Coupled Cluster con MoLe: Molecular Orbital Learning for Neural Wavefunctions

Luca Thiede, Abdulrahman Aldossary, Andreas Burger, Jorge A. Campos-Gonzalez-Angulo, Alex Zook, Melisa Alkan, Kouhei Nakaji, Taylor Lee Patti, Jérôme Florian Gonthier, Mohammad Ghazi Vakili, Alan Aspuru-Guzik
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:120463-120489, 2026.

Abstract

Density functional theory (DFT) is the most widely used method for calculating molecular properties; however, its accuracy is often insufficient for quantitative predictions. Coupled cluster (CC) theory is the most successful method for achieving accuracy beyond DFT and predicting properties that closely align with experiment. It is known as the “gold standard” of quantum chemistry. Unfortunately, the high computational cost of CC limits its widespread applicability. In this work, we present the Molecular Orbital Learning Model (MoLe), an equivariant machine learning model that directly predicts CC’s core mathematical objects, the excitation amplitudes, from the mean-field Hartree-Fock molecular orbitals as inputs. We test various aspects of our model and demonstrate its very high data efficiency and remarkable out-of-distribution generalization to larger molecules and off-equilibrium geometries, despite being trained only on small equilibrium geometries. Finally, we also examine its ability to reduce the number of cycles required to converge CC calculations. MoLe can set the foundations for high-accuracy wavefunction-based ML architectures to accelerate molecular design and complement force-field approaches.

Cite this Paper


BibTeX
@InProceedings{pmlr-v306-thiede26a, title = {Coupled Cluster con {M}o{L}e: Molecular Orbital Learning for Neural Wavefunctions}, author = {Thiede, Luca and Aldossary, Abdulrahman and Burger, Andreas and Campos-Gonzalez-Angulo, Jorge A. and Zook, Alex and Alkan, Melisa and Nakaji, Kouhei and Patti, Taylor Lee and Gonthier, J\'{e}r\^{o}me Florian and Vakili, Mohammad Ghazi and Aspuru-Guzik, Alan}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {120463--120489}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/thiede26a/thiede26a.pdf}, url = {https://proceedings.mlr.press/v306/thiede26a.html}, abstract = {Density functional theory (DFT) is the most widely used method for calculating molecular properties; however, its accuracy is often insufficient for quantitative predictions. Coupled cluster (CC) theory is the most successful method for achieving accuracy beyond DFT and predicting properties that closely align with experiment. It is known as the “gold standard” of quantum chemistry. Unfortunately, the high computational cost of CC limits its widespread applicability. In this work, we present the Molecular Orbital Learning Model (MoLe), an equivariant machine learning model that directly predicts CC’s core mathematical objects, the excitation amplitudes, from the mean-field Hartree-Fock molecular orbitals as inputs. We test various aspects of our model and demonstrate its very high data efficiency and remarkable out-of-distribution generalization to larger molecules and off-equilibrium geometries, despite being trained only on small equilibrium geometries. Finally, we also examine its ability to reduce the number of cycles required to converge CC calculations. MoLe can set the foundations for high-accuracy wavefunction-based ML architectures to accelerate molecular design and complement force-field approaches.} }
Endnote
%0 Conference Paper %T Coupled Cluster con MoLe: Molecular Orbital Learning for Neural Wavefunctions %A Luca Thiede %A Abdulrahman Aldossary %A Andreas Burger %A Jorge A. Campos-Gonzalez-Angulo %A Alex Zook %A Melisa Alkan %A Kouhei Nakaji %A Taylor Lee Patti %A Jérôme Florian Gonthier %A Mohammad Ghazi Vakili %A Alan Aspuru-Guzik %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-thiede26a %I PMLR %P 120463--120489 %U https://proceedings.mlr.press/v306/thiede26a.html %V 306 %X Density functional theory (DFT) is the most widely used method for calculating molecular properties; however, its accuracy is often insufficient for quantitative predictions. Coupled cluster (CC) theory is the most successful method for achieving accuracy beyond DFT and predicting properties that closely align with experiment. It is known as the “gold standard” of quantum chemistry. Unfortunately, the high computational cost of CC limits its widespread applicability. In this work, we present the Molecular Orbital Learning Model (MoLe), an equivariant machine learning model that directly predicts CC’s core mathematical objects, the excitation amplitudes, from the mean-field Hartree-Fock molecular orbitals as inputs. We test various aspects of our model and demonstrate its very high data efficiency and remarkable out-of-distribution generalization to larger molecules and off-equilibrium geometries, despite being trained only on small equilibrium geometries. Finally, we also examine its ability to reduce the number of cycles required to converge CC calculations. MoLe can set the foundations for high-accuracy wavefunction-based ML architectures to accelerate molecular design and complement force-field approaches.
APA
Thiede, L., Aldossary, A., Burger, A., Campos-Gonzalez-Angulo, J.A., Zook, A., Alkan, M., Nakaji, K., Patti, T.L., Gonthier, J.F., Vakili, M.G. & Aspuru-Guzik, A.. (2026). Coupled Cluster con MoLe: Molecular Orbital Learning for Neural Wavefunctions. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:120463-120489 Available from https://proceedings.mlr.press/v306/thiede26a.html.

Related Material