FABADA Goes MANTID to Answer an Old Question: How Many Lines Are There?

7th Meeting of the Spanish Neutron Scattering Association

Monserrat, D.^1 ; Vispa, A.^1 ; Pardo, L.C.^1,2 ; Tolchenov, R.^2,3 ; Mukhopadhyay, S.^2,4 ; Fernandez-Alonso, F.^2,5

Grup de Caracterització de Materials, Departament de Física i Enginyeria Nuclear, Universitat Politècnica de Catalunya, Diagonal 647, Barcelona, Catalonia

E-08028, Spain^1

ISIS Facility, Rutherford Appleton Laboratory, Chilton-Didcot, Oxfordshire

OX11 0QX, United Kingdom^2

Tessella Ltd., 26 The Quadrant, Abingdon Science Park, Abingdon

OX14 3YS, United Kingdom^3

Department of Materials, Imperial College London, Exhibition Road, London

SW7 2AZ, United Kingdom^4

Department of Physics and Astronomy, University College London, Gower Street, London

WC1E 6BT, United Kingdom^5

关键词: Bayesian fitting;    Least squares methods;    Model Selection;    Optimization algorithms;    Quasi elastic neutron scattering;    Robust inference;    Scientific method;    Spectral components;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/663/1/012009/pdf
DOI  :  10.1088/1742-6596/663/1/012009

来源: IOP

【 摘 要 】

Data fitting and model selection lie at the very heart of the scientific method, and these tasks are often tackled via the use of popular (and typically unquestioned) optimization algorithms of limited validity or flexibility, e.g., least-squares methods. To circumvent these limitations, this work introduces the implementation within the MANTID framework of robust inference methods based on the Bayesian fitting algorithm FABADA. The power of this new MANTID tool is illustrated by revisiting an overarching and recurring question in the interpretation of quasielastic neutron scattering measurements, that is, how many spectral components may (or may not) be justified by the available experimental data.