High Resolution Image Download MS PowerPoint Slide Strongly correlated molecules remain out of reach for most electronic-structure solvers because the exact wave function spans a determinant space that grows exponentially with the number of orbitals. State-of-the-art selected-CI (e.g., CIPSI and HCI) and ML-CI methods mitigate search by exploiting Hamiltonian sparsity and learned prescreening. Nonetheless, in strongly multireference regimes, the excitation-based candidate lists can grow rapidly, which may increase memory and screening costs. We present the Hamiltonian-guided autoregressive selected-configuration interaction (HAAR-SCI), a learn–sample–compress workflow that runs on a single GPU. A gated Transformer samples determinants autoregressively; Gumbel Top-K noise encourages exploration, and GPU min-heap kernels keep only configurations with the largest Hamiltonian couplings. The network is retrained after each expansion, and iterations stop when successive energies differ by ≤1 mHa. Across an 18-molecule benchmark set reaching 116 spin–orbitals, HAAR-SCI attains a mean absolute error of 0.51 mHa while using much fewer determinants than heat-bath CI. It traces the entire N 2 dissociation curve within 0.67 mHa and achieves HCI accuracy for the 40-spin-orbital [Fe 2 S 2 (SCH 3 ) 4 ] 2– cluster with a determinant reduction of 72%, demonstrating its power on systems considered intractable for conventional selected-CI solvers. A final probability-mass pruning compresses typical wave functions by a further 10–50×, retaining <0.01% of the Hilbert space yet still capturing >99.9% of the correlation energy. HAAR-SCI thus offers a compact and truly scalable route to chemical-accuracy quantum chemistry on commodity hardware.
High Resolution Image Download MS PowerPoint Slide Strongly correlated molecules remain out of reach for most electronic-structure solvers because the exact wave function spans a determinant space that grows exponentially with the number of orbitals. State-of-the-art selected-CI (e.g., CIPSI and HCI) and ML-CI methods mitigate search by exploiting Hamiltonian sparsity and learned prescreening. Nonetheless, in strongly multireference regimes, the excitation-based candidate lists can grow rapidly, which may increase memory and screening costs. We present the Hamiltonian-guided autoregressive selected-configuration interaction (HAAR-SCI), a learn–sample–compress workflow that runs on a single GPU. A gated Transformer samples determinants autoregressively; Gumbel Top-K noise encourages exploration, and GPU min-heap kernels keep only configurations with the largest Hamiltonian couplings. The network is retrained after each expansion, and iterations stop when successive energies differ by ≤1 mHa. Across an 18-molecule benchmark set reaching 116 spin–orbitals, HAAR-SCI attains a mean absolute error of 0.51 mHa while using much fewer determinants than heat-bath CI. It traces the entire N 2 dissociation curve within 0.67 mHa and achieves HCI accuracy for the 40-spin-orbital [Fe 2 S 2 (SCH 3 ) 4 ] 2– cluster with a determinant reduction of 72%, demonstrating its power on systems considered intractable for conventional selected-CI solvers. A final probability-mass pruning compresses typical wave functions by a further 10–50×, retaining <0.01% of the Hilbert space yet still capturing >99.9% of the correlation energy. HAAR-SCI thus offers a compact and truly scalable route to chemical-accuracy quantum chemistry on commodity hardware.
Accurate description of electron correlation is a central challenge in computational chemistry, with selected configuration interaction (SCI) emerging as a powerful tool to approach the full CI limit. While recent machine learning (ML) integrations have accelerated determinant selection, existing regression and classification approaches suffer from a fundamental objective-loss mismatch: they evaluate the importance of determinants in isolation without explicitly accounting for their relative importance ranking. Here, we introduce ranking configuration interaction (RCI), a novel ML-supported SCI framework that reframes determinant selection as a pairwise ranking problem. Building upon a Transformer-based architecture to capture complex, nonlocal orbital dependencies, RCI progressively optimizes the partial ordering of determinants. By doing so, RCI aligns the training objective more closely with the intrinsic ranking nature of SCI. Extensive benchmarks across both plane-wave and Gaussian basis sets, including the molecules N2, CO, H2O, NH3, and C2, demonstrate the efficiency of RCI. Compared to previously reported classification baselines, RCI consistently accelerates convergence─reducing overall computational time by 23% to over 50% depending on the system and requiring only 55% of the determinant count in representative cases such as N2 and CO. Furthermore, RCI exhibits robust performance and reaches chemical accuracy on the highly challenging iron-sulfur cluster using only 12% of the full CI space. Notably, RCI outperforms recent regression-based SCI methods by delivering a more than 15% improvement in accuracy at comparable determinant counts. RCI also demonstrates higher efficiency than heat-bath CI on the strongly correlated chromium dimer, yielding a compact and accurate wave function. This pairwise learning-to-rank model provides a lightweight and modular plugin that can be seamlessly incorporated into other supervised learning frameworks.
Accurately solving the electronic Schrödinger equation for strongly correlated systems remains a central challenge in quantum chemistry, where the exponential growth of configuration space limits the applicability of exact methods. Selected Configuration Interaction (SCI) algorithms address this challenge by adaptively constructing compact determinantal expansions, yet their efficiency depends critically on the quality of the sampling strategy used to identify chemically important configurations. Here we introduce the Handover Iterative Neural Quantum State (HI-NQS) algorithm, which embeds a classically trained autoregressive Transformer neural quantum state within the iterative sample--diagonalize--update framework of Sample-Based Quantum Diagonalization. A dual-channel Transformer architecture with explicit spin-up/spin-down cross-attention encodes fermionic spin structure as an architectural inductive bias, enabling expressive and physically informed wavefunction representations. After each subspace diagonalization, the resulting eigenvector is distilled back into the network through a factorized spin-marginal teacher signal, establishing a closed feedback loop between generative sampling and exact diagonalization. Benchmarks across a range of small molecules and a systematic nitrogen active-space series demonstrate that HI-NQS achieves chemical accuracy on all systems tested, with determinant-count scaling substantially more favorable than conventional CIPSI-based SCI for all but the smallest active spaces. All calculations are performed on GPU hardware without quantum computing resources, establishing HI-NQS as an efficient and scalable purely classical approach to the selected configuration interaction problem.
High Resolution Image Download MS PowerPoint Slide Selected Configuration Interaction (SCI) is a method in molecular electronic structure theory that allows for the construction of configuration subspaces adapted to the particular system under study. This adaptability is achieved by guiding subspace construction with respect to a selection criterion designed to identify and retain the most important configurations for an accurate description of the system. Quantum-SCI (QSCI) introduces quantum resources to inform the construction of these subspaces, motivated by the classical hardness of state sampling and system dynamics. As with conventional SCI, the CI routine is still performed classically and is therefore not corrupted by hardware noise; only the subspace quality is affected by deficiencies in the quantum hardware. Previous QSCI approaches take the measurements produced by a physically motivated circuit construction, often with a recovery step to mitigate against errors, and form a subspace from the resulting configurations. We propose an alternative approach that is more aligned with conventional selection criteria but attempts to inject classically inaccessible information into the selection step with the aim of enabling new subspace expansion pathways. The approach presented in this work uses the population statistics of a time-evolved quantum state to predict likely single and double excitations away from existing configurations to bias the subspace expansion procedure. Importantly, this occupancy-guided expansion complements rather than replaces the direct inclusion of valid configurations sampled from the time-evolved quantum state, so determinants containing higher-order excitations can still enter the variational space in a single iteration. We also include multireference perturbation theory to capture missed correlations outside the configuration subspace. This is demonstrated on hardware by using 42 qubits of an IQM superconducting device to compute the potential energy curve of SiH 4 in a 6-31G basis set as the Si–H bonds are stretched. We benchmark against the best-in-class Heatbath CI algorithm to assess the compactness of the resulting wave function.
Quantum many-body calculations are fundamentally limited by the exponential growth of the configuration space, making accurate treatment of strongly correlated systems computationally prohibitive. Here we present the Generative Transformer Neural Network Selected Configuration Interaction (GTNN-SCI), a Transformer-based machine learning approach that generatively samples important configurations to accelerate many-body quantum chemistry calculations. By leveraging the Transformer architecture’s self-attention mechanism to capture long-range electron correlations, GTNN-SCI achieves up to 10× speedup compared to state-of-the-art neural network methods while maintaining high accuracy. We demonstrate the efficacy of GTNN-SCI by calculating correlation and binding energies for representative molecules including N 2, H 2 O, and C 2 using both Gaussian (cc-pVDZ) and plane-wave basis sets, achieving faster convergence and lower energies than previously reported neural network-based selected CI techniques. Most significantly, our generative approach identifies higher-order excitations missed by conventional coupling schemes, yielding lower variational energies than established methods including heat-bath CI. This capability enables GTNN-SCI to accurately treat the strongly correlated [Fe 2 S 2 (SCH 3 ) 4 ] 2– ([2Fe–2S]) cluster system, achieving ground-state energies within chemical accuracy of DMRG benchmarks, whereas conventional selected CI methods have failed on this system. The GTNN-SCI method thus combines modern deep learning with high-performance electronic structure computation, providing an efficient and precise avenue for solving the electronic Schrödinger equation in challenging molecular systems.
Coupled Cluster (CC) theory is one of the most popular correlated wavefunction methods, yet the steep computational cost scaling of CC truncated at higher rank limits its application. Modern Selected Configuration Interaction (SCI) methods successfully exploit the sparsity of ground-state wavefunctions in the determinant space, motivating an analogous approach with the exponential CC ansatz. In this work, we introduce Selected Coupled Cluster (SCC), an arbitrary-order sparse CC method that explicitly solves the CC residual equations within a compact amplitude space guided by an SCI wavefunction. SCC optimizes the CC amplitudes in the sparse amplitude space. Using a pilot Just-in-Time (JIT) compiled implementation, we benchmark SCC against SCI, SCI+PT2, standard rank-truncated dense CC (CCSD, CCSDT), and DMRG across various chemical systems. We demonstrate that SCC exhibits fast, monotonic convergence with the SCI variational threshold, outperforming SCI and surpassing SCI+PT2 in the tight-threshold regime. SCC also provides accurate energies for high-dimensional topologies, where DMRG typically converges slowly with bond dimension. SCC thus presents an efficient drop-in enhancement for modern SCI workflows.
Recently, a new distributed implementation of the full configuration interaction (FCI) method has been reported [Gao et al. J. Chem Theory Comput. 2024, 20, 1185]. Thanks to a hybrid parallelization scheme, the authors were able to compute the exact energy of propane (\ce{C3H8}) in the minimal basis STO-3G. This formidable task involves handling an active space of 26 electrons in 23 orbitals or a Hilbert space of \SI{1.3d12} determinants. This is, by far, the largest FCI calculation reported to date. Here, we illustrate how, from a general point of view, selected configuration interaction (SCI) can achieve microhartree accuracy at a fraction of the computational and memory cost, via a sparse exploration of the FCI space. The present SCI calculations are performed with the \textit{Configuration Interaction using a Perturbative Selection made Iteratively} (CIPSI) algorithm, as implemented in a determinant-driven way in the \textsc{quantum package} software. The present study reinforces the common wisdom that among the exponentially large number of determinants in the FCI space, only a tiny fraction of them significantly contribute to the energy. More importantly, it demonstrates the feasibility of achieving comparable accuracy using more reasonable and sustainable computational resources, hence reducing the ever-growing carbon footprint of computational chemistry.