Search papers, labs, and topics across Lattice.
100 papers published across 4 labs.
Fast elastic collisions and reduced collisional loss in ultracold polar molecules could revolutionize quantum simulation and impurity physics.
einx transforms tensor programming by providing a universal, readable notation that eliminates shape errors and simplifies complex operations.
Ensemble consensus training not only boosts predictive accuracy in molecular graphs but also allows individual models to outperform traditional ensembles, reshaping our approach to semi-supervised learning in this domain.
Suppressing two-body losses by orders of magnitude opens the door to exploring itinerant quantum magnetism and novel quantum droplets in polar molecule systems.
Achieving up to 96% filling fractions in molecular tweezer arrays could revolutionize quantum simulations and computing by enabling scalable molecular systems.
GNNs can revolutionize spin dynamics simulations by learning effective magnetic force fields that match electronic calculations with unprecedented efficiency.
Claim-centered retrieval in AskChem enables researchers to achieve 100% resolvable DOIs, revolutionizing how chemistry literature is synthesized.
Intrinsic physical consistency outperforms noisy supervised learning in predicting atomic structures, achieving unprecedented accuracy without the need for costly ground-truth labels.
Spending a fixed oracle budget more selectively can lead to substantial improvements in molecular optimization outcomes without incurring additional costs.
Enforcing an AMG-style hierarchy in graph neural preconditioners can significantly enhance convergence for large-scale simulations, but it also risks introducing computational overhead.
Query complexity for bouncy particle and Zigzag samplers can be drastically reduced using windowed thinning, achieving efficiency gains that scale with the condition number.
Boundary conditions can critically determine whether a density/flux pair can support a regular Lagrangian flow, revealing potential pitfalls in ODE-based sampling for reflected diffusion models.
Learning features from past solutions can slash Newton's method iterations and drastically speed up convergence for nonlinear PDEs.
einx transforms tensor programming by providing a universal, readable notation that eliminates shape errors and simplifies complex operations.
Target-conditioned abstractions can boost scientific inspiration retrieval accuracy by over 10%, transforming how we leverage past research for new hypotheses.
PerturbMap achieves a 4.1% reduction in prediction error for single-cell perturbation responses, significantly enhancing cross-context transfer accuracy.
DivAlign reduces community-level redundancy in research ideation by tailoring suggestions to individual researcher profiles, achieving a 12% drop in similarity scores without sacrificing relevance.
Repeated stimulation in human cortical organoids can reduce network engagement from 93% to just 10%, revealing critical insights about neural circuit maturation and response dynamics.
GPU-accelerated processing of irregular data structures can now achieve unprecedented throughput, transforming high-energy physics analyses.
TRIO achieves a remarkable reduction in imaging error, enhancing PET diagnostics by leveraging the untapped potential of three-photon annihilation events.
Achieving a 60% reduction in safety-gate intercepts, LabEvolver redefines the capabilities of wet-lab agents without the need for traditional training.
Spectral changes in diatomic molecules under strong magnetic fields reveal unexpected phenomena like bond stiffening and symmetry-breaking, challenging existing theoretical frameworks.
SQD achieves near-exact energy and gradient calculations, making quantum-enabled molecular dynamics practical for real-world applications.
Performance differences among DMRG implementations can reach up to 100x, revealing critical insights for researchers in quantum computing.
Excitation wavelength is a critical control parameter for optimizing photo-CIDNP, revealing unexpected regions of hyperpolarization that challenge conventional understanding.
Equivariant MLIPs can achieve leading inference speeds while maintaining high accuracy, revolutionizing computational materials science workflows.
Nonclassical nucleation pathways emerge under high supersaturation, challenging the applicability of classical nucleation theory and revealing critical insights into cluster dynamics.
Achieving over 98% reduction in circuit complexity while retaining quantum precision could revolutionize resource management in quantum simulations.
Segmentation metrics that ignore bifurcation connectedness can lead to dangerous misdiagnoses in coronary artery disease treatment.
Landmark shape spaces can now be equipped with metrics that preserve shape invariances while preventing collisions, revolutionizing how we approach shape analysis in geometry.
DR-FRL reveals that scalar summaries can often suffice for causal inference, challenging the need for complex functional representations in many cases.
Machine learning outperforms traditional algorithms in tracing Seiberg dualities, revealing new efficiencies in theoretical physics computations.
Ensemble consensus training not only boosts predictive accuracy in molecular graphs but also allows individual models to outperform traditional ensembles, reshaping our approach to semi-supervised learning in this domain.
A unified benchmark and a physics-informed framework that together redefine the standards for reliable chemical property predictions in AI.
Achieving 30-40 FPS for time-varying implicit neural volumes, this framework transforms interactive rendering by making it both fast and high-fidelity.
Achieving dimension-independent approximation bounds for fractional parabolic PDEs could revolutionize how we model complex dynamical systems with neural networks.
Transforming passive nanoparticle networks into tunable nonlinear systems reveals new pathways to enhance computational efficiency in neuromorphic computing.
UNICON achieves specialist-level performance in previously unseen disciplines without retraining, redefining the boundaries of numerical intelligence in AI.
MarkushGlyph revolutionizes the parsing of complex chemical structures by processing them in a single pass, significantly improving accuracy over traditional multi-stage approaches.
Memory in electronics can now evolve like in nature, using low-resource and low-energy methods to store sensory data in physical structures.
Current allocation practices mismeasure productivity, prioritizing occupancy over actual utilization, which could undermine the effectiveness of UK research funding.
Mastering periodic quantum chemistry is now more accessible, with practical strategies to tackle complex Coulomb interactions and optimize computational costs.
The Kinesin molecular motor exhibits the Mpemba effect, revealing that systems further from equilibrium can relax faster than those in equilibrium, challenging conventional thermodynamic expectations.
Mo2 molecules can trap light and generate intense quantized electromagnetic fields, revolutionizing quantum optical experiments in free space.
A groundbreaking collection of 16 schemas that standardizes scientific process descriptions, paving the way for enhanced reproducibility and automation in research.
Despite LLMs excelling at identifying reviewer concerns, they falter in verifying if revisions truly resolve those issues, with the best achieving only a 0.501 score in evidence-based checks.
Life-science AI agents can now access literature more efficiently, achieving over 16-point improvements in citation accuracy through natural language queries.
By mimicking fruitfly sensory processing, this method achieves efficient regression with reduced computational overhead, transforming how we approach nonlinear dynamical systems.
A field-code compiler can turn approximate transport fields into precise value-certified samplers, dramatically enhancing communication efficiency in empirical optimal transport tasks.
Satellite scatterometers can outperform traditional models in short-term offshore wind forecasting, achieving up to 23% lower error in just hours.
PIKS achieves universal consistency for physics-informed learning, outperforming traditional methods while maintaining analytical tractability.
PTT transforms EBM training by enabling practical equilibrium maximum-likelihood training that consistently yields superior sample quality and robustness.
CostAda achieves top-tier discovery quality while using at least 50% less budget compared to existing methods, revolutionizing how we approach resource allocation in LLM search processes.
CURL effectively harnesses LLMs to stabilize CATE estimation, leading to improved performance in personalized interventions across multiple benchmarks.
Unsupervised subgrouping methods can yield interpretable policy insights, but the lack of statistical significance underscores the challenges in drawing firm conclusions from observational data.
The implementation lottery reveals that relying on a single run can mislead research conclusions, with winner reversals occurring in up to 43.6% of cases.
SciFigAlign achieves a remarkable 59% reduction in error over traditional LLM-based scoring methods by grounding figure assessments in manuscript context.
Generative LLMs can significantly reduce the manual effort required for literature retrieval and screening, transforming how researchers access scientific knowledge.
EvoPINN autonomously discovers new algorithms for physics-informed neural networks, achieving significant performance improvements while ensuring scientific validity.
Coupled multi-output symbolic regression can enforce cross-output consistency, revealing hidden relationships that independent methods miss.
PUDA revolutionizes self-driving laboratories by enabling AI agents to autonomously execute experiments with complete data provenance, bypassing the limitations of traditional graphical interfaces.
General LLMs struggle with enzyme classification, but leveraging external knowledge can dramatically improve their performance, revealing hidden gaps in reasoning capabilities.
State-of-the-art multimodal models excel in classification but falter in extracting critical data from scientific figures, revealing a significant gap in AI's reasoning capabilities.
Geometry-derived features in HERMES boost N-stage accuracy by 4.3% while simplifying the model's complexity.
QC-DFET achieves unprecedented accuracy in modeling catalytic surface reactions, bridging quantum computing and surface chemistry.
Isolating polymer signals from overwhelming large-mass scatterers reveals a universal osmotic equation of state, transforming our understanding of polymer thermodynamics.
ba-occ-DFT reveals that even with a zero DFT band gap, we can achieve accurate defect level predictions in narrow-gap semiconductors.
Q-Steer boosts molecular optimization rewards by leveraging action-value estimates during rollout, leading to consistent performance improvements across various models and optimizers.
Hybrid workflow strategies can boost throughput by 3.8x while slashing network overhead by nearly 15x, revolutionizing data processing efficiency.
Elastic effects can dictate microphase separation in elastomers, revealing a surprising link between stiffness and phase transition dynamics.
The longevity of nanobubbles and pristine emulsions defies classical theory, suggesting a revolutionary role for structured water layers in colloidal stability.
Large language models can uncover hidden relationships in atomic structures, revealing that even seven-neighbor clusters can yield indistinguishable descriptors.
Achieving over $10^{-2}$ eV accuracy in Rydberg state modeling with a significantly smaller Gaussian basis set could revolutionize ab initio studies of complex atomic and molecular systems.
Modus achieves competitive performance across diverse benchmarks by treating all modalities symmetrically, eliminating the need for modality-specific heads or pipelines.
SpectONet achieves unprecedented accuracy in predicting beam vibrations, outperforming conventional methods by over 64% on synthetic datasets through innovative sensor placement strategies.
Clay-straw adobe emerges as the top performer in thermal efficiency, but the material ranking flips under specific outdoor conditions, revealing critical insights for sustainable building practices.
Eliminating backpropagation in physics-informed learning could revolutionize how we solve partial differential equations efficiently.
Clustering patients based on disease trajectories reveals hidden patterns that could transform personalized medicine approaches.
Spectral descriptors can effectively guide CNN architecture design, leading to significant performance improvements in NIR chemometrics.
Quantum estimators can dramatically reduce query complexity in stochastic optimization, achieving optimal performance even in low-dimensional settings.
Deep learning isn't always the answer; classical PCA can outperform complex models in certain single-cell clustering scenarios.
Normalizing flows can accurately reconstruct parton distribution functions while respecting physical constraints, even from sparse data.
High binary recognition performance in AMP models fails to predict real-world assay outcomes, revealing critical gaps in current evaluation methods.
Eigenvalues align in Riemannian manifolds, but the dominant eigenspaces reveal surprising differences that could redefine how we analyze shape data.
Achieving a mean absolute error of just $7 \times 10^{-3}$, this neural operator revolutionizes how we simulate EUV lithography by drastically cutting computational costs and enhancing accuracy.
Vilya-2 redefines peptide structural modeling, achieving unprecedented accuracy and generalizability that outstrips traditional methods.
OmniQEC uncovers quantum error-correcting codes that outperform established benchmarks, paving the way for more efficient fault-tolerant quantum computing.
Functionalization can lead to chemically distinct changes in electronic structure, revealing critical insights into lithium-metal electrolyte behavior.
Contributor migration patterns reveal that quantum software concepts often transcend individual repositories, shaping broader ecosystems.
Suppressing two-body losses by orders of magnitude opens the door to exploring itinerant quantum magnetism and novel quantum droplets in polar molecule systems.
Real-time dynamics of large bosonic condensed phases can now be simulated directly, overcoming significant limitations of previous methods.
Achieving up to 96% filling fractions in molecular tweezer arrays could revolutionize quantum simulations and computing by enabling scalable molecular systems.
Fast elastic collisions and reduced collisional loss in ultracold polar molecules could revolutionize quantum simulation and impurity physics.
Solid-harmonic integral engines cut the size of Coulomb tensors while preserving crucial angular properties, revolutionizing many-electron computation efficiency.
Irreversible structural disorder in β'-Mn3(PO4)2 emerges under pressure, challenging our understanding of phase stability in complex materials.
Achieving complete deoxygenation of graphene oxide while preserving its structural integrity hinges on the delicate balance of photon and electron doses during laser reduction.
Microscopic transition processes in non-adiabatic electron transfer can take longer than previously thought, affecting overall reaction times significantly.
Mobile ions have a surprisingly minor effect on the steady-state performance of efficient perovskite solar cells, challenging previous assumptions about their role in device efficiency.
AI-generated project proposals can match human quality, but AI reviewers show a bias towards favoring their own outputs.
AI-generated literature reviews miss the mark, with less than 6% overlap with expert selections, but the 2026 model shows promise for error-free performance.
Resting heart rate from wearable devices reveals a striking socioeconomic gradient, with users in high-hardship states averaging 1.33 bpm higher than those in low-hardship states.
Confidential bidding in decentralized manufacturing can be economically feasible, but only on certain blockchain platforms like Arbitrum and opBNB.
Achieving 98.1% accuracy in IUPAC name generation could revolutionize how chemists and researchers communicate molecular structures.
Achieving high-quality eddy-current distortion correction in diffusion MRI with deep learning reduces processing time from hours to seconds.
Federated learning can achieve superior accuracy in brain tumor modeling without compromising patient privacy, outperforming traditional methods by leveraging local data insights.
