SparseVector.h

   1 // Vector of floating-point values with mostly zero entries
   2 // Copyright © 2009 The University of Chicago
   3 #ifndef SPARSEVECTOR_H
   4 #define SPARSEVECTOR_H
   5
   6 // See also CSparseIntVector.
   7
   8 #include <q3textstream.h>
   9
  10 class CLexicon;
  11 class CWordCollection;
  12 class CMorphemeCollection;
  13
  14 // this is a class which keeps track of values
  15 // associated with an integer (the "dimension"),
  16 // useful when such a vector is "sparse," that is,
  17 // the values for most of the "dimensions" is zero
  18 // (and hence you don't want to have to remember each zero).
  19 // It keeps track of the dimensions that we care about,
  20 // with their corresponding "values"; any dimension not
  21 // explicitly listed is taken to have a zero value.
  22 class CSparseVector
  23 {
  24         friend int Overlap (CSparseVector& List1, CSparseVector& List2);
  25         friend class CLexicon;
  26 //      friend class CSparseVector;   // Implicitly friends with itself
  27
  28         protected:
  29
  30         int             m_Length; // the number of slots currently in use.
  31 //      int             m_MaximalIndex; // highest index available to outside world
  32         int*    m_Dimension;
  33         float*  m_Values;
  34         int             m_NormalizedFlag;
  35         int             m_NumberOfSlots;
  36
  37         int             ContainsDim (int n);
  38         int             GetLength();
  39 //      void    SetLength(int n);
  40         float   operator[] (int n); //n is the local value
  41
  42 public:
  43
  44         CSparseVector();
  45         CSparseVector ( int n);
  46         CSparseVector (CSparseVector&);
  47         ~CSparseVector();
  48
  49         void    operator() (int n, float val);  // n is the value of Dimension
  50         float   operator() (int n);                             // n is the value of Dimension
  51         void    operator= (CSparseVector&);
  52         int             operator== (CSparseVector&);
  53
  54
  55         // for iteration: (uses MFC-style syntax):
  56         int             GetNextDimension (int& DimOut, float& Value, int& Position);
  57
  58         void    Clear();
  59         void    ClearOut();
  60         int             GetNormalizedFlag();
  61         int             GetTopDimension();
  62         float   GetAt( int dim ); //uses outside world value of dimension
  63         void    SetAt ( int, float);
  64         int             Delete (int dim); //uses outside world value of dim
  65         void    IncrementAt(int, float = 1);
  66         int             GetNumberOfSlots ();
  67         void    SetNumberOfSlots(int);
  68 //      int             GetNumberOfNonZeroMembers();
  69         Q3TextStream& OutputToStream (CWordCollection&, Q3TextStream& stream, int threshold=2);
  70         void    MakeLogFreq (CWordCollection*);
  71         void    Normalize();
  72         float   InnerProduct (CSparseVector*);
  73
  74 };
  75
  76 #endif // SPARSEVECTOR_H