonances into finding eigenstates of a non-Hermitian Hamiltonian. This allows researchers to start with good initial guesses from corresponding Hermitian Hamiltonian eigenstates, which are easier to compute using established quantum algorithms like the variational quantum eigensolver (VQE). Second, it employs high-throughput computing (HTC) to run many interconnected but independent computational tasks in parallel and asynchronously, minimizing overall completion time. The approach creates a directed acyclic graph of jobs that can scavenge available computational resources efficiently, unlike traditional high-performance computing that requires large blocks of resources simultaneously.

From extensive simulations show qDRIVE's robustness across different computing environments. In statevector simulations without noise, the algorithm achieved near-perfect accuracy for all three states studied. When shot noise was added in Aer simulations, errors remained below 1% for most cases, though one resonance showed a 2.8% error with a three-qubit ansatz. Custom simulations mimicking IBM Torino's noise characteristics produced larger but still reasonable errors, demonstrating 's practicality for current noisy intermediate-scale quantum (NISQ) computers. The research also examined how improve with better hardware: as gate noise reduction factors increased from 100 to 10,000 and qubit longevity increased from 10 to infinity microseconds, both pseudovariance and fidelity errors approached zero, indicating that future quantum processors will enable even more accurate resonance identification.

Extend beyond this specific application. By successfully integrating quantum computing with HTC, qDRIVE serves as a prototype for a wider family of heterogeneous algorithms that could tackle other challenging problems in chemistry and materials science. Molecular resonances play crucial roles in processes ranging from ultracold collision complex decay to plasmonic photocatalysis, and recent predictions suggest they could impact quantum information processing and quantum control. The ability to accurately identify these states could therefore accelerate discoveries in multiple disciplines, from designing better catalysts to developing novel quantum computing architectures based on molecular qubits.

However, the approach has limitations that must be addressed for broader application. The current implementation uses grid-based mappings that become exponentially expensive for larger systems, though this could be circumvented for electronic structure problems using scalable fermion-qubit mappings like Jordan-Wigner or Bravyi-Kitaev. Additionally, while the three-layer efficient SU(2) ansatz worked well for the benchmark system, maintaining expressibility and efficiency for larger numbers of qubits may require adaptive ansatz approaches like ADAPT-VQE. The research also notes that higher-lying states showed larger errors due to the deflation procedure, suggesting that incorporating more advanced variational algorithms could improve for complex systems with many resonance states.