Training Models of Atomic Charge by Predicting a Generalized Force
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
In atomistic simulation, ab initio methods are accurate but too computationally expensive for large systems, long trajectories, or high-throughput screening. Recently, machine-learned interatomic potentials (MLIPs) are approaching the accuracy of ab initio methods at speeds closer to traditional force fields by training on large datasets of ab initio results. Some datasets include atomic partial charges, which are convenient for training models with explicit charge assignments. We propose adding diversity to the training data by perturbing atomic charge, and annotating atoms with corresponding energy derivatives. These additional atom-level labels are identical to electronegativity as defined in conceptual density functional theory. To demonstrate this proposal, we construct a training dataset of crystals with two-atom unit cells, controlling the charge transfer between the atoms by imposing an external potential. By matching the form of the potential to the charge partitioning scheme, we compute the electronegativity difference from the strength of the imposed potential, and observe a near-linear relationship with charge transfer. Fitting these electronegativity differences with linear regression yields the parameters of a charge equilibration model. Our results demonstrate a new way of training charge prediction models, and show how to diversify MLIP training datasets while simultaneously adding atom-level response properties that can be used as training targets.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it