Abstract:The recently proposed machine learning-based physically-constrained nonlocal (MPN) kinetic energy density functional (KEDF) can be used for simple metals and their alloys [Phys. Rev. B 109, 115135 (2024)]. However, the MPN KEDF does not perform well for semiconductors. Here we propose a multi-channel MPN (CPN) KEDF, which extends the MPN KEDF to semiconductors by integrating information collected from multiple channels, with each channel featuring a specific length scale in real space. The CPN KEDF is systematically tested on silicon and binary semiconductors. We find that the multi-channel design for KEDF is beneficial for machine-learning-based models in capturing the characteristics of semiconductors, particularly in handling covalent bonds. In particular, the CPN5 KEDF, which utilizes five channels, demonstrates excellent accuracy across all tested systems. These results offer a new path for generating KEDFs for semiconductors.

What problem does this paper attempt to address?

The problem this paper attempts to address is the development of a multi-channel machine learning-based physically constrained nonlocal kinetic energy density functional (CPN KEDF) suitable for semiconductor systems. Specifically, existing machine learning-based nonlocal kinetic energy density functionals (MPN KEDF) perform well for simple metals and their alloys but perform poorly for semiconductor systems. Therefore, the authors propose a multi-channel design of CPN KEDF to improve the capture of semiconductor characteristics, particularly the ability to handle covalent bonds, by integrating information from different scales. ### Main Issues 1. **Limitations of Existing Methods**: - Existing MPN KEDF performs poorly when dealing with semiconductors. - The charge density of semiconductors includes highly localized electrons (covalent bond regions) and delocalized electrons, making it difficult for traditional KEDF to accurately describe. 2. **Objectives**: - Develop a new multi-channel MPN KEDF (CPN KEDF) that can effectively handle semiconductor systems. - Improve the accuracy and stability of the model by introducing multiple channels, each capturing information at specific scales. ### Solutions 1. **Multi-Channel Design**: - Each channel uses a specific kernel function \( w^{(\lambda)}(r - r') \), which captures information at different scales. - By integrating information from different channels, CPN KEDF can better capture the covalent bond characteristics in semiconductors. 2. **Physical Constraints**: - CPN KEDF strictly adheres to three physical constraints: scaling laws, free electron gas (FEG) limit, and non-negativity of Pauli energy density. - These constraints ensure the physical reasonableness and stability of the model. 3. **Performance Evaluation**: - The performance of CPN KEDF was systematically tested on a series of semiconductor systems (including silicon and binary semiconductors). - Results show that CPN 5KEDF (using five channels) performs excellently across all tested systems, with accuracy close to nonlocal KEDF and significantly better than semilocal KEDF. ### Conclusion - The multi-channel design significantly improves the accuracy and stability of CPN KEDF in handling semiconductor systems. - CPN 5KEDF can effectively capture covalent bond structures, which is difficult for other analytical KEDFs to achieve. - Although CPN KEDF still has some shortcomings, this study provides a new path for generating KEDFs that can handle both metal and semiconductor systems and offers a reference for the development of other machine learning-based functionals.

Multi-channel machine learning based nonlocal kinetic energy density functional for semiconductors

Machine learning based nonlocal kinetic energy density functional for simple metals and alloys

Truncated Non-Local Kinetic Energy Density Functionals for Metals and Semiconductors

Truncated Non-Local Kinetic Energy Density Functionals for Simple Metals and Silicon

Truncated nonlocal kinetic energy density functionals for simple metals and silicon

Fast and stable tight-binding framework for nonlocal kinetic energy density functional reconstruction in orbital-free density functional calculations

Nonlocal Machine-Learned Exchange Functional for Molecules and Solids

Development of a machine learning finite-range nonlocal density functional

Accelerating the calculation of electron-phonon coupling strength with machine learning

Accelerating finite-temperature Kohn-Sham density functional theory with deep neural networks

Machine learning for deep elastic strain engineering of semiconductor electronic band structure and effective mass

Universal Machine Learning Kohn-Sham Hamiltonian for Materials

Semilocal Kinetic Energy Density Functionals on Atoms and Diatoms

Machine learning of kinetic energy densities with target and feature smoothing: Better results with fewer training data

Learning electron densities in the condensed phase

A machine-learned kinetic energy model for light weight metals and compounds of group III-V elements

1D Kinetic Energy Density Functionals learned with Symbolic Regression

Machine Learning Electron Density Prediction Using Weighted Smooth Overlap of Atomic Positions

Combined Semiclassical And Effective-Mass Schrodinger Approach For Multiscale Analysis Of Semiconductor Nanostructures

Machine Learning the Physical Nonlocal Exchange–Correlation Functional of Density-Functional Theory