Faster Quantum Calculations Unlock Efficient Molecular Ground State Preparation

Scientists are continually seeking methods to improve the efficiency of quantum computation for complex chemical simulations. Max Haas, Thierry N. Kaldenbach, and Thomas Hammerschmidt, working with colleagues at the German Aerospace Center (DLR), Institute for Frontier Materials on Earth and in Space, and ICAMS, Ruhr University Bochum, detail a new protocol for preparing electronic ground states using the ExcitationSolve optimizer and established variational eigensolver techniques. Their research demonstrates a computationally efficient strategy that constructs approximate ground states from a unitary coupled cluster ansatz with a single sweep, significantly reducing complexity through classical pre-processing for operator selection. By integrating this approach with one-variational-parameter couple exchange operators, the team achieves a quadratic convergence speedup compared to current methods, representing a substantial step towards realising practical quantum advantage in materials science and beyond.

Researchers have developed a new computational protocol that dramatically accelerates the preparation of electronic ground states for quantum simulations, optimising the variational quantum eigensolver (VQE), a hybrid quantum-classical algorithm used to model complex quantum systems, particularly in quantum chemistry. By combining the ExcitationSolve optimizer with established operator selection methods like Energy Sorting, the team achieved a significant reduction in computational complexity, constructing an approximate ground state from a unitary coupled cluster ansatz with a single sweep over the available operators. This innovation addresses a critical bottleneck in VQE calculations, the exponential growth of computational demands as system size increases, circumventing issues with “barren plateaus” through efficient classical pre-processing that intelligently selects the most relevant operators before quantum computation begins. This pre-selection dramatically reduces the number of quantum circuits needed, paving the way for more complex simulations on near-term quantum computers and demonstrating seamless integration with one-variational-parameter couple exchange operators (OVP-CEOs), further minimising the number of CNOT operations, a key metric for quantum circuit complexity. Empirical results reveal a quadratic convergence speedup compared to state-of-the-art methods, meaning the algorithm reaches accurate solutions significantly faster, bringing the prospect of quantum advantage in quantum chemistry closer to reality. The team successfully applied this method to systems ranging from 4 to 20 qubits, demonstrating its scalability and efficiency. By effectively “warm-starting” the VQE with a classically-determined initial state, the researchers have created a powerful tool for exploring the behaviour of molecules and materials at the quantum level, promising to unlock new possibilities in fields ranging from drug discovery to materials science, enabling the design of novel compounds with tailored properties. Employing this novel computational protocol, the research demonstrates a quadratic convergence speedup in preparing electronic ground states compared to state-of-the-art methods, achieved by integrating the ExcitationSolve optimizer with energy sorting techniques and a unitary coupled cluster ansatz. The study reveals that all relevant operators for constructing an approximate ground state can be selected within a single operator selection step, streamlining the optimisation process and establishing a fully classical method for constructing and initialising the VQE. Notably, the work showcases the seamless integration of one-variational-parameter couple exchange operators (OVP-CEOs) into this protocol, with adoption of an OVP-CEO pool resulting in a reduced circuit depth, achieving a decrease from 13 CNOT gates to 9 CNOT gates per excitation operator, alongside a reduction in circuit depth from 11 to 7. While this introduces a minor computational overhead, the overall efficiency gains are substantial, and the researchers successfully applied this approach to molecules ranging from 4 to 20 qubits, consistently demonstrating improved performance in both operator count and quantum computer evaluation requirements. The efficacy of ExcitationSolve lies in its ability to reconstruct the cost function for each parameter, enabling direct optimisation without gradient descent, allowing for “free” energy sorting during the initial selection, identifying all operators contributing to the system’s energy with a single sweep of the operator pool. Furthermore, ExcitationSolve provides optimal parameter values for each operator, facilitating a warm start and eliminating the need for subsequent operator selection steps, particularly advantageous for linear systems like the LiH molecule where the relevant operators remain consistent across bond lengths. The application of ExcitationSolve to OVP-CEO+ and OVP-CEO− operators, despite presenting challenges related to ansatz depth, confirms its broad applicability and potential for further optimisation. A central innovation of this work lies in the combined application of the ExcitationSolve optimizer and Energy Sorting to construct efficient VQE ansätze, leveraging Energy Sorting to pre-select a majority of relevant operators via classical pre-processing. This initial sorting step circumvents the need for extensive quantum circuit evaluations typically associated with adaptive ansätze like ADAPT-VQE, significantly reducing computational complexity. Crucially, the ExcitationSolve optimizer was then integrated, exploiting its ability to reconstruct the cost function for each parameter and enabling a streamlined, gradient-free optimisation process. This methodology allowed for the construction of an approximate ground state, based on a unitary coupled cluster ansatz, within a single sweep over the operator pool, employing a UCCSD (Unitary Coupled Cluster Singles Doubles) ansatz and systematically evaluating its performance with varying numbers of qubits, ranging from 4 to 20. To further minimise circuit depth, the research seamlessly incorporated one-variational-parameter couple exchange operators (OVP-CEOs) into the operator pool, driven by the desire to overcome limitations inherent in traditional VQE implementations, such as barren plateaus and the exponential growth of the Hilbert space. By performing the initial energy sorting classically, the study effectively “warm-starts” the VQE, reducing the number of quantum computations required and accelerating convergence, representing a significant advancement towards achieving quantum advantage in quantum chemistry calculations. Scientists pursuing more accurate molecular simulations have long been hampered by computational cost, with traditional methods struggling to scale with molecular complexity, limiting our ability to model crucial chemical processes with sufficient precision. This work offers a significant step towards overcoming that barrier, not through brute force computing, but through clever algorithmic design, streamlining the process of finding the lowest energy state of a molecule, its electronic ground state, and demonstrating a substantial speedup in calculations. More efficient ground state preparation unlocks the potential for simulating larger, more complex molecules, and for studying dynamic processes over longer timescales, with applications ranging from designing novel materials with tailored properties to accelerating drug discovery by accurately predicting molecular interactions. The integration with one-variational-parameter coupled exchange operators further reduces the demands on quantum hardware, bringing practical quantum chemistry applications closer to reality. However, the demonstrated speedup, while substantial, is empirically observed and may not translate uniformly across all molecular systems. The reliance on classical pre-processing, while effective, introduces a potential bottleneck for extremely large molecules, and the study focuses on specific ansatzes and operator selection methods, leaving open the question of how well this approach generalizes to other quantum algorithms. Future work will likely explore extending this methodology to more diverse molecular systems and investigating its performance on actual quantum hardware, paving the way for a new generation of computational chemistry tools.