The characterization of natural building stones combines pore segmentation, colour measurement and aging statistics, which open tools usually offer separately. Pore Segmentation Suite is an open-source tool that unifies pore segmentation, colour-palette extraction, aging analysis and journal-ready figure collage in a single Streamlit interface. The segmentation module offers 14 algorithms, from classical thresholding to foundation models (SAM 2, Cellpose), so that the method-dependence of surface porosity becomes visible and auditable. The aging module computes image-derived CIELAB colour differences with paired statistics and a paper-ready interpretation based on metric-specific thresholds. On a Karaman travertine, twelve setup-free algorithms yield image porosities from 0.25% to 23.2%; across twelve specimens of four travertines, the algorithm accounts for 62% of the variance of log-porosity and the specimen for 14%. Against manual annotation of eleven crops, MSER had the highest mean Dice (0.35), but no pairwise difference survived correction for multiple comparisons, and a rule marking the darkest 2% of each image did as well (0.39); two annotators agreed with a Dice of 0.52. At 0.128 mm per pixel, image porosity therefore documents method-dependence rather than replacing a calibrated measurement. On 96 aged specimens, image-derived colour differences rank like contact-colorimeter readings (Spearman ρ = 0.90 over 16 stone × salt groups) but are about half as large, so their position relative to perceptual thresholds depends on the measurement. Every image-derived value reported can be regenerated from the archived images with the archived scripts, and each version of the MIT-licensed tool has a citable Zenodo DOI.
Digitizing a biological sample held in a counting chamber requires stitching overlapping microscope tiles into a whole-slide image. The chamber's etched reference grid (reticle) is a high-contrast periodic structure at the focal plane that remains superimposed on the specimen in the assembled mosaic. Existing stitching tools leave the artifact in place. This paper presents SlidePrep, an open-source pipeline that removes the reticle prior to stitching. It detects grid lines, synthesizes the occluded content via inpainting, and assembles the corrected tiles into a physically calibrated whole-slide image. SlidePrep runs as a command-line tool for batch processing and as a containerized web application that serves the calibrated OME-TIFF alongside interactive OpenSeadragon visualization. A central configuration file governs chamber geometry and detection parameters, so a new acquisition requires editing values, not code. Layout derives from acquisition order and declared overlap, removing dependence on vendor metadata files. This article details the software architecture, functionalities, and its application to a mussel larvae dataset.
Breit-Wheeler pair production, a fundamental high-energy radiative process controlling gamma-ray attenuation and supplying secondary electrons and positrons, is considered. thermal-bw is a lightweight open-source Python package that performs two primary calculations in isotropic photon fields: the absorption of gamma-rays and the energy spectrum and source term of produced pairs. This software is suited for the following spectra: blackbody, greybody, power law, broken power law, tabulated, discrete line, composite, and user-defined. For repeated evaluation, we propose a sub-percent surrogate validated over 0.01≤η≤50 (where η=EγkT/(mec2)2). For general calculations of opacity, adaptive and Gauss-Legendre quadrature are available. Additionally, in calculations of pair energies, we use expressions derived from Böttcher & Schlickeiser together with a separate calculation of the centre-of-momentum frame for largely differing interacting photon energies. thermal-bw may be used in larger radiative transfer, kinetic, or Monte Carlo modelling of high-energy astrophysical environments in which gamma-ray absorption and pair creation have an influence on emerging particle and radiation populations.
Over the past year, continued development of MOOSE (Multiphysics Object-Oriented Simulation Environment) has accelerated considerably. Use of artificial intelligence (AI) to develop software has spurred numerous projects and strongly supported the creation of a Kokkos backend for MOOSE. Collaborations with external institutions have strengthened, notably leading to a massive increase in the capabilities of the MFEM backend. We report here new equation discretization techniques, expansion of available solver methods, user interface improvements, meshing capabilities, and community growth. The framework and its modules support expanding nuclear reactor simulation capabilities, fusion energy device design, geo-mechanics and material science.
Discrete fracture networks (DFNs) are widely used to represent fractured rock masses, but converting different fracture descriptions into conforming computational meshes remains a practical challenge. OpenDFN is an open-source preprocessing tool for generating two-dimensional fracture-conforming meshes from deterministic joints, continuous and discontinuous joint sets, stochastic DFNs, and coordinate-derived fracture polylines. These fracture descriptions can be combined with primitive domains and excavation geometries within a unified Gmsh/OpenCASCADE workflow. Fracture traces are clipped to the host geometry and converted into explicit mesh edges, while parent-child relations are retained to preserve fracture-set identities in the final mesh. Candidate screening and short-segment filtering provide controls for acute intersections, narrow separations, and small clipped fragments. Representative examples demonstrate that OpenDFN can generate meshes while preserving the prescribed fracture geometry and supporting independent mesh-quality assessment. OpenDFN provides a reproducible and inspectable preprocessing route for FDEM and other numerical methods requiring explicit fracture representations.
