23burgers Directory Reference

Files
	23burgers.C

Detailed Description

Introduction

This tutorial considers a parameterized heat-transfer–type problem based on the Burgers equation. Although the class name is Burgers, the implemented equations are scalar advection–diffusion; the objective is to compute a reduced basis for the velocity field using a simple offline/online split.

A detailed look into the code

This section explains the code of tutorial 23.

Header files

The program includes both standard and ITHACA-FV headers:

#include <iostream>
#include "fvCFD.H"
#include "IOmanip.H"
#include "Time.H"
#include "Burgers.H"
#include "ITHACAPOD.H"
#include <Eigen/Dense>

<Burgers.H> defines the full-order problem derived from laplacianProblem.

Implementation

The tutorial23 class inherits from Burgers and stores a reference to the velocity field U:

class tutorial23: public Burgers
{
    public:
        explicit tutorial23(int argc, char* argv[])
            : Burgers(argc, argv), U(_U()) {}
 
        volVectorField& U;
 
        void offlineSolve(word folder = "./ITHACAoutput/Offline/")
        {
            if (offline)
            {
                ITHACAstream::readMiddleFields(Ufield, U, folder);
            }
            else
            {
                truthSolve(folder);
            }
        }
};

During offline execution the code either reads existing snapshots or runs the full-order solver once and saves the resulting velocity fields.

Main function

In main the tutorial object is created and POD mode numbers are extracted from the dictionary file:

tutorial23 train(argc, argv);
ITHACAparameters* para = ITHACAparameters::getInstance(train._mesh(),
                     train._runTime());
int NmodesUout = para->ITHACAdict->lookupOrDefault<int>("NmodesUout", 15);
train.offlineSolve();
ITHACAPOD::getModes(train.Ufield, train.Umodes, train._U().name(),
                    train.podex, 0, 0, NmodesUout);

The reduced velocity basis is written to train.Umodes and can subsequently be used for online reduced computations.

The plain code

The plain code is available here.