Multivariate Curve Resolution—Error in Curve Resolution
This article summarizes different methods for uncertainties evaluation in model-free or soft-modeling multivariate curve resolution (MCR) results. We consider two different error sources: unresolved ambiguities and propagation of experimental errors. On the one hand, it is well known that solutions obtained by curve resolution methods present rotational and scale ambiguities, meaning that these solutions are not unique and that a range of feasible solutions fitting equally well experimental data and fulfilling the constraints of the system are possible. On the other, experimental errors and uncertainties are propagated into MCR solutions and it is also important to account for them. Both types of errors, those caused by curve resolution ambiguities and those caused by experimental error propagation, are in fact intermixed in the final curve resolution solutions, and it is often difficult to discern among them. In this article, some proposals to cope with these problems are described and some examples of applications are given.
Main Authors: | , |
---|---|
Other Authors: | |
Format: | capítulo de libro biblioteca |
Language: | English |
Published: |
Elsevier
2020
|
Subjects: | Error propagation, Feasible bands, MCR-BANDS, Multivariate curve resolution (MCR), Rotation ambiguities, |
Online Access: | http://hdl.handle.net/10261/229050 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | This article summarizes different methods for uncertainties evaluation in model-free or soft-modeling multivariate curve resolution (MCR) results. We consider two different error sources: unresolved ambiguities and propagation of experimental errors. On the one hand, it is well known that solutions obtained by curve resolution methods present rotational and scale ambiguities, meaning that these solutions are not unique and that a range of feasible solutions fitting equally well experimental data and fulfilling the constraints of the system are possible. On the other, experimental errors and uncertainties are propagated into MCR solutions and it is also important to account for them. Both types of errors, those caused by curve resolution ambiguities and those caused by experimental error propagation, are in fact intermixed in the final curve resolution solutions, and it is often difficult to discern among them. In this article, some proposals to cope with these problems are described and some examples of applications are given. |
---|