David,
 
I believe that I have already made a comment on your question(exactly the same question)
which you posted at JADE newsgroup about two weeks ago.
 
I think Ernest can probably give you the detail of the rule-base side of 
"Scheduling & Inventory" applications where JESS is applied. I would like to add
more to my last comment in reply to your question from the JADE mailing list.
"Scheduling" and "Inventory" are formal branch of mathematics which is taught
in Statistics, Management Science,Engineering Science and Operations Research
courses and this means you have to be familiar in stochastic modelling, matrix
algebra and numerical analysis.
 
For the rule-base side , I will leave that to the expert such as Ernest and
Bob Orchard (FuzzyJ Author) as I have not developed inventory applications myself.
Some of the concepts applied in Inventory Control are fuzzy, such as "initial stock" ,
"current stock" , "ordered stock quantity" , "holding cost" ,"shortage cost" ,
"ordering cost" and so on, which you need to develop your application using FuzzyJ
with JESS.
 
My interest is in numerical computing and to my understanding, the applications should be
in 2 parts (Remember , I have not done expert systems inventory control development before,
but I am familiar with numerical computing in Operations Research field).
 
You need some:
 
 - numeric computation capability by using stochastic & linear programming algorithms.
 - JESS rules (or JESS + FuzzyJ) about the domain knowledge (Scheduling & Inventory).
 
The JESS (+ FuzzyJ) rules should use the output of the numeric computations as input for
the rules to work (asserted as facts).
 
There are some Java-based  numerical APIs (commercial & open source) that will
help you to get started for the numerical computing side of "Scheduling & Inventory"
software development.
 
1) JSolver (commercial Java API from ILOG)
 
 
  JSolver is a Java based API for  "Linear Programming", "Integer Programming" ,
  "Stochastic Modelling", "Optimization" and also "Quadratic Programming" . With JSolver
  you can  develop the numerical computing side of "Scheduling & Inventory" applications
  with ease and not having to worry about matrix algebra and numerical analysis
  because most of the complicated algorithms are available.
 
  This API is very expensive for a single developers license.
 
2) OR-124 (Operations Research ver-1.2.4)
 
  http://opsresearch.com/cgi-bin/mainIndex.cgi
 
  This is a popular and an excellent free Operations Research Java API. Major Linear Programming
  algorithms are implemented such as Simplex methods and so on. This API does not do "Quadratic Programming".
 
3) LPSolve (Free Linear Programming Java package)
 
 
  This package is bundled together with "JSci" (Java API for Science). This is also an excellent Java
  package for "Linear Programming" .
 
4) JMSL from Visual Numerics (commercial Java API)
 
 
   Excellent Java numerical libraries which have various algorithms in "Linear Programming",
   "Quadratic Programming" , "Optimizations" and general "Operations Research" already available.
   JMSL is one of the best Java numerical API available in the market.
 
   This API is very expensive for a single developers license (~ $ 5,000 U.S). The advantage is
   that you do not  have to worry about complex algorithm in matrix algebra and stuff like that.
 
5) NAG numerical API ( commercial API)
 
 
   The best commercial numerical APIs available for developing number crunching based applications such
   as large scale inventory systems. The only downside is that you have to use JNI(Java Native Interface)
   which destroys the Java's  "write once, run all" principle.
 
   This API is very expensive for a single developers license.
 
 
  The outlook for inventory applications in Java is good. The lead-spec of JSR-142 (J2EE Java Inventory API)
  will be implementing various stochastic modelling algorithms in future versions. I had confirmation of this
  when I inquired if this API does have stochastic modelling algorithms. He is liasing very closely with the
  lead-spec of JSR-83 (Java Multi-Array , a numerical computing API for Java) about implementation issues.
 
  A good book to start with relating to inventory is:
 
  "Introduction to Operations Research" (6th Edition) by F.S.Hillier and G.J.Lieberman,
  published by McGraw-Hill International. Chapter 17 is about "Inventory" , which details the mathematical theory
  and shows algorithm for solving inventory problems.
 
  I have a Statistics Java API that I have developed using JAMA (Java Matrix Algebra) API  which is freely
  available from NIST (National Institute of Standard & Technology):
  
    http://math.nist.gov/javanumerics/jama/
 
  If you are interested in my stats API for your inventory project, then you should contact me privately.
 
  Finally this issue has been brought up in this mailing list before and it is very important. Until you
  cannot come up with rules for inventory applications then you "SHOULD STOP RIGHT THERE , UNTIL YOU CAN".
  It seems to me , that you have no clue at all about the rules for inventory applications. You probably
  have not done the job of collecting knowledge from experts in the area of inventory. I recommend, you
  should buy JIA (Jess in Action)book and read about collecting knowledge from experts in early chapters of
  the book.
 
  Cheers,
  Sione.
----- Original Message -----
From: David
Sent: Thursday, October 16, 2003 4:11 PM
Subject: RE: JESS: Jess at JavaRanch

Hi Ernest,

 

In that website, you mentioned something about the possibility of u Jess in more complex applications like inventory control.

 

I wonder do you have any examples or implementations where Jess is deployed to better manage inventory.

 

Regards

David

 

 

 

-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of [EMAIL PROTECTED]
Sent:
Thursday, October 16, 2003 4:26 AM
To: [EMAIL PROTECTED]
Subject: JESS: Jess at JavaRanch

 

Hi Folks,

 

There's an original article (not a book excerpt) describing how to use

Jess to solve a tough word problem  in this month's JavaRanch

newsletter. It's the lead article. See

http://www.javaranch.com/newsletter/200310/NewsletterOctober2003.jsp .

 

---------------------------------------------------------

Ernest Friedman-Hill 

Distributed Systems Research        Phone: (925) 294-2154

Sandia National Labs                FAX:   (925) 294-2234

PO Box 969, MS 9012                 [EMAIL PROTECTED]

Livermore, CA 94550         http://herzberg.ca.sandia.gov

 

--------------------------------------------------------------------

To unsubscribe, send the words 'unsubscribe jess-users [EMAIL PROTECTED]'

in the BODY of a message to [EMAIL PROTECTED], NOT to the list

(use your own address!) List problems? Notify [EMAIL PROTECTED]

--------------------------------------------------------------------

 

Reply via email to