[comp.software-eng] Brand New Optimization Software

EC13@LIVERPOOL.AC.UK (05/29/91)

The most popular vector optimization methods are: the weighting objective
functions method & constraint method. Although they are good but have some
disadvantages known to people work on that subject.

  Two new vector optimization methods have been developed, based on Shannon's
informational entropy. They have some unique properties which no other
method does:

1) It minimizes the prohibitive level of time needed to generate the
whole Pareto solutions.

2) It gives Pareto solutions arranged in order with respect to one
of the objectives.

   Because of that, discovering the performance-performance space, performance
-design space and design-design space becomes obtainable which is very
important in real-world problems, particularly eng. design problems.



   On the other hand, two new single-criteria minimization methods, also
based on Shannon's entropy. They seek to eliminate the following problems:

(i) The solution obtained is heavily dependent on the starting solution;
and

(ii) The methods often converge to inferior local optima.

All these methods are COMPUTER PROGRAMMABLE and solutions can be generated
automatically.

The four methods have been applied successfully to solve problems in fields
like eng. design, production planning, and mathematics.

A. Sultan

alan@tivoli.UUCP (Alan R. Weiss) (06/01/91)

In article <91149.152946EC13@LIVERPOOL.AC.UK> EC13@LIVERPOOL.AC.UK writes:
>
>The most popular vector optimization methods are: the weighting objective
>functions method & constraint method. Although they are good but have some
>disadvantages known to people work on that subject.

Is it just me, or did I miss some commas in these sentences?

>  Two new vector optimization methods have been developed, based on Shannon's
>informational entropy. They have some unique properties which no other
>method does:
>
>1) It minimizes the prohibitive level of time needed to generate the
>whole Pareto solutions.
>
>2) It gives Pareto solutions arranged in order with respect to one
>of the objectives.
>
>   Because of that, discovering the performance-performance space, performance
>-design space and design-design space becomes obtainable which is very
>important in real-world problems, particularly eng. design problems.

Say what?  Or, for us anglophiles amongst us, "I beg your pardon?"

>   On the other hand, two new single-criteria minimization methods, also
>based on Shannon's entropy. They seek to eliminate the following problems:
>
>(i) The solution obtained is heavily dependent on the starting solution;
>and
>
>(ii) The methods often converge to inferior local optima.

Nothing like converging to inferior local optima to make my day.


>All these methods are COMPUTER PROGRAMMABLE and solutions can be generated
>automatically.

Useful, since Pareto analysis is very dull by hand.

>The four methods have been applied successfully to solve problems in fields
>like eng. design, production planning, and mathematics.
>
>A. Sultan

Would you care to give us examples, or are you content to simply
make the claim and let it lie fallow?  Case studies would be nice.

Forgive me if I sound rude, but I understood not a whit of your
gesticulating vernacular (i.e., I didn't get the point).


Is it just me?

_______________________________________________________________________
Alan R. Weiss                           TIVOLI Systems, Inc.
E-mail: alan@tivoli.com                 6034 West Courtyard Drive,
E-mail: alan@whitney.tivoli.com	        Suite 210
Voice : (512) 794-9070                  Austin, Texas USA  78730
Fax   : (512) 794-0623
_______________________________________________________________________