Second method starts but does not do anything

[shor-ty]

Hey all, I am getting more familiar with dakota at the moment and I want to execute two sample methods (LHS). The first method is called and the sampler starts while in the second method, the samples does not output anything.

I guess I miss some fundamentals of how dakota works.
Here is the input stuff:

environment
  top_method_pointer = "sequence"
  tabular_data
    tabular_data_file = 'lhs_results.dat'

method
  id_method = "sequence"
  hybrid
    sequential
    method_pointer_list = "sampling1" "sampling2"

method
  id_method = "sampling1"
  output = silent
  sampling
    sample_type = lhs
    samples = 5
    seed = 12345

  model_pointer = "single"

model
  id_model = "single"
  single
    interface_pointer = "interface1"

  variables_pointer = "variables1"
  responses_pointer = "response1"

variables
  id_variables = "variables1"
  uniform_uncertain = 2
    descriptors = 'x' 'y'
    lower_bounds = -2  -2.0
    upper_bounds = 2.0 2.0

interface
  id_interface = "interface1"
  system
      analysis_drivers = 'interface'
      parameters_file = 'parFile'
      results_file = 'resFile'

responses
  id_responses = "response1"
  response_functions = 1
  no_gradients
  no_hessians

#-------------------------------------------

method
  id_method = "sampling2"
  sampling
    sample_type = lhs
    samples = 15
    seed = 108

  model_pointer = "model2"



model
  id_model = "model2"
  single
    interface_pointer = "interface2"

  variables_pointer = "variables2"
  responses_pointer = "response2"

variables
  id_variables = "variables2"
  uniform_uncertain = 2
    descriptors = 'x' 'y'
    lower_bounds = -1  -1.0
    upper_bounds = 1.0 1.0

interface
  id_interface = "interface2"
  system
      analysis_drivers = 'interface'
      parameters_file = 'parFile2'
      results_file = 'resFile2'

responses
  id_responses = "response2"
  response_functions = 1
  no_gradients
  no_hessians

Any suggestion is warmly welcomed.
Best, Tobi

[thenectarine]

The Dakota Reference Manual states the following for the hybrid method and sequential keyword.

In a hybrid minimization method (hybrid), a set of methods synergistically seek an optimal design.

In the sequential approach, methods are run one at a time, in sequence. The best solutions from one method are used to initialize the next method.

This implies that

1. Hybrid and method_pointer_list are for optimization methods.
2. The solution from the first method is passed on to the second method.

Since the provided Dakota input file has 2 methods for sampling, both of which are not optimization methods, the hybrid procedure has no solution to map from the first to second method and terminates before starting the second method.

I dug around the Dakota documentation for 10 mins, but I did not find a way to perform 2 sampling methods in one Dakota input file. I would just create separate input files for each desired sampling method.