thank you very much to everyone who replied to my question. everyone had
the same clever solution which i was unaware of. here is an example
shown below :
-----Original Message-----
From: Erik Øiolf Sørensen <[log in to unmask]>
To: Mark Leeds <[log in to unmask]>
Date: Tuesday, October 12, 1999 7:53 AM
Subject: Re: a MaxBFGS variation ?
>If you want to constrain to an interval,
>try , for [0,2] say
>
> 2
>-----------,
>1 + exp(-x)
>
>This is much better, than providing a constant, because a locally constant
>function can provide local nonconvexities. (If you want to constrain to a
>different interval
>
> b
>a + ---------
> 1 + exp(-x)
>will do for some suitable a and b.
>
>Best regards
>Erik.
>(I do not post to ox-users because my account is on a non-local machine.)
>
>
>
>On Tue, 12 Oct 1999, Mark Leeds wrote:
>
>> hi : I was just wondering if anyone knows of ( or has ) a version of
MaxBFGS that
>> allows the user to constrain certain parameter ( maybe
>> more than one ) to an interval say [0,2].
>> I tried to do this by modifying my input function with an if statement
that
>> says whenever the parameter is outside of the constrained range, then
>> return a large negative number. Unfortunately, when I do this,
>> I get strong convergence to this large negative number.
>> I can try to look at the MaxBFGS code ( it's probably pretty complex )
>> and make an attempt at modifying it
>> but I just wanted to make sure that noone knows of another version.
>> I know there are tricks using logs and exponentiation
>> to deal with constraints such a [0,infinity] but this is not what I
>> have. I have a more restrictive interval. Thanks a lot.
>>
>>
>>
>> Mark
>>
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|