Output System

The output system is designed to be just like any other system in MOOSE: modular and expandable.

Short-cut Syntax

To enable output in an application it must contain an [Outputs] block. The simplest method for enabling output is to utilize the short-cut syntax as shown below, which enables Exodus output for writing spatial data. A complete list of output types and the associated short-cut syntax for the framework is included in Table 1.

Listing 1: Example of short-cut syntax in the Outputs block.

[Outputs]
  exodus = true
[]

Developers may refer to CommonOutputAction for more information about implementing short-cut syntax for new Output types.

Block Syntax

To take full advantage of the output system, the use of sub-blocks is required. The block listed below is nearly identical, with respect to the internal implementation, to Listing 1, including the block name (i.e., the short-cut syntax builds this exact block). The one difference between the short-cut and block syntax is the default output filename assigned

Listing 2: Example of full block syntax in the Outputs block.

[Outputs]
  [exodus]
    type = Exodus
  []
[]

Filenames

The resulting filenames produced by different syntax is important. The default naming scheme for output files utilizes the input file name (e.g., input.i) with a suffix that differs depending on how the output is defined:

  • An "_out" suffix is used for Outputs created using the short-cut syntax.

  • The sub-block syntax uses the actual sub-block name as the suffix.

For example, if the input file (input.i) contained the following [Outputs] block, two files would be created: input_out.e and input_other.e.

Listing 3: Example of output filenames, assuming an input file name of 'input.i'.

[Outputs]
  console = true
  exodus = true    # creates input_out.e
  [other]        # creates input_other.e
     type = Exodus
  []
[]

Note, the use of "file_base" anywhere in the [Outputs] block disables all default naming behavior.

Available Output Types

Table 1 provides a list of the most common output types, including the short-cut syntax as well as the type to be used when creating a sub-block. A complete list of all available output objects is provided below.

Table 1: List of common output objects included in core MOOSE framework

Short-cutSub-block ("type=")Description
consoleConsoleWrites to the screen and optionally a file
exodusExodusThe most common,well supported, and controllable output type
vtkVTKVisualization Toolkit format, requires --enable-vtk when building libMesh
gmvGMVGeneral Mesh Viewer format
nemesisNemesisParallel ExodusII format
tecplotTecplotRequires --enable-tecplot when building libMesh
xdaXDAlibMesh internal format (ascii)
xdrXDRlibMesh internal format (binary)
csvCSVComma separated scalar values
gnuplotGNUPlotOnly support scalar outputs
checkpointCheckpointMOOSE internal format used for restart and recovery

Complete list of available Output objects.

  • Moose App
  • BlockRestrictionDebugOutputDebug output object for displaying information regarding block-restriction of objects.
  • CSVOutput for postprocessors, vector postprocessors, and scalar variables using comma seperated values (CSV).
  • CheckpointOutput for MOOSE recovery checkpoint files.
  • ConsoleObject for screen output.
  • ControlOutputOutput for displaying objects and parameters associated with the Control system.
  • DOFMapOutput degree-of-freedom (DOF) map.
  • ExodusObject for output data in the Exodus format
  • GMVObject for outputting data in the GMV format
  • GnuplotOutput for postprocessors and scalar variables in GNU plot format.
  • JSONOutput for Reporter values using JSON format.
  • MaterialPropertyDebugOutputDebug output object for displaying material property information.
  • NemesisObject for output data in the Nemesis (parallel ExodusII) format.
  • PerfGraphOutputControls output of the PerfGraph: the performance log for MOOSE
  • ProgressOutput a simulation time progress bar on the console.
  • ReporterDebugOutputDebug output object for displaying Reporter information.
  • SolutionHistoryOutputs the non-linear and linear iteration solve history.
  • TecplotObject for outputting data in the Tecplot format
  • TopResidualDebugOutputDebug output object for displaying the top contributing residuals.
  • VTKOutput data using the Visualization Toolkit (VTK).
  • VariableResidualNormsDebugOutputReports the residual norm for each variable.
  • XDAObject for outputting data in the XDA/XDR format.
  • XDRObject for outputting data in the XDA/XDR format.
  • XMLOutputOutput for VectorPostprocessor using XML format.

Multiple Output Blocks

It is possible to create multiple output objects for outputting:

  • at specific time or timestep intervals,

  • custom subsets of variables, and

  • to various file types.

  • Supports common parameters and sub-blocks.

Listing 4: Example of creating multiple Output objects.

[Outputs]
  vtk = true
  [console]
    type = Console
    perf_log = true
    max_rows = true
  []
  [exodus]
    type = Exodus
    execute_on = 'timestep_end'
  []
  [exodus_displaced]
    type = Exodus
    file_base = displaced
    use_displaced = true
    interval = 3
  []
[]

Common Parameters

The Outputs block supports common parameters. For example, "execute_on" may be specified outside of individual sub-blocks, indicating that all sub-blocks should output the initial condition, for example.

If within a sub-block the parameter is given a different value, the sub-block parameter takes precedence.

The input file snippet below demonstrate the usage of a common values as well as the use of multiple output blocks.

commentnote:Console objects are a special case.

Console-based outputs have special inheritance of common parameters: the execute_on parameter, as detailed below, does not get inherited.

Listing 5: Example use of common parameters in the Outputs block.

[Outputs]
  execute_on = 'timestep_end' # disable the output of initial condition
  interval = 2                # only output every 2 timesteps
  vtk = true                  # output VTK file
  print_perf_log = true       # enable the performance log
  [initial]
    type = Exodus
    execute_on = 'initial'    # this ExodusII file will contain ONLY the initial condition
  []
  [displaced]
    type = Exodus
    use_displaced = true
    interval = 3              # this ExodusII will only output every third time step
  []
[]

Developers may refer to CommonOutputAction for more information about implementing new common output parameters.

Controlling output frequency

Several parameters are available common to all output objects to control the frequency with which output occurs.

  • interval = N will cause output objects to output only every _Nth_ simulation step

  • if start_time and/or end_time are specified, outputs will only happen after the given start time and/or before the given end time

  • sync_times = 't1 t2 t3 ... tN will cause MOOSE to hit each listed simulation time _t1 .. tN_ exactly. With sync_only = true outputs will _only_ happen at the specified sync times.

  • minimum_time_interval = dt will suppress any output if the previous output happened at an earlier time that is more recent than _dt_ time units ago (specifically this means that outputs during linear iterations are not suppressed, as they are happening at the _same_ simulation time, and output from failed, cut steps do not lead to output suppression in repeated steps as they happened in the future)

Outputs and execute_on

Outputs utilize the "execute_on" parameter like other systems in MOOSE, by default:

  • All objects have execute_on = 'initial timestep_end', with Console objects being the exception.

  • Console objects append 'initial timestep_begin linear nonlinear failed' to the "execute_on" settings at the common level.

A list of available "execute_on" flags for Output objects is provided in Table 2 and a convenience parameter, "additional_execute_on", allows appending flags to the existing "execute_on" flags.

When debugging output, use the --show-outputs flag when executing your application. This will add a section near the top of the simulation that shows the settings being used for each output object.

  • The toggles shown below provide additional operate by appending to the "execute_on" flags.

  • Currently, the following output execution times are available:

Table 2: List of available 'execute_on' values for Output objects.

Text FlagDescription
"none"disables the output
"initial"executes the output on the initial condition (on by default)
"linear"executes the output on each linear iteration
"nonlinear"executes the output on each non-linear iteration
"timestep_end"calls the output method at the end of the timestep (on by default)
"timestep_begin"executes the output method at the beginning of the timestep
"final"calls the output method on the final timestep
"failed"executes the output method when the solution fails

As detailed in the Creating Custom Output Object section, there are two types of outputs classes: BasicOutput and AdvancedOutput. Advanced outputs have additional control beyond execute_on, as detailed in the table below.

Input ParameterMethod Controlled
execute_postprocessors_onoutputPostprocessors
execute_vector_postprocessors_onoutputVectorPostprocessors
execute_elemental_onoutputElementalVariables
execute_nodal_onoutputNodalVariables
execute_scalar_onoutputScalarVariables
execute_system_information_onoutputSystemInformation
execute_input_onoutputInput

Each of the output controls accept the same output execution flags that execute_on utilizes. In AdvancedOutput objects, the execute_on settings are used to set the defaults for each of the output type controls. For example, setting execute_on = 'initial timestep_end' causes all of the output types (e.g., postprocessors, scalars, etc.) to execute on each timestep and the initial condition.

Mesh Displacements and Higher Order Meshes

The displaced mesh may be output by setting 'use_displaced = true' within your output sub-blocks. To output both the original and displaced mesh an additional output block is required.

warningwarning:Be careful when outputting a displaced mesh!

In general, it is not necessary to output the displace mesh, since most visualization tools automatically apply displacements. Because of this, writing the displaced mesh can result in visualizations that apply displacements multiple times.

If a simulation is using a higher order mesh, oversampling may be required to provide an accurate representation of the finite element solution. Visualization tools generally perform linear interpolation between data, regardless of the mesh order. Oversampling, which is controlled by setting the "refinements" parameter, will evaluate the finite element solution on a uniformly refined mesh during output to provide a improved view of the existing solution.

Listing 6: Example showing output of displaced mesh and oversampling.

[Outputs]
  console = true
  exodus = true
  [oversample]
     type = Exodus
     refinements = 2
     use_displaced = true
  []
[]

In-Block Output Control

For certain objects it is possible to control output within the block itself, these objects include: Postprocessors, VectorPostprocessors, Indicators, Markers, Variables, and AuxVariables. For example, consider the following input file that has three output objects defined and a single postprocessor as well.


[Postprocessors]
  [pp]
    type = EmptyPostprocessor
    outputs = 'csv'
  []
[]

[Outputs]
  exodus = true
  csv = true
  [vtk]
    type = VTK
    interval = 2
  []
[]

The outputs that should include the postprocessor value may be listed in the "outputs" parameter.

The "outputs" parameter limits the defined postprocessor to output only to the csv output object. This highlights the need to understand the naming convention utilized by the short-cut syntax. It is also possible to remove the postprocessor from all outputs by specifying "outputs = none".

Non-linear/Linear Residual Output

Any object inheriting from PetscOutput has the ability to output data on non-linear and/or linear iterations. To enable this add "nonlinear" and/or "linear" to the "execute_on" input parameter.


[Outputs]
  [exodus]
    type = Exodus
    execute_on = 'timestep_end linear nonlinear'
  []
[]

When outputting nonlinear/linear iterations the time is changes from the actual simulation time by creating pseudo time steps. For example, if the [Outputs] block above was associated with a simulation that was taking a time step of 0.1 the resulting output times would use the following convention:

Output TimeDescription
0.2Converged solve for time = 0.2
0.2001First non-linear iteration for time = 0.3
0.2001001First linear iteration for the first non-linear iteration
0.2001002Second linear iteration for the first non-linear iteration
0.2002Second non-linear iteration for time = 0.3
0.2002001First linear iteration for the second non-linear iteration
0.2002002Second linear iteration for the second non-linear iteration
0.3Converged solve for time = 0.3

Creating Custom Output Object

When creating a new output object, the new object must inherit from one of two templated base classes: BasicOutput or AdvancedOutput.

For either base class, the template parameter should be one of the following four output classes:

Base ClassDescription
Outputthe most general output base class, this should be used for simple output classes that require very little control over execution, see MaterialPropertyDebugOutput
PetscOutputprovides the ability to execute output calls on linear and nonlinear residual calculations.
FileOutputadds the basic functions and parameters need to write to a file.
OversampleOutputadds the ability to utilize an oversampled mesh for outputting.

Note, the four classes listed above inherit from each other, so FileOutput is a PetscOutput, see the inheritance diagram for a visual representation: Output.

Creating a BasicOutput

Basic output objects are designed for simple output cases that perform a single output task and do not require control over individual types of output.

  • When creating a basic output object, the user must override a single virtual method: output(const OutputExecFlagType & type).

  • This method should perform all the necessary commands to perform the output.

  • The OutputExecFlagType specifies what execute flag the output is currently being called with(e.g., "initial"). This type is a proper Enum and the possible values are listed in include/base/Moose.h.

Creating an AdvancedOutput

Advanced output objects provide additional control over various output types (e.g., postprocessors). When creating an advanced output object a call to the static method enableOutputTypes must exist in the new objects validParams method that indicates which types of outputs the new object will be responsible for outputting. For example, the Exodus output validParams method includes:


InputParameters params = AdvancedOutput<OversampleOutput>::validParams();
params += AdvancedOutput<OversampleOutput>::enableOutputTypes("nodal elemental scalar
                                                              postprocessor input");

Each of the keywords listed in this method enable a call to the associated output method, as detailed in the following table.

Enable KeywordAssociated Method
"nodal"outputNodalVariables
"elemental"outputElementalVariables
"scalar"outputScalarVariables
"postprocessor"outputPostprocessors
"vector_postprocessor"outputVectorPostprocessors
"input"outputInput
"system_information"outputSystemInformation

These methods are the virtual methods that must be overloaded in the custom output object. For example, if "nodal" output in enabled the outputNodalVariables should be overloaded. Each of these methods includes the OutputExecFlagType as an input variable to the method.