Re: Problem with Maximize and conditions.
- To: mathgroup at smc.vnet.net
- Subject: [mg51100] Re: Problem with Maximize and conditions.
- From: ncc1701zzz at hotmail.com (Nacho)
- Date: Mon, 4 Oct 2004 06:18:29 -0400 (EDT)
- References: <200410020719.DAA26384@smc.vnet.net> <4D224676-146C-11D9-BCA4-000A95B4967A@mimuw.edu.pl> <cjoi8t$ak0$1@smc.vnet.net>
- Sender: owner-wri-mathgroup at wolfram.com
Hello Andrzej.
I suppose that Minimize and NMinimize, as they are new functions,
still have to be debugged and documented, but I expect to see
improvements in future functions.
In any case, it is a good example of a funcion "dangerous" to use if
you don't know a priori an approximate solution, the result may be
wrong.
Regards.
Andrzej Kozlowski <akoz at mimuw.edu.pl> wrote in message news:<cjoi8t$ak0$1 at smc.vnet.net>...
> I have finally managed to get NMinimize return the right answer:
>
>
> NMinimize[{x + y + z, (1/20)*x + y + 5*z == 100, (x | y | z) $B":(B
> Integers, 0 < x < 99, 0 < y < 99,
> 0 < z < 99}, {x, y, z}, Method -> {"SimulatedAnnealing",
> "PenaltyFunction" ->
> (100*(#1 - Floor[#1]) & ), SearchPoints -> 250}, MaxIterations ->
> 500]
>
>
> {43., {x -> 20, y -> 4, z -> 19}}
>
> I am not sure if all the options, particularly MaxIterations, are
> needed. This would need more testing. According to the help
> "DifferentialEvolution" ought to be best for problems involving
> integers, but not matter what I do to it I can't force it to return the
> right answer.
>
> Andrzej
>
>
- References:
- Problem with Maximize and conditions.
- From: ncc1701zzz@hotmail.com (Nacho)
- Problem with Maximize and conditions.