As detailed in chapter GAMS Output, the execution of a GAMS program passes through several stages, where the compilation is the first stage. Often when a program is run for the first time, it cannot be solved successfully because of compilation errors. This can be very frustrating, especially for new users. In this tutorial we will explain the causes of the most common compilation errors and offer some advice on how to fix them. For an introduction to compilation errors, see section Compilation Errors. In addition, the tutorial A GAMS Tutorial by Richard E. Rosenthal offers some detailed material on this topic.

Note

Frequently, many compilation errors in the latter part of the code are actually consequential errors that will disappear as soon as the compilation errors in the beginning have been resolved. Therefore we recommend to start with fixing the first few errors and to run the program again. More often than not, many subsequent compilation errors will have vanished.

Preliminary Remarks

Before we will turn to an overview of the most common errors, examples that illustrate them and advice how to resolve them, we will provide some basic information on compilation errors and error messages.

Finding Compilation Errors

Compilation errors are marked with four asterisks (****) in the compilation output of the GAMS listing file (also called output file or lst file, since it has the extension .lst), hence it is easy to detect lines where compilation errors occured by inspection.

Note

Compilation error messages also appear in the LOG file. And, if GAMS Studio is used, a click at an error message in the LOG will navigate directly to the problematic code in the source file.

Consider the following simple example:

Set c "crops"   / barley, wheat, soy, wheat, rice /;

If this set statement appears in a program, the resulting compilation output will contain the following lines:

   1  Set c "crops"   / barley, wheat, soy, wheat, rice /;
****                                           $172

...
Error Messages

172  Element is redefined

Note that in the echo print of the input file a line starting with **** is inserted and the dollar sign $ followed by a number appears on this line. This indicates that a compilation error - in this case error 172 - was discovered in the line above. In addition, a list of all errors with explanatory error messages is given at the end of the echo print.

In our example, the error refers to wheat and the error message addresses the cause of the error: the respective element is redefined. Thus we check the other elements of the set and quickly realize that wheat appears a second time. If we delete the second instance of wheat and run the code again, this line will not cause any errors anymore. Note that many compilation errors are as easy to fix as demonstrated here.

Repositioning Compilation Error Messages

By default, the error messages are shown directly beneath the line where the respective error is marked. GAMS allows to customize this position with the command line parameter ErrMsg: the value of 0 for ErrMsg will list all error messages at the end of the echo print and the value of 2 will suppress the error messages in the lst file completely. Note that the default is 1.

Consider the following simple example.

Set       a    / a1 * a7 /;
Set       b(a) / a7 * a9 /;
Parameter p(k) / i3 47 /;

The resulting compilation output will contain the following lines:

   1  Set       a    / a1 * a7 /;
   2  Set       b(a) / a7 * a9 /;
****                         $170
**** 170  Domain violation for element
   3  Parameter p(k) / i3 47 /;
****              $120
**** 120  Unknown identifier entered as set

Observe that the error messages are placed on the line after the error is marked. This can be especially useful if the program contains many lines of code. Next, we run the same code with the following call:

  > gams test  ErrMsg=0

The resulting compilaton output follows.

   1  Set       a    / a1 * a7 /;
   2  Set       b(a) / a7 * a9 /;
****                         $170
   3  Parameter p(k) / i3 47 /;
****              $120
...
Error Messages

120  Unknown identifier entered as set
170  Domain violation for element

Now the error messages are displayed at the end of the compiler listing.

Note

Default values for command line parameters and environment variables can be specified using a configuration file gamsconfig.yaml. The GAMS Configuration Editor can be used to view and edit the GAMS configuration file but any other text editor can be used as well. Alternatively, users may change the system level defaults by entering the line ErrMsg=0 in the file gmsprmnt.txt on Windows, or gmsprmun.txt on Unix machines as in the following example:

*****************************************************
* GAMS 2.50 Default Parameterfile for Windows NT                 *
* Gams Development Corp.                                         *
* Date : 20 Mar, 1998                                            *
*****************************************************
* entries required by CMEX, put in by gams.exe:
* SYSDIR
* SCRDIR
* SCRIPTNEXT
* INPUT
errmsg=0

Resolving Common Compilation Errors

There are hundreds of compilation errors in GAMS, but some of them are particularly frequent. In the table below we present these common errors, with a brief description of the possible cause and a link to a subsection below where examples and more details are given. We recommend that users also read the error messages, since they often contain additional hints.

Error A: Misplaced Semicolons

Maybe the most common error for new users is related to semicolons. Consider the following example adapted from the well-known transportation model [TRNSPORT]:

Sets  i   "canning plants"   / Seattle, San-Diego / ;
      j   "markets"          / New-York, Chicago, Topeka / ;

The resulting compilation output will contain the following lines:

   20    Sets i   canning plants   / Seattle, San-Diego / ;
   21         j   markets          / New-York, Chicago, Topeka / ;
****         $140      $36
...
Error Messages

 36  '=' or '..' or ':=' or '$=' operator expected
     rest of statement ignored
140  Unknown symbol

What went wrong? GAMS statements like the set statement in our example have to terminate with a semicolon, unless the next line of code begins with a reserved word. Now, set statements may extend over several lines and define several sets. In our example, the set statement extends over two lines and two sets are declared. However, there is a semicolon at the end of the first line, therefore GAMS assumes that the set statement ends there. The symbol j at the beginning of the next line has not been declared yet and it is not a GAMS keyword, thus it is marked as Unknown symbol. Note that error 36 is a consequential error that will disappear as soon as the first error has been resolved.

How do we fix this? There are two ways: either we could drop the semicolon at the end of the first line and thus indicate that the set statement continues to the second line or we could insert the keyword Set at the start of the second line and thus introduce a new set statement. These two alternatives are illustrated below:

Sets  i   "canning plants"   / Seattle, San-Diego / 
      j   "markets"          / New-York, Chicago, Topeka / ;

or

Sets  i   "canning plants"   / Seattle, San-Diego / ;
Set   j   "markets"          / New-York, Chicago, Topeka / ;

Note

• In general, GAMS statements have to be terminated with a semicolon. The semicolon may be omitted if the next line starts with a GAMS keyword.
• Even if it is not required, it is good practice to always end a statement with a semicolon.

Error B: Missing Semicolons

Consider the following example adapted from the well-known transportation model [TRNSPORT]:

Equations
       cost        "define objective function"
       supply(i)   "observe supply limit at plant i"
       demand(j)   "satisfy demand at market j"

cost ..        z  =e=  sum((i,j), c(i,j)*x(i,j)) ;

The resulting compilation output will contain the following lines:

   52    Equations
   53         cost        define objective function
   54         supply(i)   observe supply limit at plant i
   55         demand(j)   satisfy demand at market j
   56
   57    cost ..        z  =e=  sum((i,j), c(i,j)*x(i,j)) ;
****         $97                      $195,96,195,96 $195,96
...
Error Messages

 96  Blank needed between identifier and text
       (-or- illegal character in identifier)
       (-or- check for missing ';' on previous line)
 97  Explanatory text can not start with '$', '=', or '..'
       (-or- check for missing ';' on previous line)

What went wrong? GAMS statements like the equation declaration in our example have to terminate with a semicolon, unless the next line of code begins with a GAMS keyword. In our example, we omitted the semicolon at the end of the equation declaration statement and started a new statement after a blank line.

How do we fix this? We simply add a semicolon after the last explanatory text at the end of the equation declaration statement.

Note

A missing semicolon is often associated with error 96 or 97.

Error C: Spelling Mistakes

This error occurs if sets, set elements, parameters, etc. are referenced with a different name than the name they were declared with. Note that differences in capitalization are not considered spelling mistakes since GAMS is case insensitive.

Consider the following example adapted from the well-known transportation model [TRNSPORT]:

Sets i   "canning plants"   / Seattle, San-Diego / 
     j   "markets"          / New-York, Chicago, Topeka / ;

Parameters a(i)  "capacity of plant i in cases"
          /   Seatle      350
              san-diego   600  / ;
...
Equations
       cost        "define objective function"
       supply(i)   "observe supply limit at plant i"
       demand(j)   "satisfy demand at market j" ;

cst ..        z  =e=  sum((i,j), c(i,j)*x(i,j)) ;

The resulting compilation output will contain the following lines:

   20    Sets i   canning plants   / Seattle, San-Diego /
   21         j   markets          / New-York, Chicago, Topeka / ;
   22
   23    Parameters a(i)  capacity of plant i in cases
   24            /   Seatle      350
****                     $170
   25                san-diego   600  / ;
...
   49    Equations
   50         cost        define objective function
   51         supply(i)   observe supply limit at plant i
   52         demand(j)   satisfy demand at market j ;
   53
   54    cst ..        z  =e=  sum((i,j), c(i,j)*x(i,j)) ;
****      $140
...
Error Messages

140  Unknown symbol
...
170  Domain violation for element

What went wrong? Note that in line 24 the set element Seattle was referenced as Seatle. GAMS does not recognize this symbol as an element of the set i and thus marks this with error 170: Domain violation. GAMS catches this error since it automatically performs domain checking. The typo in line 54, where the equation cost is referenced as cst, is marked with error 140 (Unknown symbol), since cst was not declared before.

Note

Set name misspellings are associated with error 120, set element misspellings with 170 and other misspellings with 140.

Fixing errors like these is as easy as fixing typos.

Error D: Missing Set Elements

Sometimes users forget to include an element in a set statement, but reference it later. Consider the following simple example:

Set       c    "crops"                 / barley, wheat, soy /;
Parameter d(c) "demand in metric tons" / rice 3, barley 1, wheat 4, soy 2 /;

The resulting compilation output will contain the following lines:

   1  Set            c    "crops"                 / barley, wheat, soy /;
   2  Parameter      d(c) "demand in metric tons" / rice 3, barley 1, wheat 4, soy 2 /;
****                                                 $170
...
Error Messages

170  Domain violation for element

What went wrong? Note that the symbol rice was not defined as an element of the set c, thus GAMS does not recognize it and marks it with error 170: Domain violation. GAMS catches this error since it automatically performs domain checking.

Note

If symbols are not defined as set elements, but are referenced as if they belong to the set later, error 170 will occur.

This is easy to fix: we just add the missing element(s) to the elements of the respective set.

Error E: Problems with Indices

Recall that variables, sets, parameters and equation may be defined over one or more indices. If the identifiers are referenced later in the program, the indices must appear exactly in the order that was specified in the respective definition statement. Consider the following example adapted from the well-known transportation model [TRNSPORT]:

Sets i   "canning plants"   / seattle, san-diego / 
     j   "markets"          / new-york, chicago, topeka / ;

Table d(i,j)  "distance in thousands of miles"
                    new-york       chicago      topeka
      seattle          2.5           1.7          1.8
      san-diego        2.5           1.8          1.4  ;

Scalar    f       "freight in dollars per case per thousand miles"  /90/ ;
Parameter c(i,j)  "transport cost in thousands of dollars per case" ;
c(i,j,j) = f * d(j,i) / 1000 ;

The resulting compilation output will contain the following lines:

   40    Parameter c(i,j)  transport cost in thousands of dollars per case ;
   41    c(i,j,j) = f * d(j,i) / 1000 ;
****           $148       $171,171
...
Error Messages

148  Dimension different - The symbol is referenced with more/less
        indices as declared
171  Domain violation for set

What went wrong? Note that the parameter c was declared over two indices, but in the assignment statement in line 41 it is referenced with three indices. Such a mistake is marked with error 148. Note further, that the indices of d are referenced in the wrong order, which is marked with error 171.

Note

If an identifier is referenced with more or less indices as declared, then the error 148 will be triggered. If the indices are in the wrong order, the error 171 will be triggered.

This is easy to fix: we just check the declaration or definition statement(s) and adjust the reference.

Observe that the domain error 171 is also triggered if an identifier is referenced with index say i, but was defined over say j. However, there will be no domain error if i is a subset of j or i and j reference the same set since they are aliases.

Error F: Summing over Sets that are Already Indexed

Consider the following equation definition that is adapted from the well-known transportation model [TRNSPORT]:

supply(i) ..   sum((i,j), x(i,j))  =l=  a(i) ;

The resulting compilation output will contain the following lines:

   59    supply(i) ..   sum((i,j), x(i,j))  =l=  a(i) ;
****                         $125
...
Error Messages

125  Set is under control already

What went wrong? Note that the equation is indexed over the set i, therefore the indexed operation sum in the body of the equation may not be controlled by the index i again.

Note

Summing over sets that are already indexed will trigger error 125.

How do we fix this? We need to carefully check the indexed operation and drop the surplus index. In other cases the controlling index may have to be dropped from the equation name. Note that an error like this is often indicative of a lack of clarity in thinking.

Error G: Uncontrolled Sets

Consider the following equation definition that is adapted from the well-known transportation model [TRNSPORT]:

demand ..   sum(i, x(i,j))  =g=  b(j) ;

The resulting compilation output will contain the following lines:

   61    demand ..   sum(i, x(i,j))  =g=  b(j) ;
****                            $149        $149
...
Error Messages

149  Uncontrolled set entered as constant

What went wrong? Note that the variable x is indexed over the sets i and j, but, while i is the index of the indexed operation sum, j is "free": it does neither appear as index of the equation nor as controlling index in sum. Therefore it was entered as if it were a constant. Note further, that j is also used as an index on the right-hand side of the equation without being controlled.

Note

Error 149 marks instances when an index is not controlled in the context of an equation or an indexed operation like sum or prod.

How do we fix this? We have to think what we actually want to model. In this case we need to enter j as a controlling index for the equation. But it would also be possible to add it as a controlling index for sum (but not both!). In the latter case there would still be the error on the right-hand side which needed to be taken care of with another indexed operation, e.g., sum again. Both approaches would remove the compilation error, but one has to think about what actually should be modeled to decide which is the right fix here.

Error H: Mismatched Parentheses

Consider the following two equations adapted from the model [CHENERY]:

mb(i).. x(i) =g= y(i) + sum(j, aio(i,j)*x(j)) + ( e(i) - m(i) ))$t(i) ;
...
dvv(i)$(sig(i) <> 0)..  vv(i) =e= (pi*(1-del(i))/del(i))**(-rho(i)/(1+rho(i)) ;

The resulting compilation output will contain the following lines:

   137   mb(i).. x(i) =g= y(i) + sum(j, aio(i,j)*x(j)) + ( e(i) - m(i) ))$t(i) ;
****                                                                  $408,409
...
   155   dvv(i)$(sig(i) <> 0)..  vv(i) =e= (pi*(1-del(i))/del(i))**(-rho(i)/(1+rho(i)) ;
****                                                                                 $8
...
Error Messages

  8  ')' expected
...
408  Too many ),] or }
409  Unrecognizable item - skip to find a new statement
       looking for a ';' or a key word to get started again

What went wrong? Note that in the first equation we have one surplus closing parenthesis, this is marked with error 408. The error marked with 409 is a consequential error, it will disappear once the first error has been fixed. In the second equation, one closing parenthesis is missing resulting in error 8.

Attention

Opening and closing parentheses ( ), square brackets [ ] and braces (curly brackets) { } must match.

Note

Surplus closing parentheses, bracketes and braces are marked with error 408 and missing closing parenthesis, brackets and braces are marked with error 8. Missing opening parentheses, brackets and braces may result in GAMS marking surplus closing parentheses, brackets and braces; surplus opening parentheses, brackets and braces may result in GAMS marking missing closing parentheses, brackets and braces.

While fixing errors like these entails carefully counting opening and closing parentheses, there are strategies that help to prevent mismatching parentheses errors. Many text editors (like GAMS Studio) offer a feature that identifies matching parentheses and will issue a warning if there is a mismatch. We recommend to use this feature. Further, we recommend to also use the alternatives to parentheses: square brackets and braces. They are especially useful if there are several opening parentheses since it is easier to determine by inspection if each has a matching closing symbol.

Error I: Mistakes Relating to Equations

Recall that each declared equation must be defined if it is to be used in a model statement. If the equation definition is missing, GAMS will mark an error beneath the solve statement that refers to a model which references the respective undefined equation. Consider the following example adapted from the well-known transportation model [TRNSPORT]:

Equations
       cost        "define objective function"
       supply(i)   "observe supply limit at plant i"
       demand(j)   "satisfy demand at market j" ;

cost ..        z  =e=  sum((i,j), c(i,j)*x(i,j)) ;
supply(i) ..   sum(j, x(i,j))  =l=  a(i) ;

Model transport /all/ ;
solve transport using lp minimizing z ;

The resulting compilation output will contain the following lines:

   60    Model transport /all/ ;
   61    solve transport using lp minimizing z ;
****                                          $71,256
**** The following LP errors were detected in model transport:
****  71 demand is an undefined equation
...
Error Messages

 71  The symbol shown has been declared as an equation, but no
        Symbolic equation (..) was found. hint - look for commas in the
        Documentation text for the equations. use quotes around the
        Text or eliminate the commas.
256  Error(s) in analyzing solve statement. More detail appears
     Below the solve statement above

What went wrong? Note that the equation demand was declared and included in the model transport, but it was not defined. GAMS detects that the equation definition is missing when it compiles the solve statement that relates to the model transport. Therefore the error 71 appears only there. The error 256 is a consequential error, it will disappear once the missing equation definition has been added (before the solve statement).

Note

Error 71 marks a missing equation definition. The message associated with this error is a nice example of an error message that is very descriptive and offers advice on how to fix the error.

Apart from forgetting to define a declared equation and forgetting to terminate the equation definition statement with a semicolon (see above), two other mistakes relating to equations are frequent: omitting the syntax element .. after the name of the equation and not properly defining the equation type in the equation definition statement. The following simple example adapted from the portfolio analysis model [ALAN] illustrates these mistakes:

fsum       sum(i, x(i))               =e=  1.0;
dmean..    sum(i, mean(i)*x(i))        =   target;

The resulting compilation output will contain the following lines:

   45  fsum     sum(i, x(i))                 =e= 1.0;
****             $36
   46  dmean..    sum(i, mean(i)*x(i))        =  target;
****                                       $37
...
Error Messages

36  '=' or '..' or ':=' or '$=' operator expected
     rest of statement ignored
37  '=l=' or '=e=' or '=g=' operator expected

What went wrong? Note that in the first equation the two dots .. are missing, which is marked with error 36, and in the second equation the equation type is not properly specified, which leads to error 37.

Error J: Issues with Nonlinear Expressions

The compilation error messages 51 to 60 refer to issues with nonlinear expressions involving variables in equations. For example, nonlinear expressions in an LP model will cause errors of this sort. So will discontinuous functions with endogenous arguments in an NLP model. For information on endogenous arguments in functions, see section Functions in Equation Definitions. For an overview of GAMS model types, see section Classification of Models.

Consider the following simple artificial example:

Variables x, y, z ;
Equations eq1, eq2 ;

eq1.. x**2 - y =e= z ;
eq2.. min(x,y) =l= 20 ;

Model silly / all / ;
solve silly using lp maximizing z ;

The resulting compilation output will contain the following lines:

   7  Model silly / all / ;
   8  solve silly using lp maximizing z ;
****                                    $54,51,256
**** The following LP errors were detected in model silly:
****  54 equation eq1.. VAR operands for **
****  51 equation eq2.. the function MIN is called with non-constant arguments

Error Messages

51  Endogenous function argument(s) not allowed in linear models
54  Endogenous operands for ** not allowed in linear models
256  Error(s) in analyzing solve statement. More detail appears
     Below the solve statement above

Note that error 54 marks the nonlinear expression \(x^2\) in a linear model and error 51 refers to the discontinuous function min that may be used with variables only in models of the type DNLP. The error 256 is a consequence of the two previous errors.

Error K: Using Undefined Data

Referencing data items that were not declared with a scalar, parameter or table statement will cause error 140: GAMS will indicate that the respective symbol is unknown. More often data items are declared, but not defined, i.e. they are not initialized with values.

Consider the following example adapted from the well-known transportation model [TRNSPORT]. Assume that the scalar f is declared, but has no numerical value. The resulting compilation output will contain the following lines:

   40    Scalar f  freight in dollars per case per thousand miles  ;
   41
   42    Parameter c(i,j)  transport cost in thousands of dollars per case ;
   43
   44              c(i,j) = f * d(i,j) / 1000 ;
****                       $141 
...
   61    Model transport /all/ ;
   62    solve transport using lp minimizing z ;
****                                          $257
   63
   64    display x.l, x.m ;
****              $141
...
Error Messages

141  Symbol declared but no values have been assigned. Check for missing
        data definition, assignment, data loading or implicit assignment
        via a solve statement.
        A wild shot: You may have spurious commas in the explanatory
        text of a declaration. Check symbol reference list.
257  Solve statement not checked because of previous errors

What went wrong? Note that the scalar f was declared, but there was no value assigned to it. Thus, when it is referenced in the calculation in line 44, the "empty" scalar causes error 141. Error 257 is a consequential error and leads to the second error 141. This error occurs in the context of variable attributes in the display statement. Recall that variable attributes are data associated with variables. As the solve statement could not be checked, the level and marginal values of the variable x are not defined and hence error 141 follows.

Note

Error 141 marks places where data items, that have not been defined, are referenced in computations in the context of assignment statements. It also marks places where undefined variable (or equation) attributes are referenced.

In addition, undefined data items may be referenced in the context of equation definitions. Suppose we declare the parameter c in model [TRNSPORT], but forget the corresponding assignment statement. Then the compilation output will contain the following lines:

   57    cost ..        z  =e=  sum((i,j), c(i,j)* x(i,j)) ;
 ...
   62    solve transport using lp minimizing z ;
****                                          $66,256
**** The following LP errors were detected in model transport:
****  66 equation cost.. symbol "c" has no values assigned
...
Error Messages

 66  The symbol shown has not been defined or assigned
        A wild shot: You may have spurious commas in the explanatory
        text of a declaration. Check symbol reference list.
256  Error(s) in analyzing solve statement. More detail appears
     Below the solve statement above

Note that the missing data in the body of the equation cost is marked with error 66 and the following error is a consequential error which will disappear once 66 has been fixed.

Note

Referencing undefined data items in the body of an equation causes error 66.

Error L: Referencing Set Elements Without Quotes

Recall that if set elements are referenced, they need to appear in single or double quotes. Omitting the quotes may cause different errors depending on the context. The following example is adapted from the well-known transportation model [TRNSPORT]. Suppose we start by defining the two sets and the two parameters and then add the following statements:

Scalar s;
s = a(seattle);

The resulting compilation output will contain the following lines:

   37  Scalar s;
   38  s = a(seattle);
****              $120,340,149,171
...
Error Messages

120  Unknown identifier entered as set
149  Uncontrolled set entered as constant
171  Domain violation for set
340  A label/element with the same name exist. You may have forgotten
        to quote a label/element reference. For example,
        set i / a,b,c /; parameter x(i); x('a') = 10;

What went wrong? On the right-hand side of the assignment statement we have a reference with missing quotes. Note that this creates four (!) errors:

• 120 - GAMS expects a controlling set and Seattle is not recognized and therefore considered an unknown identifier.
• 340 - Then GAMS realizes that actually there is a label called Seattle in the program and helpfully offers a hint: You may have forgotten to quote a label/element reference.
• 149 - See section Error G: Uncontrolled Sets above.
• 171 - See section Error E: Problems with Indices above.

This is a nice example for one mistake triggering several errors that may look daunting. However, fixing them is as easy as adding single or double quotes.

Error M: Missing Declaration Statements

Recall that sets, parameters, variables and equations have to be declared before they may be used in definition or assignment statements. Suppose the equation demand in the well-known transportation model [TRNSPORT] was defined but not declared. Then the compilation output would contain the following lines:

   59    demand(j) ..   sum(i, x(i,j))  =g=  b(j) ;
****         $140
...
Error Messages

140  Unknown symbol

Note that as demand was not declared, GAMS does not recognize it and therefore marks it with error 140: Unknown symbol.

Note

If error 140 is reported, the respective declaration statement is probably missing. However, the underlying cause may also be a simple typo.

Error N: Using the Same Name for Different Identifiers

Recall that each GAMS identifier must have a unique name. If the same name is used more than once, an error will be triggered. Consider the following modification of the transportation model [TRNSPORT]:

Parameters  capacity(i)  "capacity of plant i in cases"
                / Seattle 350, San-Diego 600 /

            demand(j)    "demand at market j in cases"
                / New-York 325, Chicago 300, Topeka 275 /;
...
Equations
       cost        "define objective function"
       supply(i)   "observe supply limit at plant i"
       demand(j)   "satisfy demand at market j" ;

The resulting compilation output will contain the following lines:

   47    Equations
   48         cost        define objective function
   49         supply(i)   observe supply limit at plant i
   50         demand(j)   satisfy demand at market j ;
****              $195
...
Error Messages

195  Symbol redefined with a different type

What went wrong? Note that demand was first declared as a parameter and later as an equation. GAMS will mark this repeated use of the same name with error 195.

Error O: Using ORD with an Unordered Set

Recall that the operation ord is only defined for ordered sets. If ord is used with an unordered set, error 198 will be triggered.

Consider the following example:

Set    a   "a couple of the elements" / r2, r3 /;
Set    b   "more elements"            / r1*r4 /;
Scalar c   "counter"                  / 0 /;

loop( b $ (ord(b) > 3), 
    c = c + 1 );
display c;

The resulting compilation output will contain the following lines:

   5  loop( b $ (ord(b) > 3),
****                $198
   6      c = c + 1 );
...
Error Messages

198  Set used in 'ord' or lag is not ordered.
        Hint: Some of the elements of the set were used before this
        was initialized and the order was different from the order used
        in this set. Try to initialize the set earlier.
        $offOrder allows lag operations on dynamic sets, reset with
        $onOrder

What went wrong? Two elements of the set b featured already in set a that was defind before set b. Therefore the order of the elements in set b is as follows:

r2 r3 r1 r4

Obviously, this set is not orderd and hence GAMS marks error 198 when the operator ord is applied to it.

How do we fix this? In this case it is easy: we just define the set b before set a and b will be an ordered set.

Consider one last example:

Set    a   "all elements"                / r1*r10 /;
Set    b   "elements in different order" / r3, r1, r4, r7 /;
Scalar c   "counter"                     / 0 /;

loop( b $ (ord(b) > 3),
    c = c +1 );
display c;

The resulting compilation output will contain the following lines:

   5  loop( b $ (ord(b) > 3),
****                 $198
   6      c = c +1 );

What went wrong? Note that even though all elements in set b are also elements of set a, they are not specified in the same order. Therefore the set b is unordered. For more details, see section Ordered and Unordered Sets.