next up previous
Next: 3.1.1 Names Up: 3 The Standard Data Previous: 3 The Standard Data


3.1 Introduction to the Standard Data Input Format

As we have just said, the data format is designed to be compatible with the MPS linear programming format. There are, however, extensions to allow the user to input nonlinear problems. The user must prepare an input file consisting of three types of cards:

Indicator cards contain a simple keyword to specify the type of data that follows. The first character of such cards must be in column 1; indicator cards are the only cards, with the exception of comment cards, which start in column 1. Possible indicator cards are given in Table 3.1.

Table 3.1: Possible indicator card
Keyword Comments Presence Described in §
NAME   mandatory 3.2.1
  either    
GROUPS   mandatory 3.2.6
ROWS synonym for GROUPS   3.2.6
CONSTRAINTS synonym for GROUPS   3.2.6
VARIABLES   mandatory 3.2.7
COLUMNS synonym for VARIABLES   3.2.7
  or    
VARIABLES   mandatory 3.2.8
COLUMNS synonym for VARIABLES   3.2.8
GROUPS   mandatory 3.2.9
ROWS synonym for GROUPS   3.2.9
CONSTRAINTS synonym for GROUPS   3.2.9
CONSTANTS   optional 3.2.10
RHS synonym for CONSTANTS   3.2.10
RHS' synonym for CONSTANTS   3.2.10
RANGES   optional 3.2.11
BOUNDS   optional 3.2.12
START POINT   optional 3.2.13
QUADRATIC optional 3.2.14
HESSIAN synonym for QUADRATIC   3.2.14
QUADS synonym for QUADRATIC   3.2.14
QUADOBJ synonym for QUADRATIC   3.2.14
QSECTION synonym for QUADRATIC   3.2.14
ELEMENT TYPE   optional 3.2.15
ELEMENT USES   optional 3.2.16
GROUP TYPE   optional 3.2.17
GROUP USES   optional 3.2.18
OBJECT BOUND   optional 3.2.19
ENDATA   mandatory 3.2.2


Indicator cards must appear in the order shown, except that the GROUPS and VARIABLES sections may be interchanged to allow specification of the linear terms by rows or columns. The cards CONSTANTS, RHS', RHS, RANGES, BOUNDS, START POINT, QUADRATIC, HESSIAN, QUADS, QUADOBJ, QSECTION, ELEMENT TYPE, ELEMENT USES, GROUP TYPE, GROUP USES and OBJECT BOUND are optional.

The data cards are divided into six fields. The content of each field varies with each type of data card as described in Section 3.2. Those in fields 1, 2, 3 and 5 must always be left justified within the field. Field 1, which appears in columns 2 and 3, may contain a code (that is, a one or two character string which defines the expected contents of the remaining fields on the card), fields 2, 3 and 5 may hold names and fields 4 and 6 might store numerical values. The numerical values are defined by up to 12 characters which may include a decimal point and an optional sign (a positive number is assumed unless a - is given). The value may be followed by a decimal exponent, written as an E or D, followed by a signed or unsigned one or two digit integer; the first blank after the E or D terminates the field. The names of variables, nonlinear elements or groups may be up to ten characters long. These names may include integer indices (see Section 3.1.1).

Any card with the character * in column 1 is a comment card; the remaining contents of the card are ignored. Such a card may appear anywhere in the data file. In addition, completely blank cards are ignored when scanning the input file and may thus be used to space the data. Finally, the presence of a $ as the first character in fields 3 or 5 of a data card indicates that the content of the remaining part of the card is a comment and will be ignored.



Subsections
next up previous
Next: 3.1.1 Names Up: 3 The Standard Data Previous: 3 The Standard Data