Question: NLPSolve command: Warning, initial point ignored by solver


I could solve my problem already. I edit in the solution at the end, in case some one ever runs into the same issue, and is happy finding this post when searching for a solution.



Hi there!

I am starting to learn how to use the Optimization package. The help page for the NLPSolve command contains the example

NLPSolve(sin(x)/x, x=1..30)

which would spit out the local minumum at x=23.519… with value -0.042… .

I tried to use the initialpoint option in order to solve for other local minima or maxima, but regardless which initial point choose, I get

Warning, initial point option ignored by solver

and again the same local minimum as before. I entered the option as follows:

NLPSolve(sin(x)/x, x=1..30,initialpoint={x=17})

I am not sure why it doesn't work, since the NLPSolve help page contains an example using the initialpoint option right underneath this example, and there everything works as intended.

Does anyone know what is going on here?




The problem was that NLPSolve uses the quadratic interpolation method by default if the function which should be extremised is univariate and unconstrained. This information can be found under the Optimization/General/Methods help page. This method however, is the only one which doesn't accept an initialpoint. Hence, the solution is to apply a different method, such as

NLPSolve(sin(x)/x, x=1..30,initialpoint={x=17},method=modifiednewton)


Please Wait...