srfaur — square-root algorithm
[p,s,t,l,rt,tt]=srfaur(h,f,g,r0,n,p,s,t,l)
convenient matrices of the state-space model.
E(yk*yk').
number of iterations.
estimate of the solution after n iterations.
intermediate matrices for successive iterations;
gain matrices of the filter model after n iterations.
may be given as input if more than one recursion is desired (evaluation of intermediate values of p).
//GENERATE SIGNAL
x=%pi/10:%pi/10:102.4*%pi;
rand('seed',0);rand('normal');
y=[1;1]*sin(x)+[sin(2*x);sin(1.9*x)]+rand(2,1024);
//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);
//FINDING H,F,G with 6 states
hk=hank(20,20,c);
[H,F,G]=phc(hk,2,6);
//SOLVING RICCATI EQN
r0=c(1:2,1:2);
[P,s,t,l,Rt,Tt]=srfaur(H,F,G,r0,200);
//Make covariance matrix exactly symmetric
Rt=(Rt+Rt')/2