1#@ s*: Label=FastTest
2#@ *: DakotaConfig=HAVE_AMPL
3#@ *: DakotaConfig=HAVE_NPSOL
4# Solve the AMPL diet1 problem: minimize cost subject to buying at
5# least one entree, side, and drink.  The optimal objective is 2.81.
6#
7# The AMPL model files are generated from pyomo:
8#   pyomo diet1.py diet.dat --save-model dakota_ampl_diet.nl --symbolic-solver-labels
9#
10# This test is protecting parsing of AMPL labels with spaces and AMPL
11# responses that are ordered differently than the Dakota reponses.
12#
13# Technically the decision variables are discrete, but the SQP finds
14# the correct solution on the bounds of the domain
15
16method,
17	npsol_sqp
18	  max_iterations = 50,
19	  convergence_tolerance = 1e-12
20
21variables,
22	continuous_design = 9
23	  initial_point    9 * 0.5
24	  upper_bounds     9 * 1.0
25	  lower_bounds     9 * 0.0
26	  descriptors
27             'Buy[Sausage_McMuffin]'
28             'Buy[Orange_Juice]'
29             'Buy[Big_Mac]'
30             'Buy[Filet-O-Fish]'
31             'Buy[Quarter_Pounder_w_Cheese]'
32             'Buy[1%_Lowfat_Milk]'
33             'Buy[Fries,_small]'
34             'Buy[McGrilled_Chicken]'
35             'Buy[McLean_Deluxe_w_Cheese]'
36
37interface,
38	  algebraic_mappings = 'dakota_ampl_diet.nl'
39	  asynchronous
40
41responses,
42	objective_functions = 1
43	nonlinear_inequality_constraints = 3
44	  lower_bounds = 3 * 1.0
45	  upper_bounds = 3 * 20.0
46	response_descriptors =
47	  'Total_Cost'
48	  'Entree'
49	  'Side'
50	  'Drink'
51	analytic_gradients
52	no_hessians
53
54