1Blurb::
2Use AMPL to define algebraic input-output mappings
3
4Description::
5Dakota can evaluate algebraic input-output mappings using AMPL
6\cite Fou03.  The mappings are expressed in 3 files: \c stub.nl, \c
7stub.col, and \c stub.row, where \c stub is a particular root name
8describing a particular problem. The file names are communicated to
9%Dakota using the \c algebraic_mappings keyword. It may either specify
10the full \c stub.nl filename, or alternatively, just the \c stub
11basename.
12
13%Dakota then extracts the input and output identifier strings from \c
14stub.col and \c stub.row and employs the AMPL solver library \cite Gay97
15to process the directed acyclic graphc (DAG) specification in \c
16stub.nl. The variable and objective function names declared within
17AMPL should be a subset of the variable and response descriptors
18specified in the \ref variables and \ref responses blocks. Ordering is
19not important, as %Dakota will reorder data as needed.
20
21Topics::
22
23Examples::
24An interface employing both algebraic and simulation-based mappings.
25The results from the individual algebraic and simulation mappings are
26overlaid based on the variable and response descriptors used by the
27individual mappings.
28
29\verbatim
30interface,
31	algebraic_mappings = 'ampl/fma.nl'
32	fork
33	  analysis_driver = 'text_book'
34	  parameters_file = 'tb.in'
35	  results_file    = 'tb.out'
36\endverbatim
37
38Theory::
39Faq::
40See_Also::
41