Release Notes

We would like to thank all of our users who have reported problems and made suggestions for improving this release. In particular, we thank Wolfgang Britz, Michael Ferris, Josef Kallrath, Andreas Lundell, Renger van Nieuwkoop, Sabine Ritter, Stefan Vigerske, and Kent Zhao.

• The 32bit and 64bit Linux systems are now built using the GNU Compiler Collection (GCC) toolset.
• Dropped support of Windows 95, 98, ME, and 2000.

For a long time GAMS supports calling executables via the $call and Execute in a synchronous way. The new release also supports asynchronous job handling. This means you can start a job without waiting for the result. You can continue in your model and collect the return code of the job later. There are three ways to start a job asynchronously:

• $Call.ASync ... (compile time)
• Execute.ASync '...'; (execution time)
• put_utility fx 'Exec.ASync' / '...'; / put_utility fx 'Shell.ASync' / '...'; (execution time)

After each of those the function JobHandle can be used to get the Process ID (pid) of the last job started. With JobStatus(pid) one could check for the status of a job. Possible return values are:

• 0: error (input is not a valid PID or access is denied)
• 1: process is still running
• 2: process is finished with return code which could be accessed by errorlevel
• 3: process not running anymore or was never running, no return code available

With JobTerminate(pid) a interrupt signal can be sent to a running job. If this was successful the return value is one, otherwise it is zero.

With JobKill(pid) a kill signal can be sent to a running job. If this was successful the return value is one, otherwise it is zero.

The model asynexec from the GAMS Test Library demonstrates the use of this new feature.

The new GAMS Function Library Facility allows users to import functions from an external library into a GAMS model. Apart from the import syntax, the imported functions can be used in the same way as intrinsic functions. In particular, they can be used in equation definitions. Some function libraries are included with the standard GAMS software distribution but GAMS users can also create their own libraries using an open programming interface. Simple examples in the programming languages C, Delphi and Fortran come with every GAMS system.

Function libraries are made available to a model using the compiler directive:

$FuncLibIn <InternalLibName> <ExternalLibName>

Similar to sets, parameters, variables, and equations, functions must be declared before they can be used:

Function <InternalFuncName> /<InternalLibName>.<FuncName>/;

There are a few libraries which come ready to use with the GAMS system:

• fitfclib: packages FITPACK from P. Dierckx in a way that it works with this facility
• pwpcclib: evaluates piecewise polynomial functions
• stodclib: provides random deviates, probability density functions, cumulative density functions and inverse cumulative density functions for certain distributions
• tricclib, tridclib, trifclib: these simple examples can be found compiled and as source code written in C, Delphi and Fortran respectively

Detailed information can be found in the GAMS User's Guide in chapter 6.3.

In addition, the following examples from the GAMS Test Library can be referred to:

• stolib01: Uses the stochastic library which comes with the GAMS system
• trilib01: Uses trigonometric function from a library written in C, comes with C source code
• trilib02: Uses trigonometric function from a library written in Delphi, comes with Delphi source code
• trilib03: Uses trigonometric function from a library written in Fortran, comes with Fortran source code
• pwplib01: Uses a library for piecewise polynomial functions which comes with the GAMS system
• fitlib01: Uses FITPACK from P. Dierckx which was packaged in a way that it works with this facility and comes with the GAMS system

If you need detailed instructions for connecting your library to GAMS please contact support@gams.com.

• Dropped support for EBCDIC. Functions ordebcdic and ordascii are no longer available. oldascii should be replaced by ord.

We found and corrected a bug related to the internal organization of non-linear code. This bug was introduced with version 23.6 but fortunately only occurs in rare cases.

• Version 23.6 made it more difficult to find the directory where GAMS is installed. In this release we reintroduce the EXECUTABLE entry in the GAMSINI file to make this easier.
• The PageWidth option was limited to 255
• Increased the maximum number of jobs that can be run at he same time from 5 to 20.

• The gamsinst program allows to set the default solvers and to install contributed software packages (e.g. Tom Rutherford's GAMS Programming Tools). Previous versions of gamsinst supported the installation of contributed solver packages. The new version of gamsinst does not support these types of solver packages anymore. If you need to install a contributed solver, please contact support@gams.com.

• Fixed a memory leak that could make it difficult to read large spreadsheets.
• Worked around a limitation to read large spreadsheets by reading the data in multiple slices.
• The option NAin allows a string to be specified to recognize the value for NA when reading a spreadsheet; this is in addition to recognizing the string 'NA'.
• Undf is now recognized as a value without generating an error when reading from a spreadsheet so the entry can be processed in the GAMS model using the $ONUNDF dollar option in GAMS.
• Strings that are not recognized as a special value are no longer read as NA but as Undf instead; use the $ONUNDF in GAMS to read these values.

• Update to version 2.04.01
• New options
  • CUTdelcrit Improved heuristic to reselect cuts.
  • CUTnrcuts Option that the algorithm decides was added.
  • ECPdumpsol Write encountered solutions in GDX files.
  • MIPsolstrat Strategy for multiple MIP solutions
  • NLPcall Option that the algorithm decides was added.
  • solvelink Determines the way the NLP and MIP solvers are called. By default AlphaEXP now calls the NLP and MIP solver in memory.
• Deprecated options
  • CUTautogrowth, CUTcheckgrowth, CUTgrowth, CUTgrowtheq, CUTgrowthnr, CUTmincuts
  • ECPcheckviol
  • NLPepsilon, NLPlimepsilon

• New libraries 9.3
• Compared to previous versions, BARON 9.3 comes with improved multilinear relaxations and improved branching
• Support for the LP solvers XA and OSL has been dropped

• New libraries 3.14Y

• Updated Bonmin and Couenne libraries (bugfixes).
• On Linux and Windows, Ipopt now uses a multithreaded linear algebra library. The number of threads can be set via the GAMS option THREADS.
• Added Bonmin option "print_funceval_statistics" to enable statistics on number of function evaluations during a Bonmin run.
• Added Ipopt, Bonmin, and Couenne option "print_eval_error" to enable printing of information about function evaluation errors into the listing file.
• OsiCplex, OsiGurobi, and OsiXpress now support using the variable level values as initial solution for a MIP solve. This is enabled by setting the GAMS option INTEGER4 to 1.

• New libraries 12.3
• Cplex 12.3 solves a greater variety of nonconvex quadratic programming models; that is, QP models with a quadratic term in the objective function. These models are also known as indefinite QPs (in contrast to positive or negative semi-definite QPs). In fact, Cplex 12.3 can find solutions that satisfy Karush-Kuhn-Tucker (KKT) conditions for certain indefinite QPs. A new parameter, solution target type (SolutionTarget), lets you specify to Cplex whether you will accept a solution that satisfies first-order optimality conditions (in contrast to accepting only globally optimal solutions), and Cplex computes and searches accordingly.
• Log files improved. You may observe slight differences from previous versions in the format of logs produced by Cplex 12.3. For example, the name of the column formerly titled "best node" in logs of MIP optimizations is now titled "best bound" to reflect more accurately the data recorded there.
• Supports in-core communication (solvelink=5)

• The solver DEA was dropped from our distribution.
• Data Envelopment Analysis (DEA) models can now be solved in GAMS via the Gather-Update-Solve-Scatter (GUSS), for more information see GAMS/DEA modeling

• New option
  • solvelink Determines the way the NLP and MIP solvers are called. By default Dicopt now calls the NLP and MIP solver in memory.

• New libraries 4.5.1
• Now available on AIX
• New default Method for continuous models: The new version uses a new Automatic setting as the default for solving continuous models. In previous releases, continuous models were solved with the dual simplex method by default. While the exact strategy used by the new Automatic setting may change in future releases, in this release the new approach uses the concurrent optimizer for continuous models with a linear objective (LPs), the barrier optimizer for continuous models with a quadratic objective (QPs), and the dual simplex optimizer for the root node of a MIP model. You should change the Method parameter if you would like to choose a different method.
• New Minimum Relaxation heuristic: The new version contains a new Minimum Relaxation heuristic that can be useful for finding solutions to MIP models where other strategies fail to find feasible solutions in a reasonable amount of time. Use the new MinRelNodes parameter to control this new heuristic.
• New branch direction control: The new version allows more control over how the branch-and-cut tree is explored. Specifically, when a node in the MIP search is completed and two child nodes, corresponding to the down branch and the up branch are created, the new BranchDir parameter allows you to determine whether the MIP solver will explore the down branch first, the up branch first, or whether it will choose the next node based on a heuristic determination of which sub-tree appears more promising.
• Cut pass limit: The new version allows you to limit the of cut passes performed during root cut generation in MIP. Use the new CutPasses parameter.

• The purpose of this new Gather-Update-Solve-Scatter manager or short GUSS is to provide syntax at the GAMS modeling level that makes an instance of a problem and allows the modeler limited access to treat that instance as an object, and to update portions of it iteratively. Detailed documentation of this facility is part of the Solver Manual (chapter GUSS).
• Solving MCP and CNS models with GUSS is an experimental feature. GAMS checks the consistency of MCP and CNS model, these checks are currently disabled when models of these types are solved in the GUSS framework.

• Supports logic equations
  • Logic equations can contain
    • binary variables
    • expressions that evaluate to constants
    • Boolean operators (AND,OR,XOR,NOT,IMP(→),EQV(⇔))
  • Logic equations can be used in conjunction with disjunctions
  • A version of the food model demonstrates the use of this new feature foodemp (59)
• Supports coupling and indicators through EMP's disjunction syntax
  • Indicator disjunction b e says that constraint e has to hold if b is true
  • Coupling disjunction b e else not e says that constraint e can only hold if and only if b is true
  • Indicators and Couplings can be reformulated using BigM, Convex Hull or CPLEX indicator constraints

• New libraries Lindo 6.1.1.588
• New Option checkrange: Calculates the feasible range for every variable in each equation while all other variables are fixed to their level. If set, the value of this option defines the name of the GDX file where the results are written to.

• The former LogMIP solvers lmbigm and lmchull are combined in the new solver solver logmip.
• LogMIP now uses the EMP syntax and modeltype.
• The LogMIP examples in the GAMS Model Library were revised.

• Supports in-core communication (solvelink=5)

• New libraries 6.0 Rev 114

• New option
  • solvelink determines the way the NLP and MIP solvers are called

• The solver Oslse was dropped. Oslse is now an alias to osl.

• Supports in-core communication (solvelink=5)

• New option
  • solvelink determines the way the NLP solver is called. By default SBB now calls the NLP solver in memory.

• Dropped support for LPs/RMIPs in Gams/SCIP interface.
• Added new solver OsiSoplex for solving LPs with SoPlex (via COIN-OR/OSI interface).

• Supports in-core communication (solvelink=5)

• New libraries 22.01
• Improved performance through dynamic synchronization in the deterministic concurrent solver
• The stability and performance of the barrier code has been improved
• The default performance of the simplex algorithm has been improved on some models
• Improved zero-half cuts and aggregated cuts for MIPs.
• Tuned some of the automatic strategies for cutting and branching in MIP solves.
• Threads during a MIP solve will now busy-wait by default instead of going to sleep when waiting for work. This is to overcome a performance issue with modern speed-stepping CPUs, which might step down to a lower clock frequency when the load is less than 100%. This feature can be toggled using the SLEEPONTHREADWAIT control.

• datatest (80): Driver for datalib tests of all sorts

• foodemp (59): Food Manufacturing Problem - Blending of oils
• simenlp (60): Simple example of ENLP
• nlp2mcp (61): nlp2mcp: Form the KKT conditions of an NLP as an MCP
• simplevi3 (62): Another simple VI example
• simpequil2 (63): Simple Equilibrium with external constraint
• hark-monop (64): SPE model from Harker - monopolist and competetive versions
• hark-oligop(65): SPE model from Harker - oligopoly version
• hark-oligop-ind (66): SPE model from Harker - oligopoly version
• hark-stack (67): SPE model from Harker - Stackelberg version

• kqkpsdp (355): SDP Convexifications of the Cardinality Constraint Quadratic Knapsack Problem
• prodplan (356): A Production Planning Example
• sddp (357): Multi-stage Stochastic Water Reservoir Model solved with SDDP
• ps2_f_s (358): Parts Supply Problem w/ 2 Types w/o & w/ Asymmetric Information
• ps2_f (359): Parts Supply Problem w/ 2 Types w/o Asymmetric Information
• ps2_f_eff (360): Parts Supply Problem w/ Efficient Type w/o Asymmetric Information
• ps2_f_inf (361): Parts Supply Problem w/ Inefficient Type w/o Asymmetric Information
• ps2_s (362): Parts Supply Problem w/ 2 Types w/ Asymmetric Information
• ps3_f (363): Parts Supply Problem w/ 3 Types w/o Asymmetric Information
• ps3_s (364): Parts Supply Problem w/ 3 Types w/ Asymmetric Information
• ps3_s_gic (365): Parts Supply Problem w/ 3 Types w/ Global Incentive Comp. Const.
• ps3_s_mn (366): Parts Supply Problem w/ 3 Types w/ Monotonicity Constraint
• ps3_s_scp (367): Parts Supply Problem w/ 3 Types w/o & w/ SCP
• ps10_s (368): Parts Supply Problem w/ 10 Types & w/ Asymmetric Information
• ps10_s_mn (369): Parts Supply Problem w/ 10 Types w/ Random p(i)
• ccoil (370): Oil Pipeline Design Problem using concurrent MIP solves

• interval (506): Test for interval evaluation
• empdisj3 (507): Test Disjunctions using not (equivalence)
• emplog1 (508): Test disjunctions and logical conditions
• maxima (509): Test several maxima of Cmex
• emplog2 (510): Test simple logical conditions
• qcp07 (511): Test correctness for QCP with poly function
• qcp08 (512): Test correctness for QCP with power(*,2) function
• empdisj4 (513): Test disjunctions with negative bounds on variables
• funcback (514): Test function backward compatibility
• asynexec (515): Test asynchronous execution at compile and execution time
• scen01 (516): Compile time test for the scenario facility
• empdisj5 (517): Test disjunctions using not on binary variable and constraint
• scen02 (518): Execution time test for the scenario facility
• scen03 (519): Multi-dimensional scenario solver test
• stolib01 (520): Test extrinsic functions in stodclib
• trilib01 (521): Test extrinsic functions in tricclib
• trilib02 (522): Test extrinsic functions in tridclib
• trilib03 (523): Test extrinsic functions in trifclib
• lindorg2 (524): Test for LindoGlobals CheckRange Option
• mip04 (525): Exercise new XPRESS return code for unbounded MIP
• load5 (526): Tests UNDF with $load
• pwplib01 (527): Test piecewise polynomials in pwpcclib
• fitlib01 (528): Test the use of FITPACK inside GAMS
• load6 (529): Tests dynamic error messages during $loaddc
• fnspowx (530): Test correctness of signpower intrinsic
• fnspown (531): Test correctness of signpower intrinsic
• nlcode5 (532): Test for NL code bug from Dist 23.6
• scensol2 (533): Test handling of unsorted scenario UELs in GUSS

Solver/Platform availability - 23.7    July 14, 2011
	x86 MS Windows	x86_64 MS Windows	x86 Linux	x86_64 Linux	Sun Sparc SOLARIS	Sun Sparc64 SOLARIS	Sun Intel SOLARIS	IBM RS-6000 AIX 5.3	Mac Intel32 Darwin	Mac x86_64 Darwin
ALPHAECP
BARON 9.3
BDMLP
COIN-OR
CONOPT 3
CPLEX 12.3
DECIS						32bit
DICOPT
GUROBI 4.5
KNITRO 7.0
LINDOGLOBAL 6.1					6.0	6.0
LGO
MILES
MINOS
MOSEK 6
MPSGE
MSNLP						32bit
NLPEC
OQNLP		32bit		32bit
PATH
SBB
SCIP
SNOPT
XA		32bit
XPRESS 22.01

GAMS Distribution 23.6 December 13, 2010

We would like to thank all of our users who have reported problems and made suggestions for improving this release. In particular, we thank Wolfgang Britz, Nico Keyaerts, Leon Lasdon, Xiang Li, Timo Lohmann, Sebastian Ritter, Yannick Rombauts, Tom Rutherford, Uwe Schneider, Stefan Vigerske, Achim Wechsung, and Gyeongbeom Yi.

• Added the function RandBinomial(n,p) which generates random numbers from the Binomial distribution where n is the number of trials and p the probability of success for each trial.
• Added the gams date and release functions to $eval and [..] constant evaluations: jdate, jtime, gyear, gmonth, gday, gdow, gleap, ghour, gminute, gsecond, gmillisec, jstart, jnow, gamsversion, gamsrelease.
• Modified operator precedence binding of IMP (→), EQV (⇔), OR and XOR. These operator used to have the same precedence binding, so with the usual left to right evaluation an expression like a → b OR c used to be equivalent to (a → b) OR c. IMP and EQV have now a lower precedence binding, so the expression a → b OR c is now equivalent to a → (b OR c).
• There is no longer a practical limit for the number of non-linear instructions for a single equation.
• New interpretation of value default for command line parameters gdx and rf: default is interpreted as the input file name, this means for example, that gams trnsport gdx=default will write the file trnsport.gdx. Along the same lines, gams trnsport rf=default will write the reference file trnsport.ref.
• New command line parameter Threads: Controls the number of threads to be used by a solver. It can also be set as GAMS option or model attribute. The model attribute has the highest priority, the command line parameter the lowest. Non-positive values are interpreted as the number of cores to leave free so setting threads to 0 uses all available cores while setting threads to -1 leaves one core free for other tasks.

• Determine platform dependent library name automatically, no requirement for compiler flags anymore
• Delphi API files also support Delphi 2010
• Python API files comprehensively improved
• More Fortran examples

The new tool chk4upd checks if there is a newer version of GAMS available than the one you are running. It checks for the most recent version available as well as for the newest version you can run with your license in case your maintenance and support is expired. By default it reads gamslice.txt located next to it. Alternatively, you can specify another license on the command line. In addition to the command line tool you will find an integration in the GAMS IDE at Help → Check for GAMS Update.

• Integration of 'Check for GAMS Update' in the Help Menu
• The 'Find in Files' option now allows for one or more sub-directories to be excluded from the search
• The selection to use a different GAMS system has been changed. We no longer specify the gams.exe executable but define the alternate GAMS system directory instead.

The function to test if two values are different when Eps/RelEps has been specified was changed. The documentation describes the new implementation.

GDXMRW is a collection of utilities that make it easier and more reliable to share data between GAMS and Matlab and that allow the Matlab user to call GAMS using something similar to a subroutine interface. Formerly contributed software, these utilities were integrated into GAMS Distribution 23.4 and have been updated and improved since then.

This new program will write all worksheets of an Excel workbook to a gdx file. Unlike GDXXRW, the program does not require that Excel is installed. Windows platforms only.

• Option ExtNLPsolver allows to specify an option file for the GAMS NLP solver chosen (e.g. ExtNLPSolver conopt.1)

• New libraries
  • BONMIN 1.4
  • CSDP 6.1.1
  • CSDP now also available on Solaris x64 on Intel

• The old names COINCPLEX, COINGUROBI, COINMOSEK, and COINXPRESS are not available anymore. Please use the new names OSICPLEX, OSIGUROBI, OSIMOSEK, and OSIXPRESS introduced with distribution 23.4.

• Supports in-core communication (solvelink=5)
• Dropped support of Conopt 2 (old version of Conopt)

• New libraries 12.2.0.1

• The emp information file now supports block-wise definition
• A flip operator (-) was introduced, see model flipper (50)

• New libraries 4.0
• Quadratic programming: The Gurobi Optimizer now supports models with quadratic objective functions. The new version includes primal simplex, dual simplex, and parallel barrier optimizers for continuous QP models, and a parallel branch-and-cut solver for Mixed Integer Quadratic Programming (MIQP) models.
• Concurrent optimizer: The Gurobi Optimizer now allows you to run multiple algorithms simultaneously when solving a linear continuous model on a multi-core machine. The optimizer returns when the first algorithm solves the model. We include both a standard concurrent optimizer and a deterministic concurrent optimizer. The latter returns the exact same solution every time you run it, while the former can sometimes return different optimal solutions from one run to the next. The former can sometimes be significantly faster.
• MIP performance: The MIP solver is faster in release 4.0. These improvements do not require any parameter changes.
• LP performance: The simplex and barrier solvers are slightly faster in release 4.0. We have also improved the numerical stability of the primal simplex solver and the barrier crossover algorithm.
• Delayed MIP strategy change: The Gurobi Optimizer now gives you the option to change a few MIP parameters in the middle of the optimization in order to dynamically shift the search strategy. Specifically, two new parameters, ImproveStartGap and ImproveStartTime, allow you to specify when the algorithm should modify the values of a few parameters that control the intensity of the MIP heuristics. By setting one or both of these parameters to non-default values, you can cause the MIP solver to switch from its standard parameter settings, where it tries to strike a balance between finding better solutions and proving that the current solution is optimal, to a set of parameter values that focus entirely on finding better solutions.
• New approach to choosing the LP algorithm: The functions previously controlled by parameters LPMethod and RootMethod are now controlled by parameters Method and NodeMethod. Use Method to choose the algorithm for solving a continuous model, or for solving the root relaxation in a MIP model. Use NodeMethod to choose the algorithm for solving node relaxations in a MIP model.
• Deprecated options: LPMethod and RootMethod
• New options
  • Method: Controls the algorithm used to solve continuous linear and quadratic models. This parameter also selects the algorithm used to solve the root node of a MIP model.
  • NodeMethod: Chooses the algorithm used to solve node relaxations in a MIP model.
  • ModKCuts: Controls the generation of mod-k cuts.
  • ImproveStartGap: Allows you to specify the optimality gap at which the MIP solver resets a few MIP heuristics parameters in order to shift the attention of the MIP solver to finding the best possible feasible solution.
  • ImproveStartTime: Allows you to specify the elapsed time at which the MIP solver resets a few MIP heuristics parameters in order to shift the attention of the MIP solver to finding the best possible feasible solution.
  • PreMIQPMethod: Chooses the presolve transformation performed on MIQP models.
  • PSDTol: Sets a limit on the amount of diagonal perturbation that the optimizer is allowed to do on the Q matrix for a quadratic model. If a larger perturbation is required, the optimizer will terminate with a message that Q is not positive semi-definite.

• New libraries 7.0.0
• Supports the MultiStart feature in Knitro
• Supports in-core communication (solvelink=5)
• Supported on 64bit Mac and 64bit Solaris on Intel
• Dropped support on Solaris Sparc and Sparc64

• supports in-core communication (solvelink=5)
• support of external equations

• New libraries 6.0.96
• Supports in-core communication (solvelink=5)
• Dropped support on Solaris Sparc and Sparc64

• New libraries 2.0 (more details)
• The GAMS/SCIP interface now also supports semicontinuous and semiinteger variables
• GAMS/SCIP now uses SoPlex 1.5.0 as LP solver

• Dropped support on Solaris Sparc and Sparc64

• Supports in-core communication (solvelink=5)
• New libraries 21.01

• GDXXRWExample15 (76): Write spreadsheet using a filter
• GDXXRWExample16 (77): Write spreadsheet using text and hyperlinks
• tompivot (78): Little GAMS Program from Tom Rutherford that Illustrates Report Generation with Excel
• readdata (79): Read Data from .inc, .xls, .mdb and .csv file

• flipper (50): Test of flipping equations
• scarfemp-dem (51): Scarf's Activity Analysis Example
• scarfemp-altdem (52): Scarf's Activity Analysis Example - non-closed form demand function
• scarfemp-primal (53): Scarf's Activity Analysis Example
• scarfemp-dual (54): Scarf's Activity Analysis Example
• scarfexpend (55): Scarf's Activity Analysis Example
• pies (56): PIES Energy Equilibrium
• exc2x2emp (57): pure exchange model (ie no production)
• exc2x2emp-dem (58): pure exchange model (ie no production)

• relief (353): Relief Mission
• spatequ (354): Spatial Equilibrium

• emp09 (492): Test initial levels for equilibrium EMP models
• xerr1 (493): External Function Errors: RHS wrong
• xerr2 (494): External Function Errors: Aij wrong
• cerr1 (495): Cone Equation Errors
• merr1 (496): Matrix Errors
• mip02 (497): Check on MIP solution value
• mip03 (498): Test for zero gap when optcr and optca set to zero
• emp10 (499): Test of EMP equilibrium models and flip operator
• emp11 (500): Test EMP formulations of scarfmcp
• empdisj2 (501): Test disjunctions involving the objective
• gdxxrw6 (502): Test for dset reading problem
• qcp06 (503): Nonlinear model cannot be solved as QCP
• gdxdump1 (504): Use GdxDump NoData on Transportation Problem
• lindorng (505): Test for LindoGlobal's CheckRange Option

Solver/Platform availability - 23.6    December 13, 2010
	x86 MS Windows	x86_64 MS Windows	x86 Linux	x86_64 Linux	Sun Sparc SOLARIS	Sun Sparc64 SOLARIS	Sun Intel SOLARIS	IBM RS-6000 AIX 5.3	Mac Intel32 Darwin	Mac x86_64 Darwin
ALPHAECP
BARON 9.0
BDMLP
COIN-OR
CONOPT 3
CPLEX 12.2
DECIS						32bit
DICOPT
GUROBI 4.0
KNITRO 7.0
LINDOGLOBAL 6.1					6.0	6.0
LGO
MILES
MINOS
MOSEK 6
MPSGE
MSNLP						32bit
NLPEC
OQNLP		32bit		32bit
PATH
SBB
SCIP
SNOPT
XA		32bit
XPRESS 21.01

GAMS Distribution 23.5 July 4, 2010

We would like to thank all of our users who have reported problems and made suggestions for improving this release. In particular, we thank Farhad Farnam and Alexander Gocht.

GAMS System

APIs

• New libraries 9.0.6

• New libraries 3.14U

• New libraries 12.2
• MIP kappa describes a new feature extending the computation of kappa to mixed integer programming (MIP). (In linear programming, kappa is also known as the condition number of a solution.) See GAMS/Cplex parameter mipkappastats for details.
• Intel processors prior to Pentium 4 are no longer supported.
• The MIP node log interval (GAMS/Cplex parameter MIPInterval) now accepts either a positive or negative integer value, making it easier for you to adjust the amount of information logged for your problem.
• The default value of the GAMS/Cplex ClockType parameter has changed from 0=automatic (let Cplex choose) to 2 (wallclock time).
• The concurrent optimizer launches distinct optimizers on multiple threads. Cplex now offers concurrent, deterministic optimization of linear programming models (LPs) when more than one thread is given to Cplex (GAMS/Cplex parameter Threads). Running concurrent optimizers simultaneously on multiple threads may consume more memory. If the memory reduction parameter (MemoryEmphasis) is turned on, then you have instructed CPLEX to conserve memory. In that case, Cplex does not run concurrent, deterministic optimization unless parameter LPMethod is explicitly set to 6.
• New parameter AuxRootThreads: Partitions the number of threads for Cplex to use for auxiliary tasks while it solves the root node of a problem. On a system that offers N processors or N global threads, if you set this parameter to n, where N > n > 0 then Cplex uses at most n threads for auxiliary tasks and at most N-n threads to solve the root node.

• New libraries 3.0.1
• Now also available on Mac OS X Intel 64bit

• New libraries Lindo 6.1.1 for all platforms except for Solaris Sparc 32 bit and 64 bit
• Now also available on Mac OS X Intel 64bit

• gen_r1 (483) : Generate GDX file for test of GDXMRW
• gen_r2 (484) : Generate GDX file for test of GDXMRW
• gen_rbig (485) : Generate GDX file for test of GDXMRW
• gdxmrw03 (486) : test the Matlab utility rgdx
• gdxmrw04 (487) : test the Matlab utility wgdx
• testinst (488) : test the Matlab mex-file gams
• gversion (489) : test the Matlab mex-file gams
• gdxmrw05 (490) : test the Matlab mex-file gams
• onmulti7 (491) : Test merge with $onmulti and empty data statement

Solver/Platform availability - 23.5    July 4, 2010
	x86 MS Windows	x86_64 MS Windows	x86 Linux	x86_64 Linux	Sun Sparc SOLARIS	Sun Sparc64 SOLARIS	Sun Intel SOLARIS	IBM RS-6000 AIX 5.3	Mac Intel32 Darwin	Mac x86_64 Darwin	HP 9000 HP-UX 111	SGI IRIX2	DEC Alpha Digital Unix 4.03	Mac PowerPC Darwin4
ALPHAECP
BARON 9.0
BDMLP
COIN-OR
CONOPT 3
CPLEX 12.2											10.0	9.1	8.1
DECIS						32bit
DICOPT
GUROBI 3.0
KNITRO 6.0					5.2	32bit
LINDOGLOBAL 6.1					6.0	6.0
LGO
MILES
MINOS
MOSEK 6					5.0	5.0					3.2
MPSGE
MSNLP						32bit
NLPEC
OQNLP		32bit		32bit
PATH
SBB
SCIP
SNOPT
XA		32bit				32bit
XPRESS 20.00											16.10
1)GAMS distribution for HP 9000/HP-UX is 22.1.
2)GAMS distribution for SGI IRIX is 22.3.
3)GAMS distribution for DEC Alpha is 22.7.
4)GAMS distribution for Mac PowerPC is 23.3.

GAMS Distribution 23.4 May 24, 2010

We would like to thank all of our users who have reported problems and made suggestions for improving this release. In particular, we thank Phil Bishop, Stefan Boeters, Pavel Borisovski, Wolfgang Britz, Michael Ferris, David Grace, Sascha Herrmann, Alexander Mitsos, Sebastian Ritter, and Tom Rutherford. We also thank Marcel Roelofs for helpful and insightful comments and discussions on Hessian computations.

• Now built on AIX 5.3
• 64 bit
• New AIX system no longer supports the following products BARON, CONOPT2, DECIS, OSL3, OSLSE, SCENRED, SCENRED2, and XA

• Dropped support of the Macintosh PowerPC GAMS System. Version 23.3 and older are still available for download

The GAMS parameter gridscript allows to point to a customized grid submission script.

GAMS checks a model for non-linearities at compile time and requires the user to specify using nlp in the solve statement. At run time these non-linearities might disappear (cancelation, multiplication of non-linear terms with 0, ...), but the using nlp prevented the use of a pure LP solver. The new model attribute <model>.TryLinear = 1 checks at run time whether a nonlinear model has any non-linearities and then calls the default/chosen solver for the resulting model type.

Depending on the original model type the resulting model type is:

• QCP/DNLP/NLP -> LP
• MIQCP/MINLP -> MIP
• RMIQCP/RMINLP -> RMIP

An example was added to the Test Library: trylin01 (482)

A compile time equivalent of the sameas function was added. E.g.

$eval a sameas(xyz,xYz)
$eval b sameas ( 'xyz' , "xyz" )
$eval c sameas (12-3,12-3)
$eval d sameas (12-3,13-4)
$log %a% %b% %c% %d%

will yield:

Compile Time Constants

APIs

GDXDCLIB API

Data Utilities

GDX2ACCESS

GDXXRW

MDB2GMS

SQL2GMS

EMP

IDE

Libinclude

Solvers

BARON

Coin-OR

CPLEX

GUROBI

LINDOGLOBAL

MOSEK

XPRESS

In-core communication solver links

Model Libraries

GAMS Data Library

GAMS EMP Library

GAMS Model Library

GAMS Test Library

Solver/Platform Availability Matrix

Set years /bc2000*bc1,0*2009/;
Set years /"-20"*"-1"/;

          1         -1        0.000 ExecInit
        139         15        0.000 Assignment cnf
        139         13        0.000 Assignment cnf
         .           .         .    .
         .           .         .    .
         .           .         .    .
       3549        432        0.000 Equation   nbal
       3621      39489        0.032 Solve Fini wsisn
       3621         -1        0.062 GAMS Fini
          1         -1        0.000 ExecInit
       3621         -1        0.047 Solve Read wsisn
       3621         -1        0.000 GAMS Fini

--- Profile Summary (184 records processed)
     0.062      3621 GAMS Fini
     0.047      3621 Solve Read wsisn
     0.046      3529 Equation   divcnlsea (86)
     0.032      3621 Solve Fini wsisn (39489)
     0.016      3274 Assignment wnr (2502)
     0.016      3447 Equation   cost (15)
     0.016      3475 Equation   laborc (180)
     0.016      3519 Equation   waterbaln (180)
     0.016      3546 Equation   subirrc (84)
     0.015      3030 Assignment gwtsa (273)