![[Home]](images/bhome.gif){kind=small}
![[What's New]](images/bnews.gif){kind=small}
![[Products & Services]](images/bprdsrv.gif){kind=small}
![[Contents]](images/btoc.gif){kind=small}
![[Feedback]](images/bfeed.gif){kind=small}
![[Search]](images/bsrch.gif){kind=small}
![[Dividing Line Image]](div.gif){kind=small}
For sizable optimization problems -- in most cases, models with more than about 16,000 decision variables or constraints -- we recommend that you consider using the AMPL algebraic modeling language as a potential alternative to spreadsheet modeling. Since AMPL Plus, like Microsoft Excel, can be controlled by a program in Visual Basic or another fourth generation language, you can still develop and deploy your application on PCs using Microsoft Windows. And Standard AMPL also gives you the option of running compute-intensive applications on powerful UNIX workstations.
Although AMPL is not as widely used as a program like Microsoft Excel, it does enjoy widespread support from institutions and companies like Bell Laboratories, Frontline Systems, CPLEX Optimization, and Compass Modeling Solutions. This page will provide a brief introduction to AMPL and AMPL Plus, with references to Web pages at several other sites. You'll find that there is a wealth of information about optimization modeling in AMPL available on the World Wide Web.
AMPL (A Mathematical Programming Language) was designed by Robert Fourer, former Chairman of the Department of Industrial Engineering at Northwestern University. (Click here for the Northwestern University AMPL Website.) AMPL is a very high level algebraic modeling language which allows you to express an optimization model in a form very close to conventional algebraic notation -- complete with set operations, indexing, summation and vector notation. The flexibility of AMPL set operations and indexing allows you to define models over multidimensional data (for example, products, sales regions and time periods) and to use one model with several different sets of data, even if there are different numbers of parameters, decision variables, and constraints in each data set.
During Dr. Fourer's first sabbatical at Bell Laboratories, the team of Robert Fourer, David Gay, and Brian Kernighan (well known for his role in creating the C programming language) designed and developed a language processor for AMPL. In the last few years, the AMPL language has been enhanced in many ways and the language processor has been continually upgraded. Dr. Fourer has recently completed a second sabbatical at Bell Laboratories, continuing his work with Gay and Kernighan on the AMPL system. (Click here for the Bell Laboratories AMPL Website.)
The AMPL language processor (known here as Standard AMPL) runs under UNIX on a variety of workstations, and on PCs under DOS with a built-in 32-bit DOS Extender. It uses a command line interface and accepts data in the form of ASCII text files in AMPL syntax. Although it has no graphical under interface, built-in editor or database access capabilities, it can be (and has been) used to solve very large optimization problems where all available memory and processor time must be devoted to the problem itself.
If you are a spreadsheet user and would like to learn more about algebraic modeling languages and see an example of an AMPL model, click here (11K page).
Frontline Systems developed AMPL Plus, a Microsoft Windows "front-end" for Standard AMPL. AMPL Plus was designed to give AMPL the ease of use, rapid prototyping ability and flexible database access facilities that are found today in fourth generation development tools and languages under Microsoft Windows. A good analogy is that AMPL Plus serves the same purpose for AMPL that Microsoft Visual C++ serves for the C++ language.
Problems solved with tools like AMPL can involve huge amounts of data and hundreds of thousands, or even millions of decision variables and constraints. One of the most powerful solvers available for such problems is the CPLEX family of optimizers from CPLEX Optimization, Inc. Frontline Systems and CPLEX Optimization have cooperated to ensure that users can easily move between both our AMPL solvers, and our DLL Solvers which are used with custom written programs.
Companies tackling the challenging optimization problems for which AMPL and AMPL Plus are intended often want much more support than can be provided with a standard "shrink-wrapped" software package. To meet the needs of these users, Compass Modeling Solutions, Inc. was formed., with Frontline Systems and CPLEX Optimization as strategic partners. Compass acts as a knowledgeable "single source" for Standard AMPL, AMPL Plus, and a variety of AMPL-compatible solvers from Frontline, CPLEX, and other vendors. Compass markets the full line of Frontline Systems solver products, as well as CPLEX for AMPL/AMPL Plus and the CPLEX Callable Library. Compass also offers comprehensive maintenance and technical support services for AMPL, and custom consulting, training seminars and other professional services for users of optimization modeling. (Click here for the Compass Modeling Solutions, Inc. Website.)
AMPL Plus offers a number of features to facilitate development and testing of optimization models:
Click here to see an example AMPL Plus screen (image about 31K). For more information on AMPL Plus, contact Compass Modeling Solutions at the Website above, by sending email to info@modeling.com, or by telephone (702-322-7600) or fax (702-322-3030).
