FAQ

Using with Python

Python doesn’t load MAP

If importing MAP++ into Python leads to this error on Linux/OSx:

OSError: ../src/libmap-1.20.00.so: cannot open shared object file: No such file or directory

or on Windows:

WindowsError: [Error 126] The specified module could not be found

then the issue is the MAP++ shared object/dll can’t be located by the MAP++ module. This is corrected by changing the path where the .so/.dll is picked up in mapsys.py:

lib = cdll.LoadLibrary("/directory/to/map/libmap-1.20.00.so")

On Windows, it would look something like this:

lib = cdll.LoadLibrary("C:\User\local\directory\map_x64.dll")

If the error still persists, the solution could be one of the answers on this post at Stack Overflow.

Initialization Errors

I get WARNING [5] or FATAL [41] and can’t emerge from initialization

The error thrown at initialization is either:

MAP_WARNING[5] : Cable density is approaching the density of seawater.
                 This may result in the problem becoming poorly conditioned.

or

MAP_FATAL[41] : Cable mass density is zero. Neutrally buoyant cables cannot
                be solved using quasi-statis model

We insert a check to warning the user if the cable is nearly neutrally buoyant. This causes the solver to go on the fritz, and sometimes fail, because the algebraic equation is close to dividing by zero; see the horizontal cable equation. The straightforward way to fix this is to change \(\omega\) tolerance levels in mapinit.c The default tolerance levels are omega \(\omega< 1.0\) for a warning, and omega \(\omega<10^{-3}\) for fatal. In some cases, this is unavoidable and a different mooring program is needed.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
    if (fabs(library_iter->omega)<=1.0) {
      set_universal_error_with_message(map_msg, ierr, MAP_WARNING_5, 
                                       "omega = %f <= 1.0", library_iter->omega);
    };
  };
  list_iterator_stop(&domain->library); /* ending the iteration "session" */    
  
  if (fabs(library_iter->omega)<=1.0E-3) {
    return MAP_FATAL;
  }

Todo

Include a run-time tolerance override option in the input file.

Maximum iterations are exceeded with taut mooring

The mooring system probably has a connect node. I’d start by increasing OUTER_TOL by an order of magnitude more than the default option. It doesn’t take much displacement error in the inner solve to cause a large difference in the outer loop, so there’s this constant game of playing catch-up. Alternatively, you can decrease the inner loop solve tolerance, but you might be already approaching machine precision. Taut lines are strange like that.

A good solver option/initial guess strategy should converge on a solution in under 100-500 total iterations in the first solve.

Running MAP++

What are the units of the linearized stiffness matrix?

The linearized stiffness matrix is a \(6\times 6\) entry comprised of four \(3 \times 3\) blocks:

\[\begin{split}\mathbf{K}_{6\times 6}=\begin{bmatrix} \mathbf{A}_{3\times 3} & \mathbf{B}_{3\times 3}\\ \mathbf{C}_{3\times 3} & \mathbf{D}_{3\times 3} \end{bmatrix}\end{split}\]

The units are:

  • [N/m] for \(\mathbf{A}\)
  • [N/rad] for \(\mathbf{B}\)
  • [Nm/m] for \(\mathbf{C}\)
  • [Nm/rad] for \(\mathbf{D}\)

Note that the reference position is fixed at the global origin.

The linearized stiffness values are not consistent

This is likely attributed to round-off errors. We recommend testing linearized stiffness matrix entries for sensitivity against different epsilon values, tolerances, and finite differencing methods. The matrix entries should converge towards a set of values for those different options.