C.1 The Minimos-NT Post-Processing System

Despite the fact that the contents of both output files of MINIMOS-NT can be configured by the user in a very flexible way, it is still necessary to transform or process these outputs after the actual simulation has been finished. In order to simplify the interface between MINIMOS-NT and such post-processing steps, a new post-processing system directly administered by MINIMOS-NT has been introduced.

Most of the post-processing is normally done in separate scripts, which read and parse the output files of MINIMOS-NT and start some additional calculations. For example, the purpose of this can be data conditioning for visualization purposes or the generation of input files for the next simulation tool in a simulation flow. To simplify the calling procedures and to make an additional wrap-around script obsolete, MINIMOS-NT is now able to execute arbitrary shell commands and to call external programs.

But MINIMOS-NT can also provide additional post-processing steps based on its own output data. A simple example is the lastLine post-processing steps, which stores the last line of each curve file block in a new curve file. This feature is especially useful for conditional steppings (see Appendix B.3), for example after the extraction of the $f_\textrm {T}$ curve. Although the respective behavior could also be controlled directly in the curve output module depending on an input-deck keyword, it was decided to use this feature as a demonstration on how to implement MINIMOS-NT post-processing steps.

However, MINIMOS-NT was also equipped with a more sophisticated post-processing step, namely a Fast Fourier Transformation functionality. The implementation of the transformation itself is based on the code given in [165]. In addition, a general input-deck-based interface is provided for the user.

The user can specify an arbitrary number of post-processing steps, which are processed in a consecutive order. The respective input-deck section has the name PostProcessing.

It is basically structured as follows:

PostProcessing
{
   postProcOnly = no;      // Perform only post-processing steps
   printSteps   = no;      // Print available post-processing steps
   enter        = yes;     // Enable post-processing at all

   Steps
   {
      FirstStep
      {
         function = "";
      }
      SecondStep
      {
         FirstSubStep
         {
            function = "";

         }
         ...
      }  ...
   }
}

The simulator iterates over all subsections found in section PostProcessing.Steps. The kind of post-processing step to execute is specified in the function string. Each block can be deactivated by setting enter to no. Since post-processing uses per definition already calculated data, the preceding simulation can be completely skipped in case the output files are already present (postProcOnly). In addition, the user can request a short help (printSteps) to see which post-processing functions are available at all.

As stated above, the lastLine step is a very simple one. Therefore it is used to sketch the usage and implementation idea. All available post-processing steps can be derived from default steps. In case of lastLine this predefined step looks as follows:

StepDefaults
{
   GeneralStep
   {
      enter    = yes;             // Perform function in this step?
      function = "";              // Function, such as "LastLine" or "FFT"  
   }
   <GeneralIOStep> : GeneralStep
   {
      input    = "";              // Input file
      output   = "";              // Output file
   }
   LastLineStep  : GeneralIOStep  // Write last line of each block
   {                              // to new curve file
      function = "LastLine";
   }
}

The necessary inputs are already derived from other sections, so a typical instance can be defined very quickly:

PostProcessing
{
   Steps
   {
      +LastLine : ^StepDefaults.LastLineStep
      {
         input    = ~Curve.file;
         output   = "ft_" + ~Curve.file;
      }
   }
}

In this example, the simulation task is to start a conditional stepping in order to extract the $f_\textrm {T}$ curve. The output of each step is written to the file specified in the Curve section. This file is the input file for the post-processing step, whereas the output file name is simply the same name with a preceding ft_.