Agents That Remember: Scaling Lifelong Learning in Language Models
New research introduces a comprehensive benchmark for evaluating how effectively AI agents can learn from ongoing experience and refine their own memories.
Science
Prime Numbers Unlock Efficient High-Dimensional Data Mapping
A new deterministic approach leverages the mathematical properties of prime numbers to create robust and efficient vector representations for complex datasets.
Bridging Fidelity and Speed in Simulation
A new hybrid approach combines high-accuracy and reduced-order models to dramatically accelerate complex simulations without sacrificing precision.
Untangling the Latent Space: A New Criterion for Disentangled Representations
Researchers have developed a novel method to encourage clearer, more independent factors in learned representations, improving performance under complex, nonlinear conditions.
Entangled X-rays Reveal Core-Level Secrets
New research demonstrates how the quantum link between emitted X-rays and photoelectrons varies depending on the atom’s inner-shell excitation, opening doors to more precise X-ray spectroscopy.
Cosmic Dawn in Sight: First Data from HERA’s Upgrade
New observations with the upgraded HERA radio telescope are beginning to constrain the earliest moments of cosmic structure formation.
Driven Spins: How Long-Range Interactions Dictate Thermal Chaos
New research reveals how periodically driven spin systems transition from order to thermalization depending on the strength of interactions between particles.
Quantum Magic’s Reach: How Resources Spread Through Complex Systems
![The growth of magic-quantified by $W(t)$-under random brickwork Clifford circuits demonstrates a consistent relationship with entanglement and operator spreading velocities, exhibiting an upper bound defined by [Eq. (S10)] and remaining unaffected by boundary conditions or specific unitary choices within the studied architecture, as evidenced by analysis across $10^2$ circuit realizations with negligible error.](https://arxiv.org/html/2511.21487v1/x8.png)
New research reveals the mechanisms governing the propagation of quantum ‘magic’-a vital resource for computation-within systems governed by specific dynamics.
Fragile Superradiance: The Unexpected Instability of Quantum Light-Matter Coupling
New research reveals that the superradiant phase in the dissipative Quantum Rabi Model isn’t as stable as previously thought, exhibiting a finite lifetime even at large system sizes.
Mapping Movement: The Rise of Trajectory Foundation Models
A new generation of AI is learning to predict and understand motion, opening doors for advanced applications across robotics, autonomous systems, and beyond.