Our Work

Este estudo foi desenvolvido a partir da observação de lacunas do conhecimento do paciente acerca do autocuidado e monitoramento do manuseio de drenos.

Code for the mucin pangenome manuscript

Code, generated heuristics, benchmarks and results of the NeurIPS 2026 paper "Hierarchical Task Network Planning with LLM-Generated Heuristics". The paper investigates whether large language models ca…

Code, generated heuristics, benchmarks and results of the NeurIPS 2026 paper "Hierarchical Task Network Planning with LLM-Generated Heuristics". The paper investigates whether large language models can generate effective search heuristics for HTN planning, extending the methodology of Corrêa, Pereira and Seipp (NeurIPS 2025) from classical to hierarchical planning. Using the Pytrich planner on seven standard total-order HTN benchmark domains, it evaluates heuristics generated by nine LLMs under domain-specific prompting against two established domain-independent heuristics in the same planner and against the PANDA planner. A configuration fixed in advance solves more instances than either domain-independent heuristic, the best generated heuristics expand fewer nodes than both on most instances, and the virtual best over all models is on par with PANDA in coverage.

The prevalence of quantum crosstalk is an important barrier to scaling frequency-addressable qubit architectures, with dynamic crosstalk being particularly difficult to detect and suppress. This form …

The prevalence of quantum crosstalk is an important barrier to scaling frequency-addressable qubit architectures, with dynamic crosstalk being particularly difficult to detect and suppress. This form of crosstalk refers to unintended interactions driven by the gate control fields themselves. Here, we minimize dynamic crosstalk using quantum optimal control based on the perfect entangler spectrum, where spectral peaks signal unwanted entanglement with spectator qubits. Focusing on parametric gates in tunable coupler systems, we derive pulse shapes that eliminate dynamic crosstalk. Remarkably, only minimal pulse modifications are required to mitigate the form of crosstalk that is otherwise most difficult to predict. The ability to suppress dynamic crosstalk via the perfect entangler spectrum establishes a generalizable control principle for eliminating unwanted interactions in quantum hardware.

Almost Complete Refactoring einops like native multibackend dispatch jax, torch, tf, mlx, tinygrad, plus keras variable and triton function, and numpy for cpu inference XLA/JIT works if backend provi…

Almost Complete Refactoring einops like native multibackend dispatch jax, torch, tf, mlx, tinygrad, plus keras variable and triton function, and numpy for cpu inference XLA/JIT works if backend provides it gradient available with native autodiff (or explicit vjp for triton) Better numerical stability than original

Recent work in quantum gravity has revealed that closed universes appear to admit only a one-dimensional Hilbert space, implying zero information content. The leading resolution by Harlow, Usatyuk, an…

Recent work in quantum gravity has revealed that closed universes appear to admit only a one-dimensional Hilbert space, implying zero information content. The leading resolution by Harlow, Usatyuk, and Zhao (arXiv:2501.02359, January 2025), featured in Quanta Magazine (November 2025), introduces observers as external additions to restore complexity. We demonstrate that this approach is self-contradictory: it resolves a closed universe problem by opening the universe, introducing boundaries in a system defined by their absence. We present a fundamentally different resolution. By proving that the open/closed distinction is itself a declaration and establishing observer-universe equivalence (O ≡ U), we derive the Totality Theorem: T = O + U = 1. The one-state result is not a paradox but a correct description of completeness: Shannon entropy H = 0 indicates full knowledge, not emptiness. We show that dimensionality itself is an artifact of partition, not a feature of reality. From three relations alone — T = 1, O ≡ U, dO = −dU — and a single structural principle (the Law of Identity A = A generates the binary partition A + ¬A = 1 with unique fixed point A = ¬A = 0.5), we resolve problems across every foundational domain: the unification of the four laws of thermodynamics as facets of a single identity, the black hole information paradox, the cosmological constant discrepancy, the Collatz and twin prime conjectures, Wigner's 67-year mystery of mathematical effectiveness, and the dissolution of Gödel's incompleteness as a property of notation rather than truth.

Genome-wide association study (GWAS) meta-analysis summary statistics for Ewing sarcoma (EwS) susceptibility, from Hubbard et al., “Genome-wide association study meta-analysis identifies suscept…

Genome-wide association study (GWAS) meta-analysis summary statistics for Ewing sarcoma (EwS) susceptibility, from Hubbard et al., “Genome-wide association study meta-analysis identifies susceptibility loci informing Ewing sarcoma etiology and potential mechanisms of risk,” Cell Genomics (2026). https://doi.org/10.1016/j.xgen.2026.101422 Contents: fixed-effects meta-analysis results (Metasoft) from six independent case–control analytic sets (eight studies): 1,727 EwS cases and 9,952 cancer-free controls with at least 80% genetic similarity to European reference samples. The GMKF case–parent trios and cases of other genetic ancestries are not included. Methods summary: genotypes were imputed to the TOPMed r3 reference panel (CESS was imputed to 1000 Genomes phase 3), except SJLIFE, which used whole-genome sequencing. See paper in Cell Genomics for full case-control study inclusion. Logistic regression under an additive model was run in PLINK 1.9 (SNPTEST v2 for CESS), adjusting for study-specific principal components. Results were combined by fixed-effects meta-analysis in Metasoft. Filters: variants with minor allele frequency < 0.01, Hardy-Weinberg P < 1×10⁻⁶ in controls, or imputation quality < 0.3 were removed before meta-analysis. Variants with heterogeneity across studies (Cochran’s Q P < 0.05) were excluded from this file. Format: positions are on GRCh38. Effects are reported for the effect allele (A1) relative to the reference allele (A2). Full column definitions and methods are in the accompanying README. Citation and contact: please cite the article above when using these data. Questions: Aubrey K. Hubbard (aubrey.hubbard@nih.gov); Mitchell J. Machiela (mitchell.machiela@nih.gov).   

Box title: 'Ranked-Choice Voting and Electability' Reference: Kristen Duke, Rachel Gershon, Ivuoma Onyeador, 'Ranked-Choice Voting Increases Support for Underrepresented Leaders by Reducin…

Box title: 'Ranked-Choice Voting and Electability' Reference: Kristen Duke, Rachel Gershon, Ivuoma Onyeador, 'Ranked-Choice Voting Increases Support for Underrepresented Leaders by Reducing Strategic Bias', Management ScienceUpdated file: ResearchBox_2434_updated_2026_11_03.zipUpdated on: 2026-11-03Note: this backup was updated automatically by a ResearchBox bot

The dataset supports the publication “Vibrational signatures and mimicry in ant-termite and termite-termite interactions.” It includes selected samples of raw vibration recordings from ter…

The dataset supports the publication “Vibrational signatures and mimicry in ant-termite and termite-termite interactions.” It includes selected samples of raw vibration recordings from termite and ant footsteps, normalized for scale invariance. The noise level (defined by 3σ bounds) can be used as a scaling factor. The paper fully describes the recording procedure for both vibration and video, enabling replication of the study. One accompanying video is provided to facilitate result replication using F. Wauthier’s video tracking software (Matlab Code for Motion Tracking in Video Sequences [Computer program], 18, 2011), which must be requested from the software author. Alternatively, other video tracking tools such as TRex may be used. All other methods are comprehensively documented in the full paper (DOI provided below). The analysis employs Recurrence Plots (Marwan), with the latest toolbox available from the Potsdam Institute of Climate Impact Research: https://tocsy.pik-potsdam.de/CRPtoolbox/. Additionally, Highly Comparative Time Series Analysis (HCTSA) was applied: https://github.com/benfulcher/hctsa. Abstract to full paper:  Eusocial insects fascinate researchers with their sophisticated communication systems and sensory specialisations. Two groups, ants and termites have coexisted in a long-standing predator-prey arms race, offering insight into the interplay between ecology and evolution. The termite Coptotermes acinaciformis can detect the predatory ant Iridomyrmex purpureus through footstep-induced vibrations, triggering defensive responses. Ants produce noisier walking signatures than termites, while the inquiline termite Macrognathotermes sunteri walks more quietly than its host Co. acinaciformis, suggesting species-specific, ecological vibroacoustic strategies. Using statistical analysis of video-tracked motion and footstep vibrations in confined arenas across six ant and ten termite species, we show that Co. acinaciformis, despite its body size, moves more smoothly than ants, which alternate between directed and erratic paths. Inquiline termites, by contrast, displayed erratic movements. Ants consistently produced stronger vibrations closely linked to body mass. Highly Comparative Time Series Analysis revealed that while Co. acinaciformis and I. purpureus produced distinct vibrational signatures, Mac. sunteri’s signals overlapped with its host, consistent with vibroacoustic mimicry. Although the ecological nature of this association remains unresolved, our findings underscore the central role of vibrational cues in shaping interspecific dynamics and highlight vibroacoustic communication as an underappreciated driver of social insect ecology and evolution.