minimization routine with bounds

[dsmanoli]

Hi,

Can someone tell me where I can find a routine to minimize a function of several variables with simple bounds? I am looking for a routine that is based only on function evaluations; I do not know the derivatives for the function and the function may be non-smooth.

For example, I am trying to minimize a function f(x1,...xN) where 1 < x1 <= 2 ... 0 <= xN < 4 and f(--) returns a scalar value.

I know that IMSL has routines for this, but I am trying to find the source code for such a routine.

Please let me know if you have any suggestions.

Thanks,
Day

[DavidWhite]

Quoting - dsmanoli@econ.ucla.edu

Hi,

Can someone tell me where I can find a routine to minimize a function of several variables with simple bounds? I am looking for a routine that is based only on function evaluations; I do not know the derivatives for the function and the function may be non-smooth.

For example, I am trying to minimize a function f(x1,...xN) where 1 < x1 <= 2 ... 0 <= xN < 4 and f(--) returns a scalar value.

I know that IMSL has routines for this, but I am trying to find the source code for such a routine.

Please let me know if you have any suggestions.

Thanks,
Day

Suggest you look in netlib.org; or Alan Miller's site as starters.

[anthonyrichards]

Quoting - David White

Suggest you look in netlib.org; or Alan Miller's site as starters.

There is also the CERNLIB 'MATHLIB' library available as a download from CERN ( see http://cernlib.web.cern.ch/cernlib/), from which these routines are available:

http://wwwasdoc.web.cern.ch/wwwasdoc/shortwrupsdir/d501/top.html

[Jeppe_G_]

Hi

A good starting point might be Numerical Recipies (www.nr.com) - the routines are fairly simple, but works well for small problems. For a more comprehensive code I will suggest that you look at www.coin-or.org, this is a really good solver based on open-source.

Best regards
Jeppe Grue

[bendel_boy1]

Quoting - jeppe.grue@vattenfall.com

Hi

A good starting point might be Numerical Recipies (www.nr.com) - the routines are fairly simple, but works well for small problems. For a more comprehensive code I will suggest that you look at www.coin-or.org, this is a really good solver based on open-source.

Best regards
Jeppe Grue

Always remember the optimisation routines decision guide:

http://plato.asu.edu/guide.html

[anthonyrichards]

CERLIB is free and you can access the source code.