URL: <http://gna.org/task/?7824> Summary: Model parameter ERROR estimation from Jacobian and Co-variance matrix of dispersion models. Project: relax Submitted by: tlinnet Submitted on: Mon 01 Sep 2014 02:31:47 PM UTC Should Start On: Mon 01 Sep 2014 12:00:00 AM UTC Should be Finished on: Mon 01 Sep 2014 12:00:00 AM UTC Category: relax's source code Priority: 3 - Low Status: In Progress Percent Complete: 0% Assigned to: tlinnet Open/Closed: Open Discussion Lock: Any Effort: 0.00 _______________________________________________________ Details: In theory, one can get the Model parameter ERROR, from Jacobian and Co-variance matrix of dispersion models. The error of the dispersion points R2eff, stems from error on signal intensity. The R2eff and error points are used in the dispersion models. If one knows the derivative of a function, one can get the estimated errors as well. http://www.orbitals.com/self/least/least.htm This could get compared to a large number of Monte-Carlo simulations. I relax, Monte-Carlo simulations are made by assuming an gaussian distribution of R2eff points. Estimating the errors from the Jacobian, is essential the same, since the Co-variance matrix is made by populating the weight matrix with the R2eff errors. Therefore, the end result should in principle be the same. The Jacobian matrix are easy to derive for the analytical models. For the numerical solutions, it is a little different, since we then need to take the derivative of a matrix. That can get tricky. In this situation, a numerical gradient solution would probably be better. _______________________________________________________ Reply to this item at: <http://gna.org/task/?7824> _______________________________________________ Message sent via/by Gna! http://gna.org/