Operations inside the variable system

Perform certain operations

It is possible to do operations with variables and store them in the variable table in the prescribed way. It is defined in tem_variable_module.

What is available?

Here is a list of the currently available operations for the variable system.

  • 'difference'

Evaluate the function pointers of the dependent variables, and then calculate the difference between these two. ( scalar or vector ) In lua file, first define new variable with varType operation kind as difference and provide two dependent variable via input_varname. If input_varname variable is not part of predefined solver variables then add also that variable via variable table for example spacetime function variable. For example: Define a variable called dens_difference, which depends on density and spacetime. One can get an error between simulation results and analytical solution. @note Note The number of input variables must be = 2. res(:) = a(:) - b(:)

variable = {
  {
    name = 'dens_reference',
    ncomponents = 1,
    vartype = "st_fun",
    st_fun = luaFun
  },
  {
    name = 'dens_difference',
    ncomponents = 1,
    vartype = "operation",
    operation = {
      kind = 'difference',
      input_varname = {'density', 'dens_reference'}
    }
  }
}
tracking = {
  variable = {'dens_difference'},
  folder = 'tracking/',
  shape = {
    kind = 'canoND',
    object = {
      origin = {3.0, 3.1, 3.0}
    }
  },
  format = 'ascii',
  time = {
    min = 0,
    max = tmax,
    interval = 1
  }
}
  • 'rel_difference'

@note Note Number of input variables = 2. res = ( a(:) - b(:) ) / b(:)

variable = {
  {
    name = 'dens_ref',
    ncomponents = 1,
    vartype = "st_fun",
    st_fun = luaFun
  },
  {
    name = 'dens_relDiff',
    ncomponents = 1,
    vartype = "operation",
    operation = {
      kind = 'rel_difference',
      input_varname = {'density', 'dens_ref'}
    }
  }
}
tracking = {
  variable = {'dens_relDiff'},
  folder = 'tracking/',
  shape = {
    kind = 'canoND',
    object = {
      origin = {3.0, 3.1, 3.0}
    }
  },
  format = 'ascii',
  time = {
    min = 0,
    max = tmax,
    interval = 1
  }
}
  • 'addition'

@note Note var3(:) = var1(:) + var2(:)

variable = {
  {
    name = 'var1',
    ncomponents = 3,
    vartype = "st_fun",
    st_fun = luaFun
  },
    name = 'var2',
    ncomponents = 3,
    vartype = "st_fun",
    st_fun = luaFun
  },
  {
    name = 'var3',
    ncomponents = 3,
    vartype = "operation",
    operation = {
      kind = 'addition',
      input_varname = {'var1', 'var2'}
    }
  }
}
  • 'multiplication'

Routine to multiply variables if all variables have same number of components. @note Note res(:) = a(:) * b(:) res(1) = a(1) * b(1)

variable = {
  {
    name = 'coeff',
    ncomponents = 1,
    vartype = "st_fun",
    st_fun = 0.25
  },
  {
    name = 'newVel',
    ncomponents = 1,
    vartype = "operation",
    operation = {
      kind = 'multiplication',
      input_varname = {'coeff', 'vel_mag'}
    }
  },
  ...
}
  • 'division', 'div'

Routine to divide variables if all variables have same number of components. @note Note res(:) = a(:) / b(:)

variable = {
  {
    name = 'coeff',
    ncomponents = 3,
    vartype = "st_fun",
    st_fun = 0.25
  },
  {
    name = 'newVel',
    ncomponents = 1,
    vartype = "operation",
    operation = {
      kind = 'division',
      input_varname = {'velocity', 'coeff'} -- numerator, denominator
    }
  },
  ...
}
  • 'divide_vector_by_scalar'

Routine to divide a vector by a scalar. So the second input variable must have one component. @note Note res(:) = a(:) / b

variable = {
  {
    name = 'coeff',
    ncomponents = 1,
    vartype = "st_fun",
    st_fun = 0.25
  },
  {
    name = 'newVel',
    ncomponents = 1,
    vartype = "operation",
    operation = {
      kind = 'divide_vector_by_scalar',
      input_varname = {'velocity', 'coeff'} -- numerator, denominator
    }
  },
  ...
}
  • 'gradient', 'grad'

Calculates the gradient of the input variable. Only one input variable allowed. Number of components of the input variable can be 1,2 or 3. The resulting number of components must be ncomponents + 1. Only available in Ateles so far.

variable = {
  {
    name = 'dens_grad',
    ncomponents = 2, -- density: ncomponents=1
    vartype = "operation",
    operation = {
      kind = 'gradient',
      input_varname = 'density'
    }
  },
  ...
}
  • 'magnitude'

Evaluate magnitude of any vectorial variable. Number of components and number of input variables both have to be 1. In lua file, first define new variable with varType operation kind as "magnitude" and provide name of the variable from which magnitude to be derived in input_varname. If input_varname variable is not part of predefined solver variables then add also that variable via variable table. @note Note res = sqrt(sum( a(:) ))

variable = {
  {
    name = 'velMag',
    ncomponents = 1,
    vartype = "operation",
    operation = {
      kind = 'magnitude',
      input_varname = {'velocity'}
    }
  },
}
tracking = {
  variable = {'velMag'},
  folder = 'tracking/',
  shape = {
    kind = 'canoND',
    object = {
      origin = {3.0, 3.1, 3.0}
    }
  },
  format = 'ascii',
  time = {
    min = 0,
    max = tmax,
    interval = 1
  }
}
  • 'extract'

Extract component index of any vectorial variable. In lua file, first define new variable with varType operation kind as "extract" and provide name of the variable from which to extract component index via input_varname and index to extract via input_varIndex. If input_varname variable is not part of predefined solver variables then add also that variable via variable table. @note Note Both, ncomponents and number of input variables must be 1.

variable = {
  {
    name = 'vel_y',
    ncomponents = 1,
    vartype = "operation",
    operation = {
      kind = 'extract',
      input_varname = 'velocity',
      input_varindex = 2
    }
  }
}
  • 'combine'

Combine multiple variables into single variable with nComponent of output variable as sum of all input variables nComponents. In lua file, first define new variable with varType operation kind as "combine" and provide name of the variable from which to extract component index via input_varname (it must be single variable) and index to combine via input_varIndex. If input_varname variable is not part of predefined solver variables then add also that variable via variable table.

variable = {
  {
    name = 'dens_and_vel',
    ncomponents = 4,
    vartype = "operation",
    operation = {
      kind = 'combine',
      input_varname = {'density', 'velocity'}
    }
  },
}
  • Boolean operationas and comparisons
  • 'greater_than', 'gt', '>'
  • 'greater_than_or_equal', 'ge', '>='
  • 'less_than', 'lt', '<'
  • 'less_than_or_equal', 'le', '<='
  • 'equal', 'eq', '='
  • 'not_equal', 'ne', '/='
  • 'and'
  • 'or'

Using a boolean operation one will get either 1.0 (true) or 0.0 (false) as a resulting variable value. One must provide two input variables that must have the same number of components. Here is an example:

variable = {
  {
    name = 'var3',
    ncomponents = 3,
    vartype = "operation",
    operation = {
      kind = 'equal',
      input_varname = {'var1', 'var2'}
    }
  },
}

-- if var1 == var2 then:
-- var3 = {1.0, 1.0, 1.0}

The temporal reduction operation allows the reduction of values over given iteration invervals. Thus it allows temporal averaging, as shown here:

  variable = {
    {
       name = 'press_timeavg',
       ncomponents = 1,
       vartype = "operation",
       operation = {
         kind='reduction_transient',
         input_varname={'pressure'},
         reduction_transient = {
           kind    = 'average',
           nrecord = 1000
         }
       }
    }
  }

Other implemented reductions are sum, min and max.