v1.2.3#

We are pleased to announce the release of version v1.2.3.

This is a focused maintenance and performance release that improves numerical robustness, memory efficiency, spatial boundary filtering, and region assignment performance. It also resolves several edge-case bugs in QCM/TCM workflows and cross-function confidence interval calculations.

This release is compatible with and has been tested on Python 3.10 to 3.13.

New Features#

  • Object Filtering by Alpha Boundary:

    • Added the ability to automatically remove objects lying outside alpha-shape boundaries. This provides a cleaner and more principled preprocessing step for spatial analyses where strict boundary adherence is required.

  • Improved Convex Region Identification:

    • Implemented a safer scaling strategy for q-ball distance calculations used in convex region identification, improving robustness across datasets with varying spatial scales.

Improvements#

  • TCM Memory Optimisation:

    • Replaced scipy.spatial.distance.cdist with KD-tree–based neighbour queries in TCM workflows, significantly reducing memory overhead and improving scalability for large datasets.

  • Faster Region Assignment (Hexgrid & Quadrats):

    • Region assignment now uses KD-tree acceleration, resulting in more efficient object-to-region mapping for both hexagonal grids and quadrilateral tiling.

  • Modular Region Generation Refactor:

    • Refactored region-generation logic into a more modular internal structure, improving maintainability and reducing coupling between components.

  • Boundary Filtering Checkpoints:

    • Improved internal boundary filtering logic with additional validation steps to ensure consistent object inclusion/exclusion behaviour.

  • Codebase Maintenance:

    • Tidied imports and removed unnecessary test visualisation components.

    • Added docstrings and licence headers to new helper functions.

    • General internal clean-up for improved consistency and readability.

Bug Fixes#

  • QCM NumPy Slice Hotfix:

    • Fixed a legacy NumPy array interpretation issue affecting slicing behaviour in QCM workflows. This resolves edge cases where array views were incorrectly handled.

  • Include/Exclude Logic Correction:

    • Fixed a bug where include/exclude arguments were incorrectly compared to None when passed as arrays, leading to unintended filtering behaviour.

  • Cross-PCF Confidence Intervals:

    • Corrected computation of confidence intervals for cross pair-correlation functions (cross-PCF), ensuring statistically valid interval estimation.

  • Minor Corrections:

    • Fixed typographical errors in version metadata and QCM-related components.

    • Additional small stability improvements across spatial statistics routines.

Documentation#

  • Added docstrings and licence headers for newly introduced helper utilities.

  • Minor documentation corrections and consistency updates.

This release further strengthens the numerical stability and performance of neighbourhood, boundary, and spatial statistical workflows while addressing important edge-case bugs identified in recent analyses.

Thank you for using muspan!

On this page