Determining error of function value from state estimates

Click For Summary
SUMMARY

The discussion focuses on determining the error in a scalar value produced by a function f() that operates on state estimates derived from an observation model. The user has an error covariance matrix but lacks knowledge of the true state, complicating the calculation of the error in the output of f(). The challenge lies in the function's complexity, which prevents direct input of the covariance matrix. The community suggests providing a clearer explanation or reference to enhance understanding of the observational model concept.

PREREQUISITES
  • Understanding of error covariance matrices in estimation theory
  • Familiarity with scalar functions and their derivatives
  • Knowledge of state estimation techniques
  • Basic principles of probability theory
NEXT STEPS
  • Research the propagation of uncertainty in nonlinear functions
  • Learn about the Delta Method for estimating variances
  • Explore the use of Taylor series expansions in error analysis
  • Study the application of Kalman filters in state estimation
USEFUL FOR

Statisticians, data scientists, and engineers involved in state estimation and error analysis, particularly those working with complex functions and covariance matrices.

hadron23
Messages
28
Reaction score
1
Hello,

I have an observation model that generates a vector of state estimates (of the true underlying states) and a corresponding error covariance matrix.

I have a function called f(), which operates on the state estimate to return a scalar value. Note that I do not know the true state, only the estimates and the error cov. matrix. I am trying to determine the error in the scalar value returned by f() from the information provided in the error covariance matrix.

The function f() is rather messy and operates on components of the state estimate vector, thus I cannot input the matrix into this function.

Any idea how I would determine the error in the output of f() based on the error covariance matrix?

Thanks
 
Last edited:
Physics news on Phys.org
I suggest you explain what you are doing or give a link to an explanation of it. "Observational model" may be a meaningful term in some specialized field, but apparently it doesn't ring any bells for people on the forum familiar with the mathematical theory of probability.
 

Similar threads

Undergrad Scaling covariance for an error calculation
  • · Replies 9 ·
Replies
9
Views
3K
Undergrad Extract error from simulated data
  • · Replies 13 ·
Replies
13
Views
2K
Undergrad What is the correct definition of measurement uncertainty?
  • · Replies 21 ·
Replies
21
Views
3K
Graduate How to combine correlated vector-valued estimates
  • · Replies 11 ·
Replies
11
Views
3K
Undergrad How Do You Estimate Covariance in Physics Experiments?
  • · Replies 3 ·
Replies
3
Views
2K
Undergrad How to Get Final Fisher Matrix from 2 Matrices
  • · Replies 11 ·
Replies
11
Views
2K
Estimating Error w/ Trap and Simpson's Rule when Values are Equal
  • · Replies 3 ·
Replies
3
Views
2K
Graduate Does the Error Distribution Affect Least Squares Estimation in Simple Models?
  • · Replies 1 ·
Replies
1
Views
2K
Undergrad Finding the Largest (or smallest) Value of a Function - given some constant symmetric errors
  • · Replies 3 ·
Replies
3
Views
2K
Undergrad Fitting two models in sequence with parameters in common
  • · Replies 12 ·
Replies
12
Views
3K