| Library: | kalman |
| See also: | rlsfil calibrLS |
| Quantlet: | rlsbnorm1 | |
| Description: |
Auxiliary routine for rlsfil:
solves E [ |X-MYw_b(MY)|^2]=(1+e)E [ |X-MY|^2]
- if possible - by numerical integration
for X ~ N(0,Sigt), v ~ N(0,Q) indep. M=Sigt H'(Q+HSigt H')^{-1} Y=HX+v, w_b(x)=min(1,b/|x|) |
| Usage: | {b,ctrl}=rlsbnorm1(e,Sigt,H,Q,N,eps,itmax) | |
| Input: | ||
| e | numeric; efficiency loss to attain; | |
| Sigt | n x n covariance matrix of the initial state | |
| H | m x n matrix | |
| Q | m x m variance-covariance matrix | |
| N | integer; number of grid points for integration | |
| eps | numeric; exactitude | |
| itmax | integer; maximal number of iterations | |
| Output: | ||
| b | numeric clipping height achieving e as rel. effiency loss | |
| ctrl | integer; 1 if error occurs 0 else | |
to be looked up in rlsfil
| Library: | kalman |
| See also: | rlsfil calibrLS |