CompPhotoChem/shnitsel-tools: 2026.5.1

Read full text Summarise with VIRAT
Authors: Reshief, Theo, Carolin Müller, SHNITSEL, CPC Group, rocu03, t-werner02, Julia Thuy Duong Ngo
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Bug fixes compared to previous version and addition of new features: Breaking Changes To keep the structure of shnitsel clearer, we moved the PCA, LDA, PLS, TICA and all dimensionality reduction related methods into the shnitsel.analyze.dimred submodule whereas they were previously located directly in shnitsel.analyze Features State-based and label-based coloring suppport in dimensionality reduction plots via biplot_kde() Tentative support for Linear Discriminant analysis (LDA) on datasets Initial support for Time-lagged independent component analysis (TICA) to analyze time-correlation between features. Not yet fully integrated into the default dimensionality reduction framework in shnitsel.analyze Added accessors for creating timeplots and performing pca analysis to shnitsel.xarray support. Improved VMD visualization support Add generic debug information and version information support with shnitsel.__version__ and shnitsel.show_debug_info() to help with error debugging support Implement focus_hops() and hops() functionality to obtain more detailed analysis tools related to hop-relative statistics within trajectories Add reading of state_coeffs_diag from sharc trajectories, i.e. the coefficients associated with the diagonal basis states. Add reading of prob_hop_diag (hopping probability in diagonal basis) and u_matrix (Basis transform matrix from diagonal to MCH state coefficients) from SHARC data. Stacking and unstacking trajectories will now try and restore original types like int for the active state. Harmonized fill values for layering/unstacking of data to avoid integer type fields being converted into float to allow for np.nan padding. Setting a desired fill value can be achieved by setting fill_value attribute. Note that this value must be serializable or storing the dataset on disk will become impossible. Add an unwrap_dataset() helper method to strip away bare xarray data types to help with testing and compatibility with other frameworks possibly checking for precise types. Extended traj_vmd() support for exporting trajectories to be displayed using vmd. Added on-demand padding with temporary dimensions for rotational alignment in shnitsel.geo.alignment functions to allow provision of single geometries. Add RMSD calculation support with calc_rmsd in shnitsel.analyze.rmsd. Fixes Icons and CSS files for tree visualization were not included in the initial release package version on pip Colorbar label was not set in biplot_kde() for time coloring property Imports of SHARC formats did not correctly read and set kinetic energy in imported dataset Fix broken imports in shnitsel.vis.support module Fix broken logic in hop_mask_from_active_state() function Fix construction of default_mol() if mol information is stored in atXYZ variable of a dataset for legacy support. Fix issues arising from too strict constraints on different ShnitselDataset container types leading to issues when filtering data. Fix state selection and other state-related operations failing if the state dimension was reduced to a scalar coordinate during filtering. Fix NewtonX import failing due to bad type specifier for atom type names Fix sanity_check() failing on stacked trajectories due to bad invocation of .cumprod() Fix plots failing due to change in naming conventions. Fix issues with constructing default molecular structures from layered trajectories Fix filtering for physical sanity not working on layered trajectories Fix multi-index methods failing on layered trajectories. Will now simply not modify the trajectory if they cannot be applied and print out a warning. Fix repeated stacking and unstacking not correctly filtering out padding entries in data. Fix conversion between stacked and unstacked wrappers not using the correct conversion functions to filter out invalid entries due to padding. Fix sanity_check() breaking for layered inputs. Fix StateSelection construction failing if no state info present on input data. Known issues Selection of pyramids via SMARTS patterns where StructureSelection syntax is allowed is currently not working as intended For some collections of trajectories, as_stacked() or other combiner methods may break if there are empty trajectories, e.g. after sanity_check()

Fields of Study & Subjects

Advanced Chemical Physics Studies Spectroscopy and Chemometric Analyses Computational Physics and Python Applications

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
20 March 2026
Language
EN