scientific article; zbMATH DE number 4076739
In liner inversion of a finite-dimensional data vector y to estimate a finite-dimensional prediction vector z, prior information about the correct earth model \(x_ E\) is essential if y is to supply useful limits for z. The one exception occurs when all the prediction functionals are linear combinations of the data functionals. We compare two forms of prior information: a `soft' bound on \(x_ E\) is a probability distribution \(p_ X\) on the model space X which describes the observer's opinion about where \(x_ E\) is likely to be in X; a `hard' bound on \(x_ E\) is an inequality \(Q_ X(x_ E,x_ E)\leq 1\), where \(Q_ X\) is a positive definite quadratic form on X. A hard bound \(Q_ X\) can be `softened' to many different probability distributions \(p_ X\), but all these \(p_ X's\) carry much new information about \(x_ E\) which is absent from \(Q_ X\), and some information which contradicts \(Q_ X\). For example, all the \(p_ X's\) give very accurate estimates of several other functions of \(x_ E\) besides \(Q_ X(x_ E,x_ E)\). And all the \(p_ X's\) which preserve the rotational symmetry of \(Q_ X\) assign probability 1 to the event \(Q_ X(x_ E,x_ E)=\infty\). Both stochastic inversion (SI) and Bayesian inference (BI) estimate z from y and a soft prior bound \(p_ X\). If that probability distribution was obtained by softening a hard prior bound \(Q_ X\), rather than by objective statistical inference independent of y, then \(p_ X\) contains so much unsupported new `information' absent from \(Q_ X\) that conclusions about z obtained with SI or BI would seem to be suspect.
- scientific article; zbMATH DE number 4119469
- Softening a hard quadratic bound to a prior pdf — an example from geomagnetism
- scientific article; zbMATH DE number 4024949
- scientific article; zbMATH DE number 1164965
- A distance-based prior model parameterization for constraining solutions of spatial inverse problems
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3807690)