Our Work

Companion catalog to the doctoral thesis Automatización de operaciones en el ciclo de vida de soluciones para fabricación cero defectos (Mateo-Casalí, Universitat Politècnica de València, 2026).Organi…

Companion catalog to the doctoral thesis Automatización de operaciones en el ciclo de vida de soluciones para fabricación cero defectos (Mateo-Casalí, Universitat Politècnica de València, 2026).Organises K3s MLOps solutions by deployment tier (Edge, Platform, Enterprise) and functional category, with deployment questionnaires and explicit mappings to ISO/IEC 42001 Annex B requirements.Version 2.0.0 contains 30 Helm charts, a phased installer, zone NetworkPolicies, an API audit policy, a Helm repository built with its dependencies, and the laboratory validation of the catalog (report, runner and raw evidence in docs/lab-validation/).

Material complementario del artículo "Consumo mediático y efectividad publicitaria en las generaciones Z y Alfa: influencers, eWOM y validación social en el ecosistema persuasivo …

Material complementario del artículo "Consumo mediático y efectividad publicitaria en las generaciones Z y Alfa: influencers, eWOM y validación social en el ecosistema persuasivo digital" (Observatorio OBS*, 2026). Contiene el Anexo 1, con las cadenas de búsqueda ejecutadas en Scopus y Web of Science el 2 de julio de 2025, los filtros aplicados y el flujo de cribado hasta el corpus final de 44 estudios; y el Anexo 2, con la matriz de extracción completa: una fila por estudio con 29 variables de caracterización, los pasajes textuales que sustentan la valoración de completitud del reporte metodológico y la definición de los cinco ítems empleados. Se incluyen versiones en XLSX y CSV, y un README con el diccionario de columnas. El depósito contiene únicamente los datos extraídos por la autora, no los textos completos de los estudios analizados. DOI: 10.5281/zenodo.22947774

🚀 Enhancement Add option to parallelise Lagged-GLM runs (and compute maps at the speed of light - or near that)! #170 (@smoia) Authors: 1 Stefano Moia (@smoia)

OptimaSolver v0.6.1 Diff since v0.6.0 An inner iteration that has stopped converging no longer runs to its cap The inner fixed point of the dual Newton, which recovers the phase compositions from the …

OptimaSolver v0.6.1 Diff since v0.6.0 An inner iteration that has stopped converging no longer runs to its cap The inner fixed point of the dual Newton, which recovers the phase compositions from the multipliers, stopped only when its largest step fell below 1e-14 or after 200 sweeps. On a 107-species cement solved cold, 89 % of its 69 463 calls ran to that cap without approaching the threshold: their final step had a median of 30, the bound on a single update, and the step at sweep 200 equaled the step at sweep 100 — a cycle against the clamp, at trial points whose multipliers ask for compositions that no bounded state satisfies. The iteration now also stops once its step has gone 20 sweeps without reaching a new minimum. A converging iteration reaches a new minimum at every sweep and is unaffected however many sweeps it needs; a stalled one ends above inner_tol either way, so no accepted step depended on the sweeps removed. On that cement, measured with the new benchmark/run.jl: | solve | 0.6.0 | 0.6.1 | |:--|--:|--:| | cold | 142.6 s | 16.8 s | | warm, from its answer | 1.00 s | 0.17 s | Every answer remains certified, and the largest relative change of an amount above 1e-10 mol is 4.1e-13. Added benchmark/ times the certified route on a frozen cement and compares each answer with a recorded baseline; benchmark/sweeps.jl counts the inner sweeps without instrumenting the library. Breaking changes None. This is a patch release: no exported name is added or removed, and no certified answer changes. The iteration paths of the outer Newton do change, which one downstream test had pinned; see below. Downstream ChemistryLab's test/diffuse_layer.jl pinned the size of a failure this release reduces (the eliminated diffuse-layer route above its stiffness limit used to diverge beyond 1e-3; two points now stop at 1.0e-8 and 3.2e-7, still refused). That test now classifies by a margin and passes under both 0.6.0 and 0.6.1. Merged pull requests: Add a package logo (#13) (@jfbarthelemy) Logo and references (#14) (@jfbarthelemy) Move the documentation to DocumenterVitepress (#15) (@jfbarthelemy) Logo and references (#16) (@jfbarthelemy) Perf/inner sweep criterion (#17) (@jfbarthelemy)

A racionalização da terapia empírica em cães com suspeita de ITU, portanto, exige articular diagnóstico, classificação clínica da infecç&…

A racionalização da terapia empírica em cães com suspeita de ITU, portanto, exige articular diagnóstico, classificação clínica da infecção, histórico de exposição a antimicrobianos, distribuição local dos uropatógenos e dados de suscetibilidade, restringindo a utilização de fármacos de amplo espectro às situações em que existam fundamentos clínicos e microbiológicos para tal escolha. O stewardship antimicrobiano oferece estrutura para essa articulação ao propor seleção apropriada do agente, dose, via e duração, acompanhadas por monitoramento, auditoria e revisão das prescrições. Krummenauer (2025), ao avaliar programas brasileiros de gerenciamento do uso de antimicrobianos em unidades pediátricas de terapia intensiva, insere essa prática em uma perspectiva institucional de controle do consumo e qualificação da prescrição, enquanto Roriz (2022) evidencia a inserção do cuidado farmacêutico no acompanhamento terapêutico em ambiente intensivo. Esses elementos, embora oriundos da saúde humana, sustentam uma lógica de gerenciamento compatível com hospitais e serviços veterinários, desde que adaptada às particularidades farmacológicas e epidemiológicas da espécie canina. Viana et al. (2023) reforçam que o uso inadequado favorece a seleção de resistência, e Rodrigues et al. (2023) demonstram que dados locais podem diminuir a incerteza própria da terapia empírica; no âmbito veterinário, Raposo (2025), Quadros (2023) e Sebastião (2023) fornecem o suporte para transportar esse raciocínio ao contexto das ITUs em cães. Assim, o problema que orienta este capítulo concentra-se em determinar de que modo os padrões de resistência e sensibilidade dos principais uropatógenos podem subsidiar escolhas empíricas menos aleatórias e mais ajustadas à realidade microbiológica. Nessa direção, objetiva-se analisar a resistência antimicrobiana nas infecções urinárias caninas, caracterizar seus principais agentes e respectivos perfis de suscetibilidade, examinar a contribuição da cultura e do teste de sensibilidade para a definição da conduta e discutir estratégias de uso prudente capazes de preservar a eficácia terapêutica e limitar a seleção de novos fenótipos resistentes.

Interactive Python / OpenCV tool for manually segmenting and annotating structural cracks and detachments on building facade photos. Produces LabelMe-compatible annotations and training-ready binary m…

Interactive Python / OpenCV tool for manually segmenting and annotating structural cracks and detachments on building facade photos. Produces LabelMe-compatible annotations and training-ready binary masks. Highlights Two tracing tools: cracks as polylines, detachments as closed polygons, with A*-guided edge tracing Automatic building-group detection across photos (SIFT + RANSAC), with crack import and cross-photo consistency checks Training-mask generation that follows the real crack edges, with automatic or manual width control and GSD-based physical width targeting Embedded PDF user manual inside the app (optional, requires PyMuPDF) Companion CLI tools: compare_and_filter_cracks.py and crack_segmentation_evaluator.py Downloads macOS: CrackSegmentation.dmg. The app is not notarized by Apple: on first launch, right-click the app icon and choose Open to bypass Gatekeeper (only needed once). Windows: CrackSegmentation-Setup.exe (rename to match your actual installer filename). Linux (x86_64): CrackSegmentation-linux-x86_64.tar.gz. Extract it, then run: chmod +x CrackSegmentation/CrackSegmentation ./CrackSegmentation/CrackSegmentation Or run from source — see the README for setup instructions. Notes Code comments, docstrings, on-screen (HUD) text, and console messages are all in English. Requires Python 3.10+ if running from source. Developed within the ECHO-TWIN project by the INAF Research Group — see Authors & Credits in the README.

Reproducibility package for an indicator-specific frozen-distribution benchmark of publication volume and realized journal-portfolio indicators. It contains count-level OpenAlex-derived inputs, the co…

Reproducibility package for an indicator-specific frozen-distribution benchmark of publication volume and realized journal-portfolio indicators. It contains count-level OpenAlex-derived inputs, the complete 18-member application family, exact point reconstruction diagnostics, simulation specifications and aggregate outputs, analysis code, and verification records. Raw API responses and credentials are excluded. Component-specific licenses apply: code is licensed under MIT; documentation under Creative Commons Attribution 4.0 International (CC BY 4.0); and derived data under CC0 1.0. See the package README and license files for the scope of each license.

Simulation code and numerical data for rotating-mass Newtonian gravitational communication: source fields, constrained modulation, causal decoding, packet BER qualification, steel and tungsten rotor m…

Simulation code and numerical data for rotating-mass Newtonian gravitational communication: source fields, constrained modulation, causal decoding, packet BER qualification, steel and tungsten rotor mechanics, tone-count studies, arrays, and finite-size magnetic links through conducting barriers. Includes the complete numerical data archive in datasets/. See docs/RELEASE_v0.5.0.md for scope, reproducibility and modeling limitations.

Code release for: K. Basu, A. Guerrero & F. Bertoldi (2026), "Probing submillimeter number counts below the confusion limit: extreme-value statistics of the P(D) distribution and its modulati…

Code release for: K. Basu, A. Guerrero & F. Bertoldi (2026), "Probing submillimeter number counts below the confusion limit: extreme-value statistics of the P(D) distribution and its modulation by gravitational lensing." arXiv:2609.19689

Cleaned, anonymized data supporting, "What Would Happen If Twitter Sent Consequential Messages to Only a Strategically Important Subset of Users? A Quantification of the Targeted Messaging Effect…

Cleaned, anonymized data supporting, "What Would Happen If Twitter Sent Consequential Messages to Only a Strategically Important Subset of Users? A Quantification of the Targeted Messaging Effect (TME)"