1Blurb:: Select the models that write evaluation data to HDF5
2
3Description::
4
5By default, when HDF5 output is enabled, Dakota writes evaluation
6data for all simulation interfaces (interfaces of type fork, system, direct, etc, but
7not approximation interfaces, which are constructed by models of type surrogate). This
8keyword group is used to override the default.
9
10HDF5 output is an experimental feature, and the contents and organization of
11the output file is subject to change. The current organization and a
12brief explanation of HDF5 is provided in the \ref hdf5_output section
13of this manual.
14
15The example below will be used to explain the effect of each keyword.
16
17Topics:: dakota_output
18
19Examples::
20
21\verbatim
22environment
23  results_output
24      hdf5
25       # interface_selection
26       #  simulation
27       #  all
28       #  none
29      results_output_file 'my_results'  # The .h5 extension will be added
30
31   method_pointer 'opt'
32
33method
34  id_method 'opt'
35    optpp_q_newton
36  model_pointer 'surr'
37
38model
39  id_model 'surr'
40  surrogate global gaussian_process surfpack
41  dace_method_pointer 'training'
42
43method
44  id_method 'training'
45  sampling
46    seed 1234
47    samples 20
48  model_pointer 'truth_m'
49
50model
51  id_model 'truth_m'
52  simulation
53
54interface
55  id_interface 'truth'
56  direct
57    analysis_drivers 'text_book'
58
59variables
60  continuous_design 2
61    descriptors 'x1' 'x2'
62    lower_bounds -2.0 -2.0
63    upper_bounds  2.0  2.0
64
65responses
66  objective_functions 2
67    descriptors 'f1' 'f2'
68  analytic_gradients
69  no_hessians
70
71\endverbatim
72
73
74Theory::
75Faq::
76See_Also::
77