Welcome to Morpheus!

Morpheus is an exciting new linear algebra package. It currently contains a dense matrix and dense vector class, but we hope to expand its features as it grows in popularity.

Okay, so Morpheus is really just a test project for an ATPESC tutorial...but at least it has a cool name!

gcov tutorial

Please log in to vesta
ssh <username>@vesta.alcf.anl.gov
Clone the Morpheus repository
git clone https://github.com/amklinv/morpheus.git
Build the tests
cd morpheus
make
Note that the proper coverage flags have been added to your makefile. Also note that this step generates a set of .gcno files for you.
Run the tests
./runtests
runtests is a perl script which runs the three tests for you. This step generates the .gcda files.
Run gcov on the Morpheus source code
gcov *.cpp
Ignore the system files; we are not responsible for testing them. This will generate our .gcov files

Examine Morpheus_Vector.cpp.gcov and Morpheus_Matrix.cpp.gcov
These are regular text files, so you may use your text editor of choice (vim, emacs, eclipse...or you can just cat the file). Lines that have been tested are marked by the number of times they were executed. Lines that have NOT been tested are preceeded by #####. Dashes denote lines that contain no instructions, such as blank lines or curly braces.
Example:



       5:   85:bool Matrix::isSymmetric() const
      -:   86:{
      5:   87:  if(nrows_ != ncols_)
  #####:   88:    return false;
      -:   89:
     30:   90:  for(int r=0; r<nrows_; r++)
      -:   91:  {
    150:   92:    for(int c=0; c<ncols_; c++)
      -:   93:    {
    125:   94:      if(data_[r][c] != data_[c][r])
  #####:   95:        return false;
      -:   96:    }
      -:   97:  }
      -:   98:
      5:   99:  return true;
      -:  100:}

Food for thought..

What percentage of Morpheus_Vector.cpp and Morpheus_Matrix.cpp did gcov report was being tested?
How much confidence do you have that my code is correct?
If we had 100% code coverage, would that mean there were no bugs? Why?
Which of the following Vector functions are tested?
- Constructor
- Destructor
- Subscript operator
- Const subscript operator
- getNumElements
- setValue
- scale
- add
- dot
- norm1
- normInf
- norm2
Are the two tests that exist for the Vector class good?
What types of data did I ignore?
What kinds of errors could occur as a result of me ignoring those types of data?
For which of the following vectors would the 1-norm function produce the correct result? The infinity-norm? The 2-norm?
$\left[\begin{array}{r}1\\2\\3\end{array}\right]$ $\left[\begin{array}{r}1\\0\\0\end{array}\right]$ $\left[\begin{array}{r}-1\\0\\0\end{array}\right]$
Which of the following Matrix functions are tested?
- Constructor
- Destructor
- Subscript operator
- getNumRows
- getNumCols
- getNumEntries
- Matrix-vector multiply
- Matrix-matrix multiply
- isSymmetric
- isUpperTriangular
- approxEqual
- norm1
- normInf
- print
Is it sufficient to test the isUpperTriangular function with the identity matrix? Why?
If my tests failed, would it be easy to track down the source of the problem? What could I do to make it easier?

Doxygen

To create the html pages you're currently looking at, all you have to do is type doxygen in the source directory. (Doxygen is already installed on Vesta.) It reads Doxyfile, which I generated with doxywizard on my workstation and checked into the repository. Alternatively, you can generate such files by hand.

Didn't you say Doxygen can generate LaTeX manuals?

Yes. You can find the one for this site here.

What about the beautiful GUIs you showed us?

lcov and doxywizard are great tools for a personal workstation, but not-so-great tools for computing clusters. I personally prefer using the GUIs, but it's important to know how to use gcov and doxygen too, since they're available on more systems. If you desperately need the HTML files generated by lcov, they are available here.