Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2605.07227 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914543053045760 |
|---|---|
| author | Kumar, Kammampati Sai Linda, Albert Maurya, Shubham Kumar Bhowmick, Somnath |
| author_facet | Kumar, Kammampati Sai Linda, Albert Maurya, Shubham Kumar Bhowmick, Somnath |
| contents | The fundamental quantity governing the mechanical and thermodynamic properties of a crystalline solid is its electronic charge density. Yet, its direct use for the rapid prediction of materials properties remains challenging due to its high dimensionality. Here, we present a physics-informed deep learning framework that directly predicts mechanical and thermodynamic properties from the three-dimensional electronic charge density derived from density functional theory (DFT). The proposed approach first utilizes a three-dimensional convolutional autoencoder for unsupervised dimensionality reduction, compressing a high-resolution charge-density grid (128 x 128 x 128) into a compact latent representation (16 x 16 x 16 x 16) while preserving physically meaningful features, as confirmed by negligible reconstruction errors across diverse crystal systems. The compressed latent-space representation of charge density is then used by two different regression models for property prediction: Light Gradient Boosting Machine (LightGBM) and Attention-based 3D Convolutional Neural Networks (Att CNN), and their performance is compared. Combining composition-based descriptors (Material Agnostic Platform for Informatics and Exploration or MAGPIE) with electronic charge density data further improves the model accuracy. Using a dataset of about 6059 inorganic compounds spanning multiple crystal symmetries, the models achieve strong predictive performance for bulk modulus K (R2 = 0.94), Young's modulus E (R2 = 0.88), shear modulus G (R2 = 0.87), formation energy Eform (R2 = 0.96), and Debye temperature Θ (R2 = 0.89). This work establishes electronic charge density as a transferable, physics-grounded descriptor for materials property prediction, requiring ~ 1/25 the computational resources of full-fledged DFT calculations. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_07227 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Physics Aware Representation Learning on Electronic Charge Density for Materials Property Prediction Kumar, Kammampati Sai Linda, Albert Maurya, Shubham Kumar Bhowmick, Somnath Materials Science The fundamental quantity governing the mechanical and thermodynamic properties of a crystalline solid is its electronic charge density. Yet, its direct use for the rapid prediction of materials properties remains challenging due to its high dimensionality. Here, we present a physics-informed deep learning framework that directly predicts mechanical and thermodynamic properties from the three-dimensional electronic charge density derived from density functional theory (DFT). The proposed approach first utilizes a three-dimensional convolutional autoencoder for unsupervised dimensionality reduction, compressing a high-resolution charge-density grid (128 x 128 x 128) into a compact latent representation (16 x 16 x 16 x 16) while preserving physically meaningful features, as confirmed by negligible reconstruction errors across diverse crystal systems. The compressed latent-space representation of charge density is then used by two different regression models for property prediction: Light Gradient Boosting Machine (LightGBM) and Attention-based 3D Convolutional Neural Networks (Att CNN), and their performance is compared. Combining composition-based descriptors (Material Agnostic Platform for Informatics and Exploration or MAGPIE) with electronic charge density data further improves the model accuracy. Using a dataset of about 6059 inorganic compounds spanning multiple crystal symmetries, the models achieve strong predictive performance for bulk modulus K (R2 = 0.94), Young's modulus E (R2 = 0.88), shear modulus G (R2 = 0.87), formation energy Eform (R2 = 0.96), and Debye temperature Θ (R2 = 0.89). This work establishes electronic charge density as a transferable, physics-grounded descriptor for materials property prediction, requiring ~ 1/25 the computational resources of full-fledged DFT calculations. |
| title | Physics Aware Representation Learning on Electronic Charge Density for Materials Property Prediction |
| topic | Materials Science |
| url | https://arxiv.org/abs/2605.07227 |