Discussion Forum »
OpsRes Web »
Data Fitting Problem
If I use starting values for a, b and c, how will I know that the solution is a global optimum?
--
TWikiGuest - 29 Mar 2008
You won't without doing a good investigation of the objective function. However, you should try several different starting points too see if you come up with different local minima.
--
MichaelOSullivan - 30 Mar 2008
For some of the model functions you will need to add constraints to make sure that the solutions stay feasible, e.g.,
![$x - b \geq 0$](/pub/Forum/ForumOpsRes0005/latex327437a642e8e8ee04f70c18f25b6834.png)
(or some small value such as
![$1^{-10}$](/pub/Forum/ForumOpsRes0005/latex1bb6432509eec6eb357f7a6f0c795b24.png)
).
--
MichaelOSullivan - 30 Mar 2008
How do we evaluate a function in AMPL, for the objective function?
--
TWikiGuest - 31 Mar 2008
--
MichaelOSullivan - 01 Apr 2008