Splitting the Difference: How Lie Algebra Structure Shapes Commutative Subalgebras
This review explores the conditions under which specific decompositions of Lie algebras yield elegantly structured, polynomial-like subalgebras.
Science
When Forward Isn’t Enough: The Roots of System Failure
A new analysis reveals a pervasive flaw in how systems process information, leading to semantic corruption across technologies from file syncing to human memory.
Beyond Functionals: The Limits of Wavefunction-in-DFT Accuracy
![Employing the DMRG and DMRG-in-DFT methods with the cc-pVDZ basis set on the propionitrile molecule, calculations of absolute and relative energies-expressed in Hartrees-demonstrate that incorporating the PDFT functional for nonadditive exchange-correlation energy within the DMRG-in-DFT framework offers a distinct approach compared to utilizing the PBE functional, as detailed in [latex]Eq. (53)[/latex].](https://arxiv.org/html/2603.05303v1/2603.05303v1/x6.png)
A new analysis reveals that the primary obstacle to achieving exact solutions in wavefunction-in-DFT methods isn’t technical limitations, but the inherent challenges in accurately modeling electron correlation.
Heavy Quarks, Subtle Shifts: Mapping the Edge of Nuclear Matter
![The study maps the chiral phase diagram of Quantum Chromodynamics (QCD) using a novel framework-detailed by solid lines-and validates its findings against established benchmarks from miniDSE, functional Renormalization Group (fRG), and Dyson-Schwinger Equation (DSE) methods-represented by colored markers-for both [latex]2+1[/latex] and [latex]2+1+1[/latex] flavor configurations, including the incorporation of a charm loop to refine accuracy.](https://arxiv.org/html/2603.04728v1/2603.04728v1/x5.png)
New research reveals that incorporating charm quarks into calculations of quantum chromodynamics subtly alters the predicted location of the critical endpoint in the chiral phase diagram.
Bouncing Back from Collapse: A Quantum Dust Model
![The probability density at the point of bounce, [latex]|\psi(T_b,v)|^2[/latex], exhibits distinct variations across different values of η, while maintaining a minimal value of σ, and occurs at a significantly increased volume-approximately 2000-compared to prior configurations.](https://arxiv.org/html/2603.04995v1/2603.04995v1/BounceLargeVb_rho3.png)
New research applies loop quantum gravity to a simplified universe model, offering a potential pathway to resolving the singularity at the heart of gravitational collapse.
Securing Sparse Matrix Math in the Cloud
![Homomorphic encryption of sparse matrix-vector multiplication (SpMV) in compressed sparse row (CSR) format introduces substantial computational overhead, as each element-wise product, while achievable, necessitates a costly homomorphic rotation and multiplication for subsequent aggregation-a limitation stemming from the inability to efficiently combine encrypted partial results without incurring these operations on every addition required to construct the final encrypted vector sum [latex] \sum_{j=1}^{n} A_{ij} x_j [/latex].](https://arxiv.org/html/2603.04742v1/2603.04742v1/x1.png)
A new approach to privacy-preserving sparse matrix-vector multiplication unlocks significant performance gains using homomorphic encryption.
Fortifying Homomorphic Encryption Against Known-Plaintext Attacks
A new modification to the Fully Homomorphic Modified Rivest Scheme enhances security while preserving the ability to compute on encrypted data.
Building Resilient AI: A New Loss Function for Error-Tolerant Neural Networks
A novel training approach enhances the robustness of quantized neural networks against bit errors, paving the way for more reliable deployment in resource-constrained environments.
AI Grading Gets a Reality Check
/|\text{Baseline}-\text{Human}|[/latex]- achieved by large language models on the ASAP-SAS dataset; a transformer ensemble (point A, Ormerod, 2022) and a fine-tuned GPT ensemble (point B, Ormerod & Kwako, 2024) progressively surpass earlier baselines-with a subsequent GPT-4 implementation (point C, Jiang & Bosch, 2024) projected to reach a normalized score of -1.52-indicating a substantial, ongoing refinement in automated assessment capabilities.](https://arxiv.org/html/2603.04820v1/2603.04820v1/x1.png)
A comprehensive analysis reveals that current artificial intelligence systems struggle to accurately assess short-answer responses, exposing fundamental limitations in their understanding of meaning.
Drone Hacking: 5G’s Hidden Weakness
New research reveals critical vulnerabilities in 5G Standalone networks that could allow attackers to compromise unmanned aerial vehicle (UAV) communications and control.