Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:2605.04174

A Transferable Machine Learning Approach to Predict Optimized Orbitals for Electronic Structure Problems

Lucas van der Horst, Maniraman Periyasamy, Abhishek Y. Dubey, Davide Bincoletto, Jakob S. Kottmann, Daniel D. Scherer

quant-phSubmitted May 5, 2026• Announced May 7, 2026• 2 citations
arXivHTMLPDF

Abstract

Variational quantum eigensolver ansätze hold considerable promise for ground-state energy calculations on near-term quantum hardware, yet most promising ansatz designs currently strongly depend on how well the molecular orbital basis captures the electronic correlation of the system. Computing optimized orbital coefficients via classical routines is computationally expensive and must be performed independently for each molecular geometry -- a bottleneck that limits scalability across chemical space. We present a graph neural network framework that predicts optimized orbital coefficients directly from molecular geometry and pair-wise bonding structure. Trained on hydrogenic systems of modest size ($H_4$ and $H_6$) across tens of thousands of geometries, our model transfers to larger, unseen systems ($H_8$, $H_{10}$ and $H_{12}$) without retraining -- demonstrating strong out-of-distribution generalization with respect to system size. When evaluating on structured and random configurations, and comparing against energies obtained with full classical optimization, our model reaches mean absolute energy errors $\mathcal{O}(10^2)$ and $\mathcal{O}(10)$ milli-Hartrees, respectively. Beyond energy estimation, the predicted orbitals serve as high-quality warm-start initializations that substantially reduce optimizer iterations to ground-state energy convergence. These results establish graph neural networks as an effective and scalable strategy for accelerating orbital optimization in hybrid quantum-classical workflows, directly reducing the classical pre-processing overhead that currently limits the practical deployment of variational quantum eigensolver on near-term quantum hardware.

Comments
13 Pages, 4 Figures, 5 Tables

Related papers

Machine-Learned Representations of Basis Sets and Their Application in Quantum Computational Chemistry

Wenhao He

Quantum simulations of electronic structure promise to deliver significant speedups over classical methods, but remain limited by the number of qubits on near-term devices. A key strategy to reduce quantum resource requirements is to truncate the molecular Hilbert space via compact and efficient basis sets. However, most optimized basis sets either rely on predefined heuristics or require expensive classical computations, such as CASSCF orbital optimization or ℓ1-norm minimization of the Hamiltonian. In this work, we introduce a general machine learning framework for fast basis set prediction in quantum computational chemistry. Our method employs an equivariant graph neural network that outputs a Hermitian matrix encoding optimized molecular orbitals. The eigenvectors of this matrix define a transferable and efficient basis set, trained on orbitals obtained via CASSCF and Hamiltonian ℓ1 norm optimization. We evaluate our model on hydrogen chains and demonstrate that the predicted bases achieve energy accuracy and Hamiltonian sparsity comparable to orbital-optimized methods, while reducing classical preprocessing time. In addition, the predicted orbitals can be directly used as high-quality initial guesses for CASSCF calculations, further accelerating their convergence.

•Sep 1, 2025•0 citations

Smart Machine Learning for Solving the Full Electron-Nuclear Schrödinger Equation with a Quantum-Chemically Inspired Wavefunction Ansatz

Yaolong Zhang, Bin Jiang, Hua Guo

Directly solving the full electron-nuclear Schrödinger equation remains one of the grand challenges in quantum mechanics. Here, we propose a quantum-chemically motivated orbital-based neural network wavefunction ansatz with explicit many-body electron-electron correlation terms, with parameters depending on the entire electron-nuclear configuration, enabling an accurate description of strongly correlated and/or multi-reference electronic structures by a single Slater-determinant across a broad nuclear configuration space. To train this wavefunction, we further develop an efficient hybrid optimization strategy that combines the variational Monte Carlo technique with local-energy constraints, mitigating the intrinsic sampling bias of the former and considerably reducing the statistical error. These advances are embedded in a unified neural network framework, Schrödinger, enabling the determination of globally accurate potential energy surfaces with a single training for systems ranging from strongly correlated diatomic molecules, triatomic reactions containing conical intersections, to polyatomic multi-reference molecules. This work offers a practical machine-learning tool for solving the electron-nuclear Schrödinger equation and capturing full quantum effects beyond the Born-Oppenheimer approximation.

physics.chem-phCross-listedarXiv:2609.40112•Oct 1, 2026•0 citations

Foundation Neural-Network Quantum States for Molecular Potential Energy Surfaces in Second Quantization

Lizhong Fu, Jianan Wei, Wenguan Wang +1 more

Second-quantized neural-network quantum states have achieved accurate molecular energies, but extending them across molecular geometries requires a shared representation of the geometry-dependent wavefunction coefficients. We introduce geometry-conditioned foundation neural-network quantum states for molecular electronic structure in second quantization. A single autoregressive model learns a family of ground states from sparse anchor geometries and provides wavefunctions at untrained geometries without further optimization. Orbital alignment matches orbital identities and transports their phases, establishing an aligned orbital basis across geometries. Frozen energies reach chemical accuracy at every untrained query geometry for N$_2$, CO, and H$_4$. On additional molecular paths, the energy-trained wavefunctions yield dipoles, quadrupoles, and natural occupations without property labels. Across three paired N$_2$ training seeds, orbital alignment lowers the mean absolute energy error over all untrained query geometries from 34-37 mHa to 0.049-0.085 mHa. At approximately 1 mHa mean absolute error, frozen evaluation reduces the per-geometry cost by $986\times$ relative to independent optimization, yielding an estimated $25.8\times$ end-to-end GPU-cost reduction on a 161-point N$_2$ grid.

physics.chem-phCross-listedarXiv:2609.37733•Sep 30, 2026•0 citations

A Unified Benchmarking Study of Variational Quantum Eigensolver Ansatz for Electronic Structure Problems

Julie Maria Raju, Abhoy Kole, Julen Larrucea +3 more

Electronic structure calculations become computationally expensive on classical computers as electron correlation increases. Quantum computing offers a promising alternative through hybrid quantum-classical Variational Quantum Eigensolver (VQE) on Noisy-Intermediate Scale Quantum (NISQ) devices. The choice of ansatz plays a pivotal role in the performance and scalability of the VQE, yet selecting an optimal ansatz remains challenging. Existing approaches including chemically inspired ansatz, adaptive schemes and Hardware Efficient Ansätze (HEA) lack a unified comparison. This gap makes the scalability and trade-offs of each method unclear for varying molecular sizes. This work presents a systematic benchmarking study comparing Unitary Couples Cluster Singles and Doubles Ansatz (UCCSD), fermionic ADAPT-VQE, BasicPCA HEA, and Excitation Preserving HEA across molecules of varying size like $H_2$, $LiH$, $N_2$. We evaluate trade-offs among circuit depth, parameter count, ground-state energy error, and NISQ feasibility. ADAPT-VQE achieves accuracy comparable to UCCSD with lower resource overhead, but becomes impractical for modest system sizes ($\sim$10 electrons) due to increasing circuit complexity. In contrast, HEA reduces hardware costs but requires more parameters and sacrifices accuracy, with performance strongly dependent on molecular size. The Excitation Preserving ansatz remains below chemical accuracy for $H_2$ and $LiH$, and fails for $N_2$, underscoring the need to balance expressibility and trainability. Overall, these results highlight key trade-offs between accuracy and resource efficiency across VQE ansätze, emphasizing the need to improve ADAPT-VQE scalability and HEA accuracy while addressing optimization challenges for near-term quantum hardware.

arXiv:2611.15365•Nov 11, 2026•0 citations

Improving Generalization and Trainability of Quantum Eigensolvers via Graph Neural Encoding

Jungyun Lee, Daniel K. Park

Determining the ground state of a many-body Hamiltonian is a central problem across physics, chemistry, and combinatorial optimization, yet it is often classically intractable due to the exponential growth of Hilbert space with system size. Even on fault-tolerant quantum computers, quantum algorithms with convergence guarantees -- such as quantum phase estimation and quantum subspace methods -- require an initial state with sufficiently large overlap with the true ground state to be effective. Variational quantum eigensolvers (VQEs) are natural candidates for preparing such states; however, standard VQEs typically exhibit poor generalization, requiring retraining for each Hamiltonian instance, and often suffer from barren plateaus, where gradients can vanish exponentially with circuit depth and system size. To address these limitations, we propose an end-to-end representation learning framework that combines a graph autoencoder with a classical neural network to generate VQE parameters that generalize across Hamiltonian instances. By encoding interaction topology and coupling structure, the proposed model produces high-overlap initial states without instance-specific optimization. Through extensive numerical experiments on families of one- and two-local Hamiltonians, we demonstrate improved generalization and trainability, manifested as reduced test error and a significantly milder decay of gradient variance. We further show that our method substantially accelerates convergence in quantum subspace-based eigensolvers, highlighting its practical impact for downstream quantum algorithms.

quant-pharXiv:2602.19752•Machine Learning Science and Technology•Feb 24, 2026•1 citation

QUANTUM DEEP LEARNING FRAMEWORK FOR ACCURATE MOLECULAR GROUND-STATE ENERGY PREDICTION IN COMPLEX CHEMICAL SYSTEMS

Giribabu Sadineni, Rama Chaithanya Tanguturi

Correct prediction of the molecular ground-state energy is a fundamental problem in quantum chemistry, materials science and drug screening and discovery which directly impacts the understanding of the stability and chemical behavior of molecules. For complex molecular systems, traditional computational techniques like the Hartree-Fock and the Density Functional Theory (DFT) can be very costly. The improvement of molecular energy prediction efficiency and accuracy has recently opened new possibilities thanks to the progress in quantum computation and deep learning. In this paper, a hybrid Quantum Deep Learning Framework for Accurate Molecular Ground-State Energy Prediction in Complex Chemical Systems combining Variational Quantum Algorithms and Deep Neural Network Architectures is proposed. The framework builds molecular Hamiltonians, optimizes the quantum state with the Variational Quantum Eigensolver (VQE), and builds quantum-generated features for deep learning-based energy estimation. Performance measures used for the evaluation of the proposed model are Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and Coefficient of Determination (R²). The proposed Quantum Deep Learning Framework has shown Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and R2 values equal to 0.042, 0.031, and 0.989 correspondingly, which proves higher performance from the standpoint of prediction. Prediction precision of the developed approach reaches 98.9% and exceeds that of all other methods known, including VQE, ADAPT-VQE, FermiNet, and Deep Schrödinger. Besides that, the proposed approach showed faster convergence, lower prediction variance, and higher computational efficiency during the molecular energies' prediction process. The results suggest that the proposed hybrid quantum-classical approach is promising and efficient for ground-state energy prediction of molecules, as well as useful in computational chemistry, materials discovery, and molecular design, and is scalable.

•Jun 30, 2026•0 citations
Variational quantum eigensolver ansätze hold considerable promise for ground-state energy calculations on near-term quantum hardware, yet most promising ansatz designs currently strongly depend on how well the molecular orbital basis captures the electronic correlation of the system. Computing optimized orbital coefficients via classical routines is computationally expensive and must be performed independently for each molecular geometry -- a bottleneck that limits scalability across chemical space. We present a graph neural network framework that predicts optimized orbital coefficients directly from molecular geometry and pair-wise bonding structure. Trained on hydrogenic systems of modest size (H4H_4H4​ and H6H_6H6​) across tens of thousands of geometries, our model transfers to larger, unseen systems (H8H_8H8​, H10H_{10}H10​ and H12H_{12}H12​) without retraining -- demonstrating strong out-of-distribution generalization with respect to system size. When evaluating on structured and random configurations, and comparing against energies obtained with full classical optimization, our model reaches mean absolute energy errors O(102)\mathcal{O}(10^2)O(102) and O(10)\mathcal{O}(10)O(10) milli-Hartrees, respectively. Beyond energy estimation, the predicted orbitals serve as high-quality warm-start initializations that substantially reduce optimizer iterations to ground-state energy convergence. These results establish graph neural networks as an effective and scalable strategy for accelerating orbital optimization in hybrid quantum-classical workflows, directly reducing the classical pre-processing overhead that currently limits the practical deployment of variational quantum eigensolver on near-term quantum hardware.
Quantum simulations of electronic structure promise to deliver significant speedups over classical methods, but remain limited by the number of qubits on near-term devices. A key strategy to reduce quantum resource requirements is to truncate the molecular Hilbert space via compact and efficient basis sets. However, most optimized basis sets either rely on predefined heuristics or require expensive classical computations, such as CASSCF orbital optimization or ℓ1-norm minimization of the Hamiltonian. In this work, we introduce a general machine learning framework for fast basis set prediction in quantum computational chemistry. Our method employs an equivariant graph neural network that outputs a Hermitian matrix encoding optimized molecular orbitals. The eigenvectors of this matrix define a transferable and efficient basis set, trained on orbitals obtained via CASSCF and Hamiltonian ℓ1 norm optimization. We evaluate our model on hydrogen chains and demonstrate that the predicted bases achieve energy accuracy and Hamiltonian sparsity comparable to orbital-optimized methods, while reducing classical preprocessing time. In addition, the predicted orbitals can be directly used as high-quality initial guesses for CASSCF calculations, further accelerating their convergence.
Directly solving the full electron-nuclear Schrödinger equation remains one of the grand challenges in quantum mechanics. Here, we propose a quantum-chemically motivated orbital-based neural network wavefunction ansatz with explicit many-body electron-electron correlation terms, with parameters depending on the entire electron-nuclear configuration, enabling an accurate description of strongly correlated and/or multi-reference electronic structures by a single Slater-determinant across a broad nuclear configuration space. To train this wavefunction, we further develop an efficient hybrid optimization strategy that combines the variational Monte Carlo technique with local-energy constraints, mitigating the intrinsic sampling bias of the former and considerably reducing the statistical error. These advances are embedded in a unified neural network framework, Schrödinger, enabling the determination of globally accurate potential energy surfaces with a single training for systems ranging from strongly correlated diatomic molecules, triatomic reactions containing conical intersections, to polyatomic multi-reference molecules. This work offers a practical machine-learning tool for solving the electron-nuclear Schrödinger equation and capturing full quantum effects beyond the Born-Oppenheimer approximation.
Second-quantized neural-network quantum states have achieved accurate molecular energies, but extending them across molecular geometries requires a shared representation of the geometry-dependent wavefunction coefficients. We introduce geometry-conditioned foundation neural-network quantum states for molecular electronic structure in second quantization. A single autoregressive model learns a family of ground states from sparse anchor geometries and provides wavefunctions at untrained geometries without further optimization. Orbital alignment matches orbital identities and transports their phases, establishing an aligned orbital basis across geometries. Frozen energies reach chemical accuracy at every untrained query geometry for N2_22​, CO, and H4_44​. On additional molecular paths, the energy-trained wavefunctions yield dipoles, quadrupoles, and natural occupations without property labels. Across three paired N2_22​ training seeds, orbital alignment lowers the mean absolute energy error over all untrained query geometries from 34-37 mHa to 0.049-0.085 mHa. At approximately 1 mHa mean absolute error, frozen evaluation reduces the per-geometry cost by 986×986\times986× relative to independent optimization, yielding an estimated 25.8×25.8\times25.8× end-to-end GPU-cost reduction on a 161-point N2_22​ grid.
Electronic structure calculations become computationally expensive on classical computers as electron correlation increases. Quantum computing offers a promising alternative through hybrid quantum-classical Variational Quantum Eigensolver (VQE) on Noisy-Intermediate Scale Quantum (NISQ) devices. The choice of ansatz plays a pivotal role in the performance and scalability of the VQE, yet selecting an optimal ansatz remains challenging. Existing approaches including chemically inspired ansatz, adaptive schemes and Hardware Efficient Ansätze (HEA) lack a unified comparison. This gap makes the scalability and trade-offs of each method unclear for varying molecular sizes. This work presents a systematic benchmarking study comparing Unitary Couples Cluster Singles and Doubles Ansatz (UCCSD), fermionic ADAPT-VQE, BasicPCA HEA, and Excitation Preserving HEA across molecules of varying size like H2H_2H2​, LiHLiHLiH, N2N_2N2​. We evaluate trade-offs among circuit depth, parameter count, ground-state energy error, and NISQ feasibility. ADAPT-VQE achieves accuracy comparable to UCCSD with lower resource overhead, but becomes impractical for modest system sizes ($\sim$10 electrons) due to increasing circuit complexity. In contrast, HEA reduces hardware costs but requires more parameters and sacrifices accuracy, with performance strongly dependent on molecular size. The Excitation Preserving ansatz remains below chemical accuracy for H2H_2H2​ and LiHLiHLiH, and fails for N2N_2N2​, underscoring the need to balance expressibility and trainability. Overall, these results highlight key trade-offs between accuracy and resource efficiency across VQE ansätze, emphasizing the need to improve ADAPT-VQE scalability and HEA accuracy while addressing optimization challenges for near-term quantum hardware.
Determining the ground state of a many-body Hamiltonian is a central problem across physics, chemistry, and combinatorial optimization, yet it is often classically intractable due to the exponential growth of Hilbert space with system size. Even on fault-tolerant quantum computers, quantum algorithms with convergence guarantees -- such as quantum phase estimation and quantum subspace methods -- require an initial state with sufficiently large overlap with the true ground state to be effective. Variational quantum eigensolvers (VQEs) are natural candidates for preparing such states; however, standard VQEs typically exhibit poor generalization, requiring retraining for each Hamiltonian instance, and often suffer from barren plateaus, where gradients can vanish exponentially with circuit depth and system size. To address these limitations, we propose an end-to-end representation learning framework that combines a graph autoencoder with a classical neural network to generate VQE parameters that generalize across Hamiltonian instances. By encoding interaction topology and coupling structure, the proposed model produces high-overlap initial states without instance-specific optimization. Through extensive numerical experiments on families of one- and two-local Hamiltonians, we demonstrate improved generalization and trainability, manifested as reduced test error and a significantly milder decay of gradient variance. We further show that our method substantially accelerates convergence in quantum subspace-based eigensolvers, highlighting its practical impact for downstream quantum algorithms.
Correct prediction of the molecular ground-state energy is a fundamental problem in quantum chemistry, materials science and drug screening and discovery which directly impacts the understanding of the stability and chemical behavior of molecules. For complex molecular systems, traditional computational techniques like the Hartree-Fock and the Density Functional Theory (DFT) can be very costly. The improvement of molecular energy prediction efficiency and accuracy has recently opened new possibilities thanks to the progress in quantum computation and deep learning. In this paper, a hybrid Quantum Deep Learning Framework for Accurate Molecular Ground-State Energy Prediction in Complex Chemical Systems combining Variational Quantum Algorithms and Deep Neural Network Architectures is proposed. The framework builds molecular Hamiltonians, optimizes the quantum state with the Variational Quantum Eigensolver (VQE), and builds quantum-generated features for deep learning-based energy estimation. Performance measures used for the evaluation of the proposed model are Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and Coefficient of Determination (R²). The proposed Quantum Deep Learning Framework has shown Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and R2 values equal to 0.042, 0.031, and 0.989 correspondingly, which proves higher performance from the standpoint of prediction. Prediction precision of the developed approach reaches 98.9% and exceeds that of all other methods known, including VQE, ADAPT-VQE, FermiNet, and Deep Schrödinger. Besides that, the proposed approach showed faster convergence, lower prediction variance, and higher computational efficiency during the molecular energies' prediction process. The results suggest that the proposed hybrid quantum-classical approach is promising and efficient for ground-state energy prediction of molecules, as well as useful in computational chemistry, materials discovery, and molecular design, and is scalable.