I feel a bit silly to keep replying to myself, but...

On Thu, 31 Mar 2005 13:25:34 -0500, Kohei Yoshida
<[EMAIL PROTECTED]> wrote:
... 
> Consider this scenario:
> 
> A1: = 2*G5*G6 + 3*B3
> G5: = B2
> G6: = 20
> B2 and B3 are decision variables.

This is not a good example either, because to get the coefficient for
the first term, setting B2=1 and B3=0 will just do.  The same for the
2nd term.  So let's scratch this scenario.
 
> Things get really complicated if you consider:
> 
> A1: = G5*(G6 + B3) + B4
> G5: = 20
> G6: = B2
> B2, B3 and B4 are all decision variables.

The same is true with this one.  By setting 1 to each of the decision
variables with all the others set to 0, you can get the coefficient
for each variable.  So, this is not a good counter-example either.

> A1: = G5*(G6 + B3) + B4
> G5: = B5
> G6: = B2
> B2, B3, B4 and B5 are all decision variables.

With this one, since this is not linear, the linearity test should
fail.  I still don't know wether the linearity test can be done by
merely changing the cell values.  But this seems like the only case
where the descent parser may be necessary.  But again, I may still be
over-complicating the problem...

But how about this scenario:

A1: = B2*B2*2 + B3
B2 and B3 are decision variables

By setting B2=1 and B3=0, you will still get 2 as the "coefficient"
for B2, but this is really not linear.

I guess I should re-think my approach a bit.  Meanwhile, I'm open to
any other ideas.


Regards,

Kohei

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to