Parameter identifiability under approximation (Q3759889)
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 4009739
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | Parameter identifiability under approximation |
scientific article; zbMATH DE number 4009739 |
Statements
Parameter identifiability under approximation (English)
0 references
1986
0 references
parameter identifiability
0 references
Galerkin approximations
0 references
least square problem
0 references
output least square identifiability
0 references
0.8638361692428589
0 references
0.8472009301185608
0 references
0.8176708817481995
0 references
0.8119500279426575
0 references
The paper deals with the parameter identifiability, i.e. the problem of injectivity of the parameter-to-state map, for Galerkin approximations of a linear parabolic equation of the form \(u_ t=(au_ x)_ x\) defined for \(t>0\) and \(0\leq x\leq 1\), with initial condition and homogeneous Dirichlet boundary conditions. In the parameter estimation problem, an infinite dimensional least square problem, designed to actually determine or approximate the parameter a from the observation, is replaced by a sequence of finite dimensional optimization problems using Galerkin approximations, and a sequence of approximating parameters, say \(a_ n\), is thus found. The convergence of \(a_ n\) to a depends also on the parameter identifiability of both the distributed and the discrete problems. While parameter identifiability of distributed parameter systems has been extensively studied in the literature, the present paper seems to be the first one to deal with parameter identifiability of the approximating problem. The paper contains a discussion of the difficulties arising in this investigation, and proposes a quite general method of analysis. A necessary and sufficient condition for the model problem is proved, and some aspects of output least square identifiability are also discussed.
0 references