Track synthesis, materials and catalysis as it is published.
Search papers, labs, and topics across Lattice.
The authors construct NNLap, a machine-learned exchange-correlation functional that augments PBE with a neural network correction evaluated on the local electron density, its gradient, and its Laplacian. By training on exact exchange-correlation potentials and energies derived via inverse DFT from configuration-interaction densities across only eight systems (five atoms and three molecules), the model eliminates the need for explicit Kohn-Sham orbital dependence. The resulting Laplacian-level functional matches the thermochemical accuracy of advanced meta-GGAs like SCAN and r2SCAN while outperforming both r2SCAN and B3LYP on total energies.
Training on just eight small chemical systems is sufficient for a Laplacian-level neural functional to match orbital-dependent meta-GGAs like SCAN on thermochemistry while surpassing B3LYP on total energies.
We present NNLap, a machine-learned Laplacian-level exchange-correlation (XC) functional that augments PBE with a neural-network correction depending on the electron density, its gradient, and its Laplacian. The model is trained on exact XC potentials and energies, obtained through inverse density-functional theory (DFT) calculations on configuration-interaction densities. Despite training on only a few systems -- five atoms and three molecules -- the model achieves remarkable accuracy on thermochemistry benchmarks, competing with the meta-GGA functionals SCAN and r2SCAN. It also attains accurate total energies, comparable to SCAN and better than r2SCAN and B3LYP. This shows that a Laplacian-level model, trained on exact XC potentials and energies, can reach the accuracy of meta-GGAs without their orbital dependence.