ALPSCore reference

#include <propagation.hpp>
Public Member Functions  
linear_prop ()  
linear_prop (double dx)  
double  dx () const 
Perform linaralized error propagation by estimating the Jacobian.
Given a transformation f
and a random sample X
, estimate the propagated uncertainties by performing a Taylor series and keeping the linear term:
Cov[f(X)] = df/dX Cov[X] (df/dX)^T + O(d^2f/dx^2)
where df/dX
is the Jacobian of f
at X
, as estimated by finite differences of dx
. This procedure is exact for linear transformations; for nonlinear transformation, it will introduce bias.
Definition at line 43 of file propagation.hpp.

inline 
Definition at line 45 of file propagation.hpp.

inline 
Definition at line 47 of file propagation.hpp.

inline 
Definition at line 49 of file propagation.hpp.