1#@ *: DakotaConfig=HAVE_NPSOL
2# DAKOTA INPUT FILE - dakota_sbo_barnes.in
3
4# Demonstrates the use of approximation models and a trust region
5# optimization environment in performing constrained minimization on
6# the textbook test function.
7
8# These tests exercise global, local, and multipoint surrogates for
9# feasible and infeasible starting points with and without constraint
10# relaxation.
11
12# Test s16 is s14, but with new gauss_proc surrogate; worse
13# performance, but tests analytic gradient; hopefully will improve
14# once trend is added.
15
16environment,
17	method_pointer = 'SBLO'
18
19method,
20	id_method = 'SBLO'
21	surrogate_based_local
22	model_pointer = 'SURROGATE'
23	approx_method_pointer = 'NLP'
24	max_iterations = 50
25	trust_region
26	  initial_size = 0.10
27	  contraction_factor = 0.5
28#	  constraint_relax homotopy		#s6,#s8,#s9,#s11
29	  expansion_factor   = 1.50
30	  soft_convergence_limit = 1		# Slowed convergence criteria
31	  minimum_size = 0.001
32
33method,
34	id_method = 'NLP'
35	npsol
36	  max_iterations = 50
37	  convergence_tolerance = 1e-4
38
39model,
40	id_model = 'SURROGATE'
41	responses_pointer = 'SURROGATE_RESP'
42	surrogate global                  	#s0,#s2,#s3,#s5,#s6,#s8,#s9,#s11,#s14,#s15,#s16,#s17
43	  dace_method_pointer = 'SAMPLING'	#s0,#s2,#s3,#s5,#s6,#s8,#s9,#s11,#s14,#s15,#s16,#s17
44	  polynomial quadratic			#s0,#s2,#s3,#s5,#s6,#s8,#s9,#s11
45#	  gaussian_process surfpack		#s14,#s15
46#	  experimental_gaussian_process		#s16
47#	    find_nugget	1                       #s16
48#	    trend none                          #s16
49# Emulate s2 with new surrogate poly:
50#         experimental_polynomial		#s17
51#           basis_order 2			#s17
52#	  use_derivatives			#s15
53	  correction additive first_order	#s0,#s3,#s6,#s9
54# 	surrogate local taylor_series		#s1,#s4,#s7,#s10
55#	  actual_model_pointer = 'TRUTH'	#s1,#s4,#s7,#s10
56# 	surrogate multipoint tana		#s12,#s13
57#	  actual_model_pointer = 'TRUTH'	#s12,#s13
58
59variables,
60	continuous_design = 2
61#   Feasible starting point
62	  initial_point    30.   40.	#s0,#s1,#s2,#s3,#s4,#s5,#s12,#s14,#s15,#s16,#s17
63# Infeasible starting point
64#	  initial_point    65.    1.
65# Infeasible starting point
66#	  initial_point    10.   20.	#s6,#s7,#s8,#s9,#s10,#s11,#s13
67	  lower_bounds      0.    0.
68	  upper_bounds     80.   80.
69	  descriptors      'x1'  'x2'
70
71responses,
72	id_responses = 'SURROGATE_RESP'
73	objective_functions = 1
74	nonlinear_inequality_constraints = 3
75	nonlinear_inequality_lower_bounds = 0.     0.     0.
76	nonlinear_inequality_upper_bounds = 1.e+50 1.e+50 1.e+50
77	analytic_gradients
78	no_hessians
79
80###############################################
81# SAMPLING method specifications for building #
82# surrogate function(s)			      #
83###############################################
84method,
85	id_method = 'SAMPLING'
86	model_pointer = 'TRUTH'
87	sampling
88	  seed = 12345  samples = 10
89
90model,
91	id_model = 'TRUTH'
92	single
93	  interface_pointer = 'TRUE_FN'
94	  responses_pointer = 'TRUE_RESP'
95
96interface,
97	direct #system #asynchronous
98	id_interface = 'TRUE_FN'
99 	  analysis_driver = 'barnes'
100
101responses,
102	id_responses = 'TRUE_RESP'
103	objective_functions = 1
104	nonlinear_inequality_constraints = 3
105	nonlinear_inequality_lower_bounds = 0.     0.     0.
106	nonlinear_inequality_upper_bounds = 1.e+50 1.e+50 1.e+50
107#	no_gradients		 	#s2,#s8,#s14,#s16,#s17
108#	numerical_gradients		#s7
109#	  method_source dakota		#s7
110	analytic_gradients		#s0,#s1,#s3,#s4,#s5,#s6,#s9,#s10,#s11,#s12,#s13,#s15
111	no_hessians
112