Implementation of Lamports method for scheduling

The implementation for this method is collected in only one file, scheduler/lamport.cc, with the following main function:

//-----------------------------------------------------------------------------
// function   : lamport_scheduler
// description: This function implements the scheduler described by Lamport.
//----------------------------------------------------------------------------- 
void lamport_scheduler( Source_code *source )
{
        cout << LAMPORT_START;
        // create the distance matrix from the h-transformations
        Matrix distance_matrix = create_distance_matrix( source );
        // partition the distance matrix by distance levels
        Matrix **matrix_list = partition_by_levels( distance_matrix );
        // calculate the first column vector of the transformation matrix
        Matrix vector_u = find_first_column_vector( matrix_list,
                                                    source->get_n_vars() );
        // create the transformation matrix
        Matrix transformation_matrix = 
                                create_transformation_matrix( vector_u );
        Affin_function *schedule = calculate_schedule( transformation_matrix,
                                                                source );
        Affin_function *placement = calculate_placement( transformation_matrix,
                                                                source );
        cout << LAMPORT_RESULT;
        transformation_matrix.print();
        source->print_schedules();
        source->print_placements();
        source->save_placements();
        source->save_schedules();
        
        #ifndef NDEBUG
        cout << ROUTINE_DONE;
        #endif
}

This routine is called by the main function of the scheduling algorithms in mainsched.cc, if the parameter job given to the binary scheduler was "l". The implementation above is straightforward. Since the algorithm expects a distance matrix, we first have to convert the more general representation of dependences, given by the h-transformations, to matrix form. After this the elements of the distance matrix, namely the distance vectors, are sorted after their dependence level. Now all is done to start the algorithm, that consists of two parts:

Find a first column vector representing a schedule
Expand this vector to a transformation matrix

The output of each algorithm calculating a schedule is a function stored in a file with extension SCHED. Since we have at this point only a matrix containing a schedule and an allocation, we have to extract schedule and allocation from it. After this the two functions are saved.

Note: The method of lamport calculates besides the schedule also an allocation, which is in most cases useless for minimizing communication.