Running the Numbat application

Commandline

Most often, Numbat will be run from the commandline using


./numbat-opt -i input.i

Numbat (and all MOOSE applications) have a large number of commandline options available to the user. The complete list can be viewed using the --help option


./numbat-opt --help

Recovering

If you output checkpoint files (using checkpoint = true in your Outputs block) then the --recover option will allow you to continue a solve that died in the middle of the solve, perhaps because the job ran out of time on the cluster you were using.

We recommend that all input files for large Numbat simulations enable checkpointing. This can be enabled using


[Outputs]
  perf_graph = true
  [./csvoutput]
    type = CSV
    file_base = 2DSF
    execute_on = 'INITIAL TIMESTEP_END'
  [../]
  [./checkpoint]
    type = Checkpoint
    num_files = 2
  [../]
[]
(tests/2D/2DSF_recover.i)

For all of the options available for checkpointing, see the MOOSE documentation.

If a long-running simulation does fail to complete, it can be recovered by running


./numbat-opt --recover checkpoint_dir/XXXX -i input.i

where checkpoint_dir is the subdirectory where the checkpoint files are saved, and XXXX is the number of one of the available checkpoint files.

Overriding parameters

MOOSE provides a handy feature where any parameter in the input file can be overridden from the commandline, making it possible to script studies where only parameters are changed from simulation to simulation.

For example, assume that the anisotropy gamma of the porous medium is set as 1 in the input file


[Kernels]
  [./Darcy]
    type = NumbatDarcySF
    variable = streamfunction
    concentration = concentration
    gamma = 1
  [../]
[]

This value can be changed to 0.5 by running Numbat with the following commandline option


./numbat-opt -i input.i Kernels/Darcy/gamma=0.5

Graphical user interface (Peacock)

MOOSE provides a graphical user interface, Peacock, which can be used to both run simulations and create input files. Starting Peacock from within the base Numbat directory allows Peacock to extract the Numbat syntax, so that all Numbat objects are available in the menus.

Peacock can be run using


$MOOSE_DIR/python/peacock/peacock -i input.i

where $MOOSE_DIR is the directory where the MOOSE repository is located.

note

It is not recommended to use Peacock to run very large models (e.g. three dimensional simulations) that require lots of memory.