Package gov.nih.mipav.view.renderer.J3D.surfaceview.brainflattenerview

Class MjSparseMatrix

java.lang.Object
gov.nih.mipav.view.renderer.J3D.surfaceview.brainflattenerview.MjSparseMatrix
public class MjSparseMatrix extends Object
Sparse matrix implemented with a hash map. All undefined entries are zero.

  • Nested Class Summary

    Nested Classes
    Modifier and Type
    Class
    Description
    static class 
    MjSparseMatrix.Index
    Used to reference a single element in the matrix.

  • Field Summary

    Fields
    Modifier and Type
    Field
    Description
    private MjSparseMatrix.Index
    m_kIndexKeyTmp
    Instantiate once instead of each time a value key (Index) is needed.
    private HashMap
    m_kMap
    Mapping of Index(row,col).
    private static final float
    ms_fTolerance
    DOCUMENT ME!

  • Constructor Summary

    Constructors
    Constructor
    Description
    MjSparseMatrix()
    Matrix of undefined dimensions.

  • Method Summary

    Modifier and Type
    Method
    Description
    private static float
    dot(int iSize, float[] afU, float[] afV)
    conjugate gradient methods.
    float
    getElement(int iRow, int iCol)
    Get the element value at the specified row,col.
    Iterator
    iterator()
    Return iterator to non-zero elements in the matrix.
    private void
    multiply(int iSize, float[] afX, float[] afProd)
    DOCUMENT ME!
    void
    setElement(int iRow, int iCol, float fValue)
    Set the element value at the specified row,col.
    int
    size()
    return number of non-zero entries in the matrix.
    (package private) boolean
    solveSymmetricCG(int iSize, float[] afB, float[] afX)
    Conjugate gradient method for sparse, symmetric matrices.
    private static void
    updateP(int iSize, float[] afP, float fBeta, float[] afR)
    DOCUMENT ME!
    private static void
    updateR(int iSize, float[] afR, float fAlpha, float[] afW)
    DOCUMENT ME!
    private static void
    updateX(int iSize, float[] afX, float fAlpha, float[] afP)
    DOCUMENT ME!

    Methods inherited from class java.lang.Object

    clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

  • Field Details

    • ms_fTolerance

      private static final float ms_fTolerance
      DOCUMENT ME!
      See Also:

    • m_kIndexKeyTmp

      private MjSparseMatrix.Index m_kIndexKeyTmp
      Instantiate once instead of each time a value key (Index) is needed.

    • m_kMap

      private HashMap m_kMap
      Mapping of Index(row,col).

  • Constructor Details

    • MjSparseMatrix

      public MjSparseMatrix()
      Matrix of undefined dimensions. Initially all elements are zero.

  • Method Details

    • getElement

      public float getElement(int iRow, int iCol)
      Get the element value at the specified row,col.
      Parameters:
      iRow - DOCUMENT ME!
      iCol - DOCUMENT ME!
      Returns:
      DOCUMENT ME!

    • iterator

      public Iterator iterator()
      Return iterator to non-zero elements in the matrix. Each entry in the iterator is of Map.Entry type where the key is of Index type and the value is of Float type.
      Returns:
      DOCUMENT ME!

    • setElement

      public void setElement(int iRow, int iCol, float fValue)
      Set the element value at the specified row,col.
      Parameters:
      iRow - DOCUMENT ME!
      iCol - DOCUMENT ME!
      fValue - DOCUMENT ME!

    • size

      public int size()
      return number of non-zero entries in the matrix.
      Returns:
      DOCUMENT ME!

    • solveSymmetricCG

      boolean solveSymmetricCG(int iSize, float[] afB, float[] afX)
      Conjugate gradient method for sparse, symmetric matrices. Input: The nonzero entries of the symmetrix matrix A are stored in a map whose keys are pairs (i,j) and whose values are real numbers. The pair (i,j) is the location of the value in the array. Only one of (i,j) and (j,i) should be stored since A is symmetric. The code assumes this is how you set up A. The column vector B is stored as an array of contiguous valuess. If B itself is sparse, another solver can be implemented which uses a map to store B. Output: B[iSize] is the solution x to Ax = B
      Parameters:
      iSize - DOCUMENT ME!
      afB - DOCUMENT ME!
      afX - DOCUMENT ME!
      Returns:
      DOCUMENT ME!

    • dot

      private static float dot(int iSize, float[] afU, float[] afV)
      conjugate gradient methods.
      Parameters:
      iSize - DOCUMENT ME!
      afU - DOCUMENT ME!
      afV - DOCUMENT ME!
      Returns:
      DOCUMENT ME!

    • updateP

      private static void updateP(int iSize, float[] afP, float fBeta, float[] afR)
      DOCUMENT ME!
      Parameters:
      iSize - DOCUMENT ME!
      afP - DOCUMENT ME!
      fBeta - DOCUMENT ME!
      afR - DOCUMENT ME!

    • updateR

      private static void updateR(int iSize, float[] afR, float fAlpha, float[] afW)
      DOCUMENT ME!
      Parameters:
      iSize - DOCUMENT ME!
      afR - DOCUMENT ME!
      fAlpha - DOCUMENT ME!
      afW - DOCUMENT ME!

    • updateX

      private static void updateX(int iSize, float[] afX, float fAlpha, float[] afP)
      DOCUMENT ME!
      Parameters:
      iSize - DOCUMENT ME!
      afX - DOCUMENT ME!
      fAlpha - DOCUMENT ME!
      afP - DOCUMENT ME!

    • multiply

      private void multiply(int iSize, float[] afX, float[] afProd)
      DOCUMENT ME!
      Parameters:
      iSize - DOCUMENT ME!
      afX - DOCUMENT ME!
      afProd - DOCUMENT ME!