Optimization with NMinimize
- To: mathgroup at smc.vnet.net
- Subject: [mg39496] Optimization with NMinimize
- From: Andreas Kabutke <a.kabutke at web.de>
- Date: Thu, 20 Feb 2003 05:13:18 -0500 (EST)
- Sender: owner-wri-mathgroup at wolfram.com
Dear members of MathGroup, i want to optimize my funktion t[u1_,u2_], with the genetic algorithm "DifferentialEvolution" in NMinimize: t[u1_,u2_] is a funktion with fractions, roots, squares and the mathematica operations Table and Sum (but to long to send it to you now). NMinimize[{t[u1,u2], 0 <= u1 <= 4999 && 0 <= u2 <= 4999 && u2 <= u1 && (u1 | u2) \[Element] Integers}, {u1, u2}, Method -> {"DifferentialEvolution", "RandomSeed" -> 0, "PostProcess" -> Automatic, "SearchPoints" -> Automatic, "CrossProbability" -> 0.5, "ScalingFactor" -> 0.6}] A calculation for testing works and find the desired minimium! But on the other hand it shows the following error massages: LessEqual::"nord": "Invalid comparison with 3.5591120000000056*e6 + 6.654961962772207*e-8 \[ImaginaryI] attempted." General::"stop": "Further output of LessEqual::"nord" will be suppressed during this calculation." NMinimize::"fcplx": "The objective function returned a complex value at the point 3.5591120000000056*e6 + 6.654961962772207*e-8 \[ImaginaryI]. Proceeding with the absolute value of the objective function." General::"stop": "Further output of NMinimize :: "fcplx will be suppressed during this calculation." I don`t no in which way i can handle this problem! Are the constraints not strong enough? Do i need a penalty funktion (how i have to implicate it, which one do i need???). I tried it with the Mathematica UnitStep-funktion but it don`t help! I`m realy not a professional in Mathematica...... Thanks for your help... Andreas Kabutke ______________________________________________ Dipl.-Min. Andreas Kabutke Humboldt University Berlin Institute of Physical and Theoretical Chemistry Brook-Taylor-Str. 2, 12489 Berlin, Tel.: ++49-30 - 2093 5556 Fax: ++49-30 - 2093 5559 E-Mail: a.kabutke at web.de