home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!think.com!mintaka.lcs.mit.edu!ai-lab!great-grain!mdlm
- From: mdlm@great-grain.ai.mit.edu (Michael De-la-Maza)
- Newsgroups: sci.math
- Subject: Novice ?: Availability of LP programs and absolute value tricks
- Date: 21 Jan 1993 22:11:00 GMT
- Organization: MIT Artificial Intelligence Lab
- Lines: 47
- Sender: mdlm@great-grain (Michael De-la-Maza)
- Distribution: world
- Message-ID: <1jn71kINNn04@life.ai.mit.edu>
- NNTP-Posting-Host: great-grain.ai.mit.edu
-
- I have a couple of questions on the subject of linear (and slightly
- nonlinear) programming.
-
- 1. Does anyone know where to get free linear programming programs?
- An ftp site would be best.
-
- 2. I know that there are specific results for the case in which the
- function to be maximized is quadratic. Are there specific results
- for the case in which the constraints can be quadratic? Any programs
- that implement these (quadratic)^2 algorithms?
-
- 3. Does anyone know of a reference that summarizes tricks for getting
- rid of absolute values or know of any such tricks?
-
- The most obvious trick is:
-
- |x - y| < a <=> (x - y) < a _and_ (y - x) < a
-
- But this does not generalize to the greater than case:
-
- |x - y| > a <=> (x - y) < a _or_ (y - x) < a
-
- However,
-
- |x - y| > a <=> -|x - y| < -a
-
- So after multiplying by -1 the greater than case reduces to the
- less than case.
-
- I would like tricks for more complicated absolute value forms, e.g.,
-
- |x1 - x2| + |y1 - y2| < a
-
- The obvious trick (which is a simple generalization of the
- |x - y| < a trick) produces a set of equations that is
- exponential in the number of absolute value expressions, e.g.,
-
- |x1 - x2| + |y1 - y2| < a <=>
-
- x1 - x2 + y1 - y2 < a and
- x2 - x1 + y1 - y2 < a and
- x1 - x2 + y2 - y1 < a and
- x2 - x1 + y2 - y1 < a
-
-
- Since this is a novice question, please reply to me directly, so as not
- to clutter up the net.
-