Smarter Text Chunks, Better Answers: A New Approach to Knowledge Retrieval
Researchers have developed a system that intelligently breaks down complex texts to improve the accuracy and relevance of information retrieved for question answering.
Science
Certifying Materials Simulations with Machine Learning
A new framework establishes rigorous safety guarantees for machine learning potentials used to predict material behavior, paving the way for more reliable discovery.
Fragile Order: How Small Changes Disrupt Predictable Systems
![The study of the Ishimori chain under perturbation reveals a crossover from ballistic to diffusive transport governed by Fermi’s Golden Rule, demonstrated by the collapse of rescaled dynamical exponents and a characteristic timescale of [latex]t\_{\star}\sim\lambda^{-2}[/latex], while weaker perturbations exhibit anomalously long crossover timescales-potentially scaling as [latex]t\_{\star}\sim\lambda^{-4}[/latex]-suggesting a nuanced relationship between perturbation strength and transport dynamics.](https://arxiv.org/html/2603.11712v1/x1.png)
New research reveals how even minor disturbances can dismantle the predictable behavior of classically integrable models on a lattice.
Beyond Bandwidth: Optimizing Chiplet Links for Reliability
As chiplet designs grow in complexity, ensuring robust die-to-die communication requires a nuanced approach to link selection that considers error correction overhead.
When Machine Learning Loses Its Way: Symmetry’s Role in Predicting Material States
![The model describes a system where electrons traverse a lattice-defined by intra-cell hopping amplitudes [latex]\{t_{1,1}, t_{1,2}, t_{1,3}, ...\}[/latex] and inter-cell amplitudes [latex]\{t_{2,1}, t_{2,2}, t_{2,3}, ...\}[/latex]-supplemented by a long-range hopping term [latex]t_3[/latex] and modulated by on-site potentials [latex]\{V_1, V_2, V_3, ...\}[/latex], enabling electron transport across the structure.](https://arxiv.org/html/2603.11688v1/x1.png)
A new study reveals that convolutional neural networks struggle to accurately classify topological materials when key symmetries are disrupted, hindering their ability to generalize beyond familiar configurations.
Decoding Higgs Decay: A New Approach to Quantum State Reconstruction
Researchers are developing advanced techniques to precisely map the quantum state of Higgs bosons decaying into pairs of W or Z bosons, even when theoretical approximations fall short.
Unlocking AI System Vulnerabilities: The Cascade Effect
New research reveals how coordinated software and hardware attacks can bypass AI safety measures and amplify adversarial threats in complex systems.
Building Better Code with Thoughtful AI
A new reasoning framework empowers AI agents to systematically improve code quality by continuously questioning and verifying design choices.
Beyond the Hype: Securing the Next Generation of AI Agents
As artificial intelligence agents become increasingly powerful, understanding and mitigating their unique security vulnerabilities is paramount.
Decoding Protein Acidity with Quantum-Inspired AI
![The pursuit of accurate residue-level [latex] pK_a [/latex] prediction has evolved through distinct methodological approaches, initially leveraging descriptor-driven resources like DeepKaDB-built upon curated features from soluble proteins-and subsequently expanding with simulation-driven datasets such as PHMD549-which utilizes GPU acceleration to extend PHMD279 to over 26,000 residues-culminating in hybrid quantum-classical frameworks like DQNN that integrate curated descriptors with quantum-inspired feature transformations.](https://arxiv.org/html/2603.11061v1/Threepanelschemati.png)
A new approach combines the power of quantum computing principles with deep learning to significantly improve the prediction of pKa values – critical for understanding protein behavior.