No label defined (Q4953929)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 1446779
Language Label Description Also known as
default for all languages
No label defined
    English
    No label defined
    scientific article; zbMATH DE number 1446779

      Statements

      0 references
      14 May 2000
      0 references
      model fitting
      0 references
      object-oriented paradigm
      0 references
      dynamic differentiation
      0 references
      parameter estimation
      0 references
      automatic differentiation
      0 references
      0 references
      0 references
      One way to study a set of observed data is to approximate them with a model function, which could be as simple as the sum of several polynomials. Once such a function is provided, the modeler can keep on adjusting various parameters built in the function to reduce the distance between the model and the data, until a preset criterion is met. This model fitting problem is essentially a minimization one, for which there are indeed several solutions, mostly constructed from the numerical analysis perspective. The main contribution of this paper is to suggest an alternative, object-oriented, framework to represent such a model function, via a case study of the nuclear magnetic resonance spectra estimation. The goal is to increase flexibility and user kindness, at a reasonable computational cost. From the object-oriented viewpoint, it is suggested that a function hierarchy collects the basic component functions to hide from the users such details as the number and/or type of basic functions, as well as the respective mathematical expressions, of those functions, when used in constructing the model function; a parameter hierarchy collects all the parameters to hide the interdependencies and their usage among various component functions; and a computationally attractive automatic differentiation algorithm be included to provide various gradients for the distance calculation. Various related technical issues are thoroughly discussed, which culminates with a comparative study, demonstrating that this approach, while meeting all the requirements, is nearly as efficient as the traditional one. The latter fixes the model function at the design time and has it hard coded.
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references