Attached is a snippet function for calculating the constrained minimum of a scalar function J(x) of several variables.

More specifically it solves the problem:

minimize J(x) in the domain xm <= x <= xM

subject to:
g(x)>=0; (a vector function of m potentially nonlinear inequality constraints)
A*x = b; (r linear equality constraints, A being an rxn matrix)

The algorithm has been adapted from this article (in particular from eqn. (7) on page 4):
http://www.ajbasweb.com/ajbas/2011/894-909.pdf.

---

minimizerExamples_v01.sm (39 KiB) downloaded 165 time(s).

I attached a version with function arguments. In the definition, the formal parameter is a function name with the numbers of arguments given, in the call the actual parameter is the function with an UNDEFINED variable as dummy argument, Thus, minimize(J(x), g(x), x ... won't work because x is bound to the initial values vector.

---

minimizerExamples_Kr.sm (38 KiB) downloaded 171 time(s).

@ Martin Kraska:

With your changes (and changing the names in the body of the algorithm as well), it seems to work fine now, and I will soon update both files accordingly.

Thank you very much for your explanations!

mb

The attachments in the first post have been updated according to the suggestions of M. Kraska.