GAMS [ Home | Downloads | Documentation | Solvers | APIs | Tools | Model Libraries | Resources | Sales | Support | Contact Us | Search ]

Release Notes

Major releases contain substantial changes to the GAMS system. The License Check date is set to the release date of the major release.

Minor releases are mainly issued to provide bug fixes, performance improvements, and maintenance releases of solver libraries. Additionally, they can provide a few new features that do not change existing behavior. The License Check Date remains the same as for the prior major release. This means that any license file that worked with the prior major release will also work with this minor release.

Maintenance releases do not provide any new features. They are issued to provide bug fixes, performance improvements, and maintenance releases of solver libraries. The License Check Date remains the same as for the prior major release. This means that any license file that worked with the prior major release will also work with this maintenance release.

GAMS Distribution 21.0 - May 15, 2003

This is a Major release.

GAMS System

  • GAMS Model Library
    We added quite a few global optimization models to the GAMS model library that demonstrate the power of the new suite of Global Optimizers available in GAMS: BARON, LGO, and OQNLP. In addition, we added a model that calculates the US holiday schedule (HOLIDAY) and a model for the Five Leaper Tour problem (FIVELEAP). The model NASH gives an example of a Mathematical Program with Equilibrium Constraints (MPEC) which can be solved by the beta solver NLPEC.

  • IDE
    • GDX data viewer: Symbols can be sorted by various attributes, and the display uses two grids to display the data. The top grid shows the plane index, the bottom grid the current data. The indices can be arranged using drag and drop. The arrangement is preserved in the project file. The data viewer can write a single symbol or all symbols to an html file.
    • Removed Setup button from the print dialog screen. Printer setup is now available from the window that follows this dialog. This allows more printer options to be selected, such as duplex printing.
    • Added option to update the GAMS system using a ZIP file in the GAMS system directory. (File | Options | Execute | Update)
    • Added entries on the help menu for easy access to some documentation files.
    • Revised the on-line help and added a number of screen shots.
    • Added file:/// as a recognized hyperlink.

  • GDX Utilities
    • New documentation includes GDX facilities in GAMS.
    • GDXXRW (Excel interface)
      - Added 'usage' output
      - Removed SSET as a type, and added Values option for a set.
      - Added option to specify a single field for comparison (for variables and equations)
      - Added option to ignore comparison of associated text
      - Added RelEps option for relative comparisons
      - Added UelTable option


  • New Global Solvers

    NLP and MINLP problems frequently have multiple local optima. The three new solvers, BARON, LGO and OQNLP aim at finding the best possible local optima, the global optimum solution. The solvers differ in the methods they use, in whether they find globally optimal solution with proven optimality, in the size of models they can handle, and in the functional forms they accept.

    • BARON
      BARON is from The Optimization Firm, LLC and the University of Illinois at Urbana-Champaign. The Branch And Reduce Optimization Navigator derives its name from its combining interval analysis and duality in its reduce arsenal with enhanced branch and bound concepts as it winds its way through the hills and valleys of complex optimization problems in search of global solutions.

    • LGO
      LGO is from Pinter Consulting Services, Inc. This algorithm combines rigorous statistical methods with traditional mathematical programming methods to find solutions within well defined bounds. LGO stand for Lipschitz Global Optimization.

    • OQNLP
      OQNLP from Optimal Methods, Inc and OptTek Systems, Inc. is a solver for global optimization of non-linear problems with discrete and continuous variables. This multi start solver combines robust nonlinear optimization technologies with state-of-the-art meta heuristics like scatter search. In this official release of OQNLP, nonlinear subproblems with different starting points can be solved by any GAMS NLP solver as well as the build-in LSGRG solver.

    The CONOPT family of NLP solvers now have three members, CONOPT1, CONOPT2, and CONOPT3. The alias CONOPT which used to point to CONOPT2 will now call CONOPT3. In case you need to run an older version of CONOPT you must specify CONOPT1 or CONOPT2.

    CONOPT3 is an enhancement of CONOPT2. An additional solver component based on sequential quadratic programming (SQP) principles has been added and it uses the new 2nd order facilities in GAMS. In addition, an improved automatic scaling routines is now used as default.

    CONOPT3 is a true multi-method solver that combines its many solver components (SQP, SLP or sequential linear programming, Quasi-Newton, and Steepest Descend) with dynamic selection of the best component based on performance statistics.

    CONOPT2 is, apart from minor bug fixes, identical to CONOPT2 from the previous release. CONOPT1 has not been changed and it will not be updated any more. We encourage all CONOPT1 user to move on to CONOPT3 or CONOPT2.

  • CPLEX 8.0
    Find detailed release notes here

    The new GAMS/MINOS, which is based on MINOS 5.51 uses a new LU factorization with Threshold Rook Pivoting. It also has better memory estimation and improved reporting of infeasibilities.

    MOSEK from MOSEK ApS, Copenhagen Denmark is a large scale system for solving linear, mixed-integer linear, and convex nonlinear mathematical optimization problems. MOSEK is particularly well suited for solving sparse large-scale problems using an extremely efficient interior point algorithm. This official release of MOSEK using library version 2.5 also solves second order conic programs. Please find details at

  • OSL
    The OSL optimizer family is now represented by its newest member, OSL3. You can still run OSL2 and OSL1 (where available) by specifying them explicitly, but the default OSL is OSL3.
    Improvements to OSL3 over past OSL versions include:
    • Better performance for the simplex, barrier, and MIP codes
    • Better memory management
    • More robustness, especially in the barrier code

  • PATH
    The PATH presolve has been extended and strengthened.

    The new GAMS/SNOPT, which is based on SNOPT 6.2-1 uses a new LU factorization with Threshold Rook Pivoting. Memory estimation, the reporting of infeasibilities, and the handling of LP's has been improved.

    GAMS/XPRESS now links to XPRESS 2003, the latest optimizer from DASH. Optimizer improvements include:
    • Completely redesigned and rewritten branch and bound framework
    • New code for the Newton Barrier optimizer
    • Enhanced branching technology, cut strategies and heuristics
    • Better memory management


  • McCarl's GAMS User Guide
    The new McCarl's GAMS User Guide will be installed automatically with CD installation. If you download the GAMS system from our web site, you have to install it separately, see [not available anymore]

  • Solver Manuals
    There is a revised version for most of the solver manuals, all of them are in a uniform format. A composite manual, The Solver Manuals, is also available for printing.