ABSTRACT
Parameter values in physical models are typically inferred from
data observed in the field. In this work we make the following
assumptions about this inference: (1) the discrete observation vector
contains random variables which have non-necessarily Gaussian
noise
with known covariance matrix cov(
)=
, (2) the physics are well understood and can be represented
linearly by
, and (3) the true parameter values
are not known
but estimated by
which is a vector of random
variables from an unknown non-necessarily Gaussian
prior distribution with cov(
)=
. The parameter estimate
could be from any method where the uncertainty in
is unknown, and we need to determine the covariance matrix
. We will present an approach for estimating a range of
viable matrices
under the above stated assumptions. This
approach is based on the fact that the minimum value of the
cost function