SimplePredictor

Algorithm that will predict the next solution based on previous solutions.

The simple predictor uses the solution from the previous time step to update the solution before a non linear solve. The simple predictor update is:

var element = document.getElementById("moose-equation-74262a8e-b39d-4597-8d47-9ffd0ff48155");katex.render("\\phi = \\phi (1 + s \\dfrac{\\Delta t}{\\Delta t_{old}}) - \\phi_{old} s \\dfrac{\\Delta t}{\\Delta t_{old}}", element, {displayMode:true,throwOnError:false});

with $var element = document.getElementById("moose-equation-e9876620-0126-465b-a71f-355ec592dd27");katex.render("\\phi", element, {displayMode:false,throwOnError:false});$ the solution vector, which includes all the application's non linear variables, and $var element = document.getElementById("moose-equation-6b02bf86-168f-4ce3-84a8-a4333b2ad4e7");katex.render("s", element, {displayMode:false,throwOnError:false});$ a scaling factor, specified by the "scale" parameter. That scaling factor is further scaled with the size of the current time step $var element = document.getElementById("moose-equation-aecdbae8-b1cc-4af0-b3c7-4bed93d3d737");katex.render("\\Delta t", element, {displayMode:false,throwOnError:false});$ divided by the previous one.

Example input syntax

In this example, a SimplePredictor is specified in the executioner to use the previous time step solution to compute better initial guesses for each non linear solve.

[Executioner]
  type = Transient
  solve_type = 'PJFNK'

  nl_max_its = 15
  nl_rel_tol = 1e-14
  nl_abs_tol = 1e-14

  start_time = 0.0
  dt = 0.5
  end_time = 1.0

  [Predictor]
    type = SimplePredictor
    scale = 1.0
  []
[]

Input Parameters

scaleThe scale factor for the predictor (can range from 0 to 1)
C++ Type:double
Unit:(no unit assumed)
Controllable:No
Description:The scale factor for the predictor (can range from 0 to 1)

Required Parameters

nl_sysnl0The nonlinear system that this predictor should be applied to.
Default:nl0
C++ Type:NonlinearSystemName
Unit:(no unit assumed)
Controllable:No
Description:The nonlinear system that this predictor should be applied to.
skip_after_failed_timestepFalseSkip prediction in a repeated time step after a failed time step
Default:False
C++ Type:bool
Unit:(no unit assumed)
Controllable:No
Description:Skip prediction in a repeated time step after a failed time step
skip_timesSkip the predictor if the current solution time is in this list of times
C++ Type:std::vector<double>
Unit:(no unit assumed)
Controllable:No
Description:Skip the predictor if the current solution time is in this list of times
skip_times_oldSkip the predictor if the previous solution time is in this list of times
C++ Type:std::vector<double>
Unit:(no unit assumed)
Controllable:No
Description:Skip the predictor if the previous solution time is in this list of times

Optional Parameters

control_tagsAdds user-defined labels for accessing object parameters via control logic.
C++ Type:std::vector<std::string>
Unit:(no unit assumed)
Controllable:No
Description:Adds user-defined labels for accessing object parameters via control logic.
enableTrueSet the enabled status of the MooseObject.
Default:True
C++ Type:bool
Unit:(no unit assumed)
Controllable:No
Description:Set the enabled status of the MooseObject.

Advanced Parameters

(moose/test/tests/predictors/simple/predictor_test.i)

# The purpose of this test is to test the simple predictor.  This is a very
# small, monotonically loaded block of material.  If things are working right,
# the predictor should come very close to exactly nailing the solution on steps
# after the first step.

# The main thing to check here is that when the predictor is applied in the
# second step, the initial residual is almost zero.

[Mesh]
  type = GeneratedMesh
  dim = 2
  xmin = 0
  xmax = 1
  ymin = 0
  ymax = 1
  nx = 3
  ny = 3
[]

[Functions]
  [ramp1]
    type = ParsedFunction
    expression = 't'
  []
[]

[Variables]
  [u]
    order = FIRST
    family = LAGRANGE
  []
[]

[Kernels]
  [diff_u]
    type = Diffusion
    variable = u
  []
[]

[BCs]
  [bot]
    type = DirichletBC
    variable = u
    boundary = bottom
    value = 0.0
  []
  [ss2_x]
    type = FunctionDirichletBC
    variable = u
    boundary = top
    function = ramp1
  []
[]

[Executioner]
  type = Transient
  solve_type = 'PJFNK'

  nl_max_its = 15
  nl_rel_tol = 1e-14
  nl_abs_tol = 1e-14

  start_time = 0.0
  dt = 0.5
  end_time = 1.0

  [Predictor]
    type = SimplePredictor
    scale = 1.0
  []
[]

[Postprocessors]
  [final_residual]
    type = Residual
    residual_type = FINAL
  []
  [initial_residual]
    type = Residual
    residual_type = INITIAL
  []
[]

[Outputs]
  csv = true
[]

Example input syntax
Input Parameters