Lindquist's algorithm
[P,R,T]=lindquist(n,H,F,G,R0)
number of iterations.
estimated triple from the covariance sequence of y
.
E(yk*yk')
solution of the Riccati equation after n iterations.
gain matrices of the filter.
computes iteratively the minimal solution of the algebraic
Riccati equation and gives the matrices R
and T
of the
filter model, by the Lindquist's algorithm.
//Generate signal x=%pi/10:%pi/10:102.4*%pi; y=[1; 1] * sin(x) + [sin(2*x); sin(1.9*x)] + rand(2,1024,"normal"); //Compute correlations c=[]; for j=1:2 for k=1:2 c=[c;corr(y(k,:),y(j,:),64)]; end end c=matrix(c,2,128); //Find H,F,G with 6 states hk=hank(20,20,c); [H,F,G]=phc(hk,2,6); //Solve Riccati equation R0=c(1:2,1:2); [P,R,T]=lindquist(100,H,F,G,R0); | ![]() | ![]() |