sci.op-research

On Apr 10, 5:47=A0pm, "saneman" wrote:
> In each iteration the simplex algorithm visit one of the possible corner
> points. But what is the intuition behind this behaviour and why does it no=
t
> investigate a point somewhere in the middle of a line (assuming the number=

> of variables n =A0 2 <=3D n <=3D 3)?
>
> All corner points can be found by setting (m-n) variables =3D 0 and then
> solving the remaining system. But I still don't see why this is necessary =
a
> corner point.

It has been awhile but corner points are where you reach the limit of
one or more constraints. You cannot get any better without violating a
constraint. In rare cases, there will be line between corner points
that has an optimal value constant across the line. Simplex will
ususally find the first value but unless programmed to look again, it
may miss a near by optimal point of the same value.