Here follows simple polynomial (of order 0, 1 or 2) line-fitting is considered. Example m-files can be found in the SIPPI/examples/case_linefit folder.
First, the forward problem is defined. Then examples of stochastic inversion using SIPPI is demonstrated using a a synthetic data set.
The forward problem consists of computing the y-value as a function of the x-position of the data, and the polynomial coefficients determining the line. sippi_forward_linefit.m:
% sippi_forward_linefit Line fit forward solver for SIPPI
%
% [d,forward,prior,data]=sippi_forward_linefit(m,forward,prior,data);
%
function [d,forward,prior,data]=sippi_forward_linefit(m,forward,prior,data);
if length(m)==1;
d{1}=forward.x*m{1};
elseif length(m)==2;
d{1}=forward.x*m{1}+m{2};
else
d{1}=forward.x.^2*m{1}+forward.x*m{2}+m{3};
end
the forward.x must be an array of the x-locations, for which the y-values of the corresponding line will be evaluated.
Note that the prior must be defined such that prior{1} refer to the intercept, prior{2} to the gradient, and prior{3} to the 2nd order polynomial coefficient.
If only one prior type is defined then the forward response will just be a constant, and if two prior types are defined, then the forward response will be a straight line.
| This site is hosted by sourceforge.net |