Here are some examples of potential uses :

1.
There are N objects with some properties (weight, volume, cost, etc.).

There are M containers for these objects. The sum of the properties of
objects enclosed in each container is limited.

All M containers can't hold all N objects.

The task is to distribute of N objects into all M containers with highest
sum of properties in each.

2.
There are number of substances with some properties.
The task is to mix any substances with minimum/maximum of one of properties
with the other properties being equal to a certain value.

3.
The are some points placed in space. The task is to link every point with
N other points with a minimum or maximum sum of the lengths between linked points.

4.
N potential securities have a certain income and risk. The task is to invest
some money to receive certain income with minimum risk.

i.e.  .........................................................

The algorithm was measured at the following calculation speed.

On a Pentium-100 computer with 16M RAM the linear programming task for 350x700 dimension
was resolved in about 10 minutes. The task with dimension 400x800 has been resolved in 18
minutes.

Note that no accuracy was sacrificed during solution.

If you have any questions or any similar tasks, you are welcome to discuss.
The algorithm been tested only for 2 tasks and I'm interested in testing it on
other real world problems.

Thanks to Loren Christensen for corrections to my English.