list_parameter_study
Samples variables as a specified values
Topics
parameter_studies
Specification
Alias: None
Arguments: None
Child Keywords:
Required/Optional |
Description of Group |
Dakota Keyword |
Dakota Keyword Description |
---|---|---|---|
Required (Choose One) |
Points Source |
List of variable values to evaluate in a list parameter study |
|
File containing list of variable values to evaluate in a list parameter study |
|||
Optional |
Identifier for model block to be used by a method |
Description
Dakota’s list parameter study allows for evaluations at user selected points of interest.
Default Behavior
By default, the list parameter study operates over all types of variables.
The number of real values in the list_of_points
specification or
file referenced by import_points_file
must be a multiple of the
total number of variables (including continuous and discrete types)
contained in the variables specification.
Expected Outputs
A list parameter study produces a set of responses for each parameter set that is specified.
Expected HDF5 Output
If Dakota was built with HDF5 support and run with the
hdf5
keyword, this method
writes the following results to HDF5:
Usage Tips
This parameter study simply performs simulations for the first parameter set (the first
n
entries in the list), followed by the next parameter set (the nextn
entries), and so on, until the list of points has been exhausted.Since the initial values from the variables specification will not be used, they need not be specified.
When the points are specified in the Dakota input file using the
list_of_points
keyword, discrete set values must be referred to using 0-based indexes, and not the values themselves. However, when usingimport_points_file
, refer to discrete set values directly, not by index.For both
list_of_points
andimport_points_file
, Dakota expects the values for each evaluation to be ordered by type. The type ordering matches that of thevariables
section and of the listing in the Parameters file format section. When multiple variables are present for a single type, the ordering within that type must match the order specified by the user in the variables section of her input file.
Examples
This shows the method and variables block of a Dakota input file that runs a list_parameter_study.
method
list_parameter_study
list_of_points =
3.1e6 0.0029 0.31
3.2e6 0.0028 0.32
3.3e6 0.0027 0.34
3.3e6 0.0026 0.36
variables
continuous_design = 3
descriptors = 'E' 'MASS' 'DENSITY'
Note that because of the way Dakota treats whitespace, the above example is equivalent to:
method
list_parameter_study
list_of_points =
3.1e6 0.0029 0.31 3.2e6 0.0028
0.32 3.3e6 0.0027
0.34 3.3e6 0.0026 0.36
variables
continuous_design = 3
descriptors = 'E' 'MASS' 'DENSITY'
Although the first example is much more readable.
And here’s a full input file:
# tested on Dakota 6.0 on 140501
environment
tabular_data
tabular_data_file 'List_param_study.dat'
method
list_parameter_study
list_of_points = 0.1 0.1
0.2 0.1
0.3 0.0
0.3 1.0
model
single
variables
active design
continuous_design = 2
descriptors 'x1' 'x2'
continuous_state = 1
descriptors = 'constant1'
initial_state = 100
interface
analysis_drivers 'text_book'
fork
asynchronous
evaluation_concurrency 2
responses
response_functions = 1
no_gradients
no_hessians
This example illustrates the list_parameter_study.
The function evaluations are independent, so any level of
evaluation_concurrency
can be usedDefault behavior for parameter studies is to iterate on all variables. However, because
active
design
is specified, this study will only iterate on thecontinuous_design
variables.