Expander attention as exchange-correlation
Karim K. Alaa El-Din, Antonius v. Strachwitz, Sam M. Vinko · 2026
Read Expander attention as exchange-correlation by Karim K. Alaa El-Din, Antonius v. Strachwitz, Sam M. Vinko (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-10265.txt.
Description
Karim K. Alaa El-Din, Antonius v. Strachwitz, and Sam M. Vinko tackle a scaling problem in quantum chemistry. Density functional theory is the workhorse of the field, but its exchange-correlation term, which encodes many-body quantum effects beyond a mean-field picture, must be approximated. Machine-learned versions can handle strongly correlated systems yet scale poorly, limiting their use. Their 2026 paper proposes a cheaper non-local alternative.
The new approximation uses an expander graph transformer ansatz, a network design built on expander graphs, and scales linearly with system size rather than quadratically or worse. It recovers the correct hydrogen molecule dissociation curve in the strongly correlated regime, where bonds stretch and electrons correlate strongly. Results on planar H4 also look promising, a system where even top-tier coupled-cluster methods fail. The path opens toward accurate machine-learned functionals cheap enough to deploy at scale.
Catalog record
| CREATOR | Karim K. Alaa El-Din, Antonius v. Strachwitz, Sam M. Vinko |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 38 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.10265