Jinyuan Wu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Jinyuan Wu has a total of 9 co-authors. Here are the top co-authors by number of shared publications:
- Diana Yuan Qiu
Yale University
4 shared publications Bowen Hou
Department of Materials Science, Yale University, New Haven, USA
3 shared publications- Cong Su
Department of Materials Science, Yale University, New Haven, USA
1 shared publication Luna Y. Liu
Department of Applied Physics, Yale University, New Haven, USA
1 shared publication- Steffi Y. Woo
Center for Nanophase Materials Sciences, Oak Ridge National Laboratory
1 shared publication - Thomas K. Allison
Stony Brook University
1 shared publication
Latest Publications
- 24 August 2026
- Published by American Physical Society (APS) in Physical Review Letters
- Vol. 137 (9)
Signature of a Trion in Photoemission
- 24 August 2026
- Published by American Physical Society (APS) in Physical Review Letters
- Vol. 137 (9)
Signature of a Trion in Photoemission
- 16 March 2026
- Published by Springer Nature in Nature Physics
- Vol. 22 (4) , 521-526
Direct observation of massless excitons and linear exciton dispersion
- 16 March 2026
- Published by Springer Nature in Nature Physics
- Vol. 22 (4) , 521-526
Direct observation of massless excitons and linear exciton dispersion
- Open Access
- 2 November 2024
- Published by Springer Nature in Nature Communications
- Vol. 15 (1) , 1-11
Unsupervised representation learning of Kohn–Sham states and consequences for downstream predictions of many-body effects
Representation learning for the electronic structure problem is a major challenge of machine learning in computational condensed matter and materials physics. Within quantum mechanical first principles approaches, density functional theory (DFT) is the preeminent tool for understanding electronic structure, and the high-dimensional DFT wavefunctions serve as building blocks for downstream calculations of correlated many-body excitations and related physical observables. Here, we use variational autoencoders (VAE) for the unsupervised learning of DFT wavefunctions and show that these wavefunctions lie in a low-dimensional manifold within latent space. Our model autonomously determines the optimal representation of the electronic structure, avoiding limitations due to manual feature engineering. To demonstrate the utility of the latent space representation of the DFT wavefunction, we use it for the supervised training of neural networks (NN) for downstream prediction of quasiparticle bandstructures within the GW formalism. The GW prediction achieves a low error of 0.11 eV for a combined test set of two-dimensional metals and semiconductors, suggesting that the latent space representation captures key physical information from the original data. Finally, we explore the generative ability and interpretability of the VAE representation.
- 2 November 2024
- Published by Springer Nature in Nature Communications
- Vol. 15 (1) , 1-11
Open AccessUnsupervised representation learning of Kohn–Sham states and consequences for downstream predictions of many-body effects
Representation learning for the electronic structure problem is a major challenge of machine learning in computational condensed matter and materials physics. Within quantum mechanical first principles approaches, density functional theory (DFT) is the preeminent tool for understanding electronic structure, and the high-dimensional DFT wavefunctions serve as building blocks for downstream calculations of correlated many-body excitations and related physical observables. Here, we use variational autoencoders (VAE) for the unsupervised learning of DFT wavefunctions and show that these wavefunctions lie in a low-dimensional manifold within latent space. Our model autonomously determines the optimal representation of the electronic structure, avoiding limitations due to manual feature engineering. To demonstrate the utility of the latent space representation of the DFT wavefunction, we use it for the supervised training of neural networks (NN) for downstream prediction of quasiparticle bandstructures within the GW formalism. The GW prediction achieves a low error of 0.11 eV for a combined test set of two-dimensional metals and semiconductors, suggesting that the latent space representation captures key physical information from the original data. Finally, we explore the generative ability and interpretability of the VAE representation.
- 16 August 2024
- Published by American Physical Society (APS) in Physical Review B
- Vol. 110 (7) , 075133
Quasiparticle and excitonic properties of monolayer within many-body perturbation theoryIn the monolayer limit, is a two-dimensional topological insulator exhibiting the quantum spin Hall effect and is believed to host an excitonic insulator ground state. However, theoretical analysis of this system is complicated by the difficulty of obtaining descriptions of the single-quasiparticle band structure consistent with experimental measurement within conventional first-principles techniques. Previous band-structure calculations using the Perdew-Burke-Ernzerhof functional and a one-shot approximation result in a semimetallic band structure, while calculations with hybrid functionals appear to open a band gap. Here, we demonstrate that self-consistently updating wave functions within a static approximation (static COHSEX) can reproduce the insulating band structure experimentally observed by angle-resolved photoemission spectroscopy without resorting to mechanisms beyond the quasiparticle picture. Finally, a finite-momentum Bethe-Salpeter equation calculation on top of self-consistent results in negative exciton excitation energies, leaving open the possibility of excitonic instability in monolayer .- 16 August 2024
- Published by American Physical Society (APS) in Physical Review B
- Vol. 110 (7) , 075133
Quasiparticle and excitonic properties of monolayer within many-body perturbation theoryIn the monolayer limit, is a two-dimensional topological insulator exhibiting the quantum spin Hall effect and is believed to host an excitonic insulator ground state. However, theoretical analysis of this system is complicated by the difficulty of obtaining descriptions of the single-quasiparticle band structure consistent with experimental measurement within conventional first-principles techniques. Previous band-structure calculations using the Perdew-Burke-Ernzerhof functional and a one-shot approximation result in a semimetallic band structure, while calculations with hybrid functionals appear to open a band gap. Here, we demonstrate that self-consistently updating wave functions within a static approximation (static COHSEX) can reproduce the insulating band structure experimentally observed by angle-resolved photoemission spectroscopy without resorting to mechanisms beyond the quasiparticle picture. Finally, a finite-momentum Bethe-Salpeter equation calculation on top of self-consistent results in negative exciton excitation energies, leaving open the possibility of excitonic instability in monolayer .