"Extended Kalman Filter covariance matrix definition"

Hello,

I am using the Extended Kalman Filter Simulink model for my project. When I run, I get the following warning:

"Matrix is singular to working precision."

The state transition covariance matrix(Q) I defined is eye(4)*0.01, which is not singular. Is there any idea of what the warning means?

Thanks

[Benjamin Thompson]

Might need to see some code to help with this. Q does not have to be non-singular, in fact it can be zero for theoretical analysis (perfect process). Not sure what EKF form you are using but maybe per the link below, the HPH' + R result is singular so that a Kalman gain cannot be computed?

https://en.wikipedia.org/wiki/Extended_Kalman_filter

[M.M]

Thank you for your reply @Benjamin Thompson. Some of the concepts of KF and EKF were confusing to me. The problem has been solved with your help.

[M.M]

Does it make sense to use EKF for linear systems? I used transfer function system identification method for my data. As trnasfer function provide a linear model, I used KF to estimate the states of the system, but the KF could not provide a good approoximation. I tried EKF and it estimated perfectly. I'm looking for a source/paper to see if EKF may be used in linear systems where KF cannot provide a good estimate.