Rollplex: Cross-Phase GPU Spatial Sharing for Vision Language Model Post-Training

Posted by klaurent Caltech

Kavli Affiliate: Wei Gao| Summary:Vision-language models (VLMs) enable embodied agents to reason and act from visual observations and language instructions. Reinforcement learning (RL) post-training enhances these capabilities using task feedback, but current on-policy RL runtimes execute rollout, reference scoring, and actor training in strict serial phases. While effective for text-only RL, this phase-granular execution is […]


Continue.. Rollplex: Cross-Phase GPU Spatial Sharing for Vision Language Model Post-Training

CoDS: Robust Collaborative Perception via Expert-driven Detection and BEV Segmentation

Posted by klaurent Caltech

Kavli Affiliate: Wei Gao| Summary:Collaborative perception breaks through single-view limitations via multi-agent information exchange. However, multi-source noise such as pose errors and communication delays degrades fusion feature quality, constraining perception performance. Joint training of detection and BEV segmentation provides a natural remedy, where segmented road regions help constrain target distributions and detection bounding boxes help […]


Continue.. CoDS: Robust Collaborative Perception via Expert-driven Detection and BEV Segmentation

Improved Measurement Cost Scaling in the Nonorthogonal Quantum Eigensolver

Posted by klaurent Berkeley

Kavli Affiliate: Birgitta Whaley | Summary:Quantum subspace diagonalization methods are promising algorithms for quantum chemistry on near-term quantum computers. These methods can estimate low-lying energies of molecular systems using shallow quantum circuits, at the cost of many circuit repetitions to estimate the projected matrix elements. Errors in these matrix elements can be converted into much […]


Continue.. Improved Measurement Cost Scaling in the Nonorthogonal Quantum Eigensolver

Floquet Quasiparticle Poisoning of Frozonium

Posted by klaurent Cornell

Kavli Affiliate: Debanjan Chowdhury | Summary:Periodic driving can suppress the Josephson nonlinearity of a fluxonium superconducting circuit, producing a nearly harmonic Floquet spectrum at isolated freezing points [K. Lewellen et al., Newton 2, 100434 (2026)]. Here we show that this dynamically frozen behavior does not generically suppress quasiparticle-induced dissipation in the resulting frozonium circuit. We […]


Continue.. Floquet Quasiparticle Poisoning of Frozonium

CIBER $times$ galaxy cross-correlations reveal a bright, low-redshift NIR background

Posted by klaurent Caltech

Kavli Affiliate: James Bock | Summary:We perform the first tomographic analysis of near-IR extragalactic background light (EBL) anisotropies, cross-correlating CIBER 1.1 and 1.8 $μ$m imager data with photometric galaxy catalogs from DESI Legacy Survey DR8 and Hyper-Suprime-Cam Ultra-Deep Survey. We measure significantly higher cross-power than expectations from an integrated galaxy light (IGL) model on scales […]


Continue.. CIBER $times$ galaxy cross-correlations reveal a bright, low-redshift NIR background

Strain-controlled magnetism and magnetoelasticity in monolayer NiPS$_3$ and CrPS$_4$

Posted by klaurent Delft

Kavli Affiliate: Herre Van Der Zant | Summary:We develop a first-principles framework for magnetoelastic coupling in two-dimensional magnets based on a strain-dependent Heisenberg model. In this approach, strain derivatives of the exchange interactions provide direct access to magnetostriction and to the magnetic renormalization of the elastic tensor, establishing a microscopic link between spin interactions and […]


Continue.. Strain-controlled magnetism and magnetoelasticity in monolayer NiPS$_3$ and CrPS$_4$

Early Exploration of the Scientific Discovery Space for the Habitable Worlds Observatory

Posted by klaurent Caltech

Kavli Affiliate: Dimitri Mawet| Summary:The Habitable Worlds Observatory (HWO) is a future NASA flagship mission concept identified by the Astro2020 Decadal Survey as the highest priority for large space missions. HWO should conduct "transformative astrophysics" and search for biosignatures in the atmospheres of approximately 25 potentially Earth-like planets. To further the early-stage development of HWO, […]


Continue.. Early Exploration of the Scientific Discovery Space for the Habitable Worlds Observatory

CausalSplat: Towards Comprehensive Hierarchical Reasoning in 3D Gaussian Splatting

Posted by klaurent Caltech

Kavli Affiliate: Wei Gao| Summary:While 3D Gaussian Splatting (3DGS) has advanced open vocabulary scene understanding, existing methods remain confined to explicit queries. They struggle to interpret implicit intents, complex spatial constraints, and commonsense reasoning required for practical embodied interactions. To address this gap, we introduce the task of reasoning 3D Gaussian segmentation and construct two […]


Continue.. CausalSplat: Towards Comprehensive Hierarchical Reasoning in 3D Gaussian Splatting

CausalSplat: Towards Comprehensive Hierarchical Reasoning in 3D Gaussian Splatting

Posted by klaurent Caltech

Kavli Affiliate: Wei Gao| Summary:While 3D Gaussian Splatting (3DGS) has advanced open vocabulary scene understanding, existing methods remain confined to explicit queries. They struggle to interpret implicit intents, complex spatial constraints, and commonsense reasoning required for practical embodied interactions. To address this gap, we introduce the task of reasoning 3D Gaussian segmentation and construct two […]


Continue.. CausalSplat: Towards Comprehensive Hierarchical Reasoning in 3D Gaussian Splatting

Assessing fidelity-limiting factors and achieving single-qubit gate fidelity beyond 99.999% in driven silicon spin qubits

Posted by klaurent Delft

Kavli Affiliate: Giordano Scappucci | Summary:In semiconductor single-spin qubits, high-fidelity quantum gates have been demonstrated; however, achieving consistent performance remains challenging due to variations in driven qubit coherence, which is less explored than free-evolution coherence such as $T_2^*$. Here, we report single-qubit gate fidelities above 99.999%, achieved by dramatically extending the driven-spin coherence time and […]


Continue.. Assessing fidelity-limiting factors and achieving single-qubit gate fidelity beyond 99.999% in driven silicon spin qubits