%0 Conference Paper
%A Daniel, Davis Thomas
%A Bartsch, Christian Hippolyt
%A Bereck, Franz Philipp
%A Scheurer, Christoph
%A Köcher, Simone Swantje
%A Granwehr, Josef
%T IltPy: A python library for inverse Laplace transform of magnetic resonance data
%M FZJ-2025-04081
%D 2025
%X Relaxation and diffusion processes offer rich information about interactions anddynamics in materials and can be correlated to chemical composition and structure. [1]Experimental magnetic resonance data obtained from relaxation and diffusionmeasurements are often analysed by fitting a suitable mathematical function to the datafor extracting underlying relaxation time and diffusion rate constants. However,depending on the compositional heterogeneity, processes may exhibit functionaldependencies which are not governed by a single characteristic parameter but adistribution. Therefore, experimentally measured data may feature a superposition ofdifferent contributions, and their disentanglement becomes challenging by conventionaldata analysis methods. In such cases, inversion algorithms allow for quantitativeanalysis by inverting the data with a suitable kernel, mitigating the need for possiblyambiguous assumptions regarding the number of components or the shape of theunderlying distribution. Herein, IltPy,[2] an open-source python library is introduced forperforming regularized inverse Laplace transforms (ILT) of one- and multi-dimensionaldata. Conventional approaches to ILT of magnetic resonance data require anassumption of signal contributions to be strictly positive. [3] However, this approachsuppresses negative contributions which may be physically relevant, particularly insystems undergoing chemical exchange or cross-relaxation.[4]IltPy implements regularized inverse Laplace transform (ILT) without requiring non-negativity (NN) constraints.[5] Tikhonov regularization in its generalized form is used,and the solution is stabilized with a uniform penalty and a zero-crossing penaltyallowing for extraction of parameter distributions preserving both positive and negativefeatures in the data without preferring one of the signs. IltPy supports user-definedkernels and validity of NN constraint can be tested by comparing inversions with andwithout NN. For large data sets, singular value decomposition is used to compressdata. For experimental data with non-uniform noise or oscillatory features, such asESEEM, IltPy supports weighted inversions. Furthermore, resolution of multi-dimensional data may be improved by regularization of non-inverted dimensions.[2]IltPy is particularly suited for EPR and NMR spectroscopic data. The performance ofthe library is demonstrated using application examples from relaxation and diffusiondata sets, revealing insights into interactions and environments in complex systems.[1] Daniel, D.T. et. al. Phys. Chem. Chem. Phys. 2023, 25, 12767–12776[2] https://apps.fz-juelich.de/iltpy [Accessed 17.04.2025][3] Provencher, S. Comput. Phys. Commun. 1982, 27, 213−227[4] Rodts S, Bytchenkoff D. Journal of Magnetic Resonance. 2010, 205, 315-318.[5] Granwehr, J. et. al. Journal of chemical theory and computation. 2012, 8,3473-3482
%B 46. FGMR Annual Discussion Meeting 2025
%C 15 Sep 2025 - 18 Sep 2025, Bonn (Germany)
Y2 15 Sep 2025 - 18 Sep 2025
M2 Bonn, Germany
%F PUB:(DE-HGF)6
%9 Conference Presentation
%U https://juser.fz-juelich.de/record/1047025